Aren
/
NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
NetBSD/sys/kern/init_sysctl.c

3002 lines
79 KiB
C
Raw Normal View History

Merge scheduler changes from the vmlocking branch. All discussed on tech-kern: - Invert priority space so that zero is the lowest priority. Rearrange number and type of priority levels into bands. Add new bands like 'kernel real time'. - Ignore the priority level passed to tsleep. Compute priority for sleep dynamically. - For SCHED_4BSD, make priority adjustment per-LWP, not per-process.
2007-11-06 03:42:39 +03:00
/* $NetBSD: init_sysctl.c,v 1.113 2007/11/06 00:42:40 ad Exp $ */
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*-
* Copyright (c) 2003 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Andrew Brown.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* This product includes software developed by the NetBSD
* Foundation, Inc. and its contributors.
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* 4. Neither the name of The NetBSD Foundation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
RCSid police
2003-12-29 01:12:00 +03:00
#include <sys/cdefs.h>
Merge scheduler changes from the vmlocking branch. All discussed on tech-kern: - Invert priority space so that zero is the lowest priority. Rearrange number and type of priority levels into bands. Add new bands like 'kernel real time'. - Ignore the priority level passed to tsleep. Compute priority for sleep dynamically. - For SCHED_4BSD, make priority adjustment per-LWP, not per-process.
2007-11-06 03:42:39 +03:00
__KERNEL_RCSID(0, "$NetBSD: init_sysctl.c,v 1.113 2007/11/06 00:42:40 ad Exp $");
RCSid police
2003-12-29 01:12:00 +03:00
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#include "opt_sysv.h"
#include "opt_multiprocessor.h" This makes hw.ncpu and kern.cp_time work better on those platforms.
2003-12-06 05:52:29 +03:00
#include "opt_multiprocessor.h"
Include opt_posix.h for the P1003_1B_SEMAPHORE define. Include <machine/cpu.h> just to be sure.
2003-12-06 23:06:11 +03:00
#include "opt_posix.h"
When colecting a 32 bit process' argument or environement vector, we need to convert 32 bits pointers to the 64 bit environement
2006-09-08 15:59:52 +04:00
#include "opt_compat_netbsd32.h"
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#include "pty.h"
#include "rnd.h"
#include <sys/types.h>
#include <sys/param.h>
#include <sys/sysctl.h>
remove a homegrown definition of CPU_INFO_FOREACH.
2007-08-06 15:51:46 +04:00
#include <sys/cpu.h>
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#include <sys/errno.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/unistd.h>
#include <sys/disklabel.h>
#include <sys/rnd.h>
#include <sys/vnode.h>
#include <sys/mount.h>
#include <sys/namei.h>
#include <sys/msgbuf.h>
#include <dev/cons.h>
#include <sys/socketvar.h>
#include <sys/file.h>
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
#include <sys/filedesc.h>
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#include <sys/tty.h>
#include <sys/malloc.h>
#include <sys/resource.h>
#include <sys/resourcevar.h>
#include <sys/exec.h>
#include <sys/conf.h>
#include <sys/device.h>
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
#include <sys/stat.h>
integrate kauth.
2006-05-15 01:15:11 +04:00
#include <sys/kauth.h>
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
#include <sys/ktrace.h>
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
When colecting a 32 bit process' argument or environement vector, we need to convert 32 bits pointers to the 64 bit environement
2006-09-08 15:59:52 +04:00
#ifdef COMPAT_NETBSD32
#include <compat/netbsd32/netbsd32.h>
#endif
machine/{bus,cpu,intr}.h -> sys/{bus,cpu,intr}.h
2007-10-19 15:59:34 +04:00
#include <sys/cpu.h>
Include opt_posix.h for the P1003_1B_SEMAPHORE define. Include <machine/cpu.h> just to be sure.
2003-12-06 23:06:11 +03:00
Introduce ``security.curtain'', new node for security features and settings, and new variable for controlling access to objects based on user-id.
2005-09-07 20:26:15 +04:00
/* XXX this should not be here */
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
int security_setidcore_dump;
char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
uid_t security_setidcore_owner = 0;
gid_t security_setidcore_group = 0;
mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
Introduce ``security.curtain'', new node for security features and settings, and new variable for controlling access to objects based on user-id.
2005-09-07 20:26:15 +04:00
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
static const u_int sysctl_flagmap[] = {
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
PK_ADVLOCK, P_ADVLOCK,
PK_EXEC, P_EXEC,
PK_NOCLDWAIT, P_NOCLDWAIT,
PK_32, P_32,
PK_CLDSIGIGN, P_CLDSIGIGN,
PK_PAXMPROTECT, P_PAXMPROTECT,
PK_PAXNOMPROTECT, P_PAXNOMPROTECT,
PK_SYSTRACE, P_SYSTRACE,
PK_SUGID, P_SUGID,
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
0
};
static const u_int sysctl_sflagmap[] = {
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
PS_NOCLDSTOP, P_NOCLDSTOP,
PS_PPWAIT, P_PPWAIT,
PS_WEXIT, P_WEXIT,
PS_STOPFORK, P_STOPFORK,
PS_STOPEXEC, P_STOPEXEC,
PS_STOPEXIT, P_STOPEXIT,
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
0
};
static const u_int sysctl_slflagmap[] = {
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
PSL_TRACED, P_TRACED,
PSL_FSTRACE, P_FSTRACE,
PSL_CHTRACED, P_CHTRACED,
PSL_SYSCALL, P_SYSCALL,
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
0
};
static const u_int sysctl_lflagmap[] = {
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
PL_CONTROLT, P_CONTROLT,
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
0
};
static const u_int sysctl_stflagmap[] = {
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
PST_PROFIL, P_PROFIL,
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
0
};
static const u_int sysctl_lwpflagmap[] = {
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
LW_INMEM, P_INMEM,
LW_SINTR, P_SINTR,
LW_SYSTEM, P_SYSTEM,
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
0
};
static const u_int sysctl_lwpprflagmap[] = {
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
LPR_DETACHED, L_DETACHED,
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
0
};
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
* try over estimating by 5 procs/lwps
*/
#define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc))
#define KERN_LWPSLOP (5 * sizeof(struct kinfo_lwp))
Changes to make ktrace LKM friendly and reduce ifdef KTRACE. Proposed on tech-kern.
2007-08-15 16:07:23 +04:00
static int dcopyout(struct lwp *, const void *, void *, size_t);
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
Changes to make ktrace LKM friendly and reduce ifdef KTRACE. Proposed on tech-kern.
2007-08-15 16:07:23 +04:00
static int
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
dcopyout(l, kaddr, uaddr, len)
struct lwp *l;
const void *kaddr;
void *uaddr;
size_t len;
{
int error;
error = copyout(kaddr, uaddr, len);
Changes to make ktrace LKM friendly and reduce ifdef KTRACE. Proposed on tech-kern.
2007-08-15 16:07:23 +04:00
ktrmibio(-1, UIO_READ, uaddr, len, error);
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
return error;
}
Count the number of CPUs at boot and stash in 'ncpu'. Eventually should have each CPU register at attach, so we can figure out the topology for the scheduler.
2007-02-15 23:32:47 +03:00
When DIAGNOSTIC is defined, provide a kern.panic_now sysctl to conviniently and reliably panic the system
2006-03-27 00:07:21 +04:00
#ifdef DIAGNOSTIC
static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
#endif
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
Make kern.rtc_offset writable at securelevel <= 0. This allows boot-time adjustment when a machine runs other OSes with RTC == localtime.
2003-12-27 02:49:39 +03:00
static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
static int sysctl_kern_hostid(SYSCTLFN_PROTO);
Rename sysctl_kern_hostname() to sysctl_setlen() and use it also for domainname. Note that there's no need to copy rnode since we're not changing any of it, nor protecting anything from change. Thanks to martin for initial work.
2003-12-29 01:19:59 +03:00
static int sysctl_setlen(SYSCTLFN_PROTO);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
static int sysctl_kern_file(SYSCTLFN_PROTO);
static int sysctl_msgbuf(SYSCTLFN_PROTO);
static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
static int sysctl_kern_cptime(SYSCTLFN_PROTO);
Also make declaration of sysctl_kern_maxptys() depend on NPTY > 0. Makes the mvme68k RAMDISK kernel compile again.
2003-12-07 13:31:32 +03:00
#if NPTY > 0
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
Also make declaration of sysctl_kern_maxptys() depend on NPTY > 0. Makes the mvme68k RAMDISK kernel compile again.
2003-12-07 13:31:32 +03:00
#endif /* NPTY > 0 */
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
static int sysctl_kern_urnd(SYSCTLFN_PROTO);
implement kern.arandom properly, instead of lying about it and only filling the first 4 bytes of the array with random data.
2006-11-02 01:27:43 +03:00
static int sysctl_kern_arnd(SYSCTLFN_PROTO);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
static int sysctl_kern_lwp(SYSCTLFN_PROTO);
static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
static int sysctl_kern_drivers(SYSCTLFN_PROTO);
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
static int sysctl_kern_file2(SYSCTLFN_PROTO);
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
static int sysctl_security_setidcore(SYSCTLFN_PROTO);
static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
Add a new sysctl 'cp_id' that returns the array of cpu id values. Requested by me, implemented by atatat.
2005-06-16 18:55:58 +04:00
static int sysctl_kern_cpid(SYSCTLFN_PROTO);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
static int sysctl_doeproc(SYSCTLFN_PROTO);
static int sysctl_kern_proc_args(SYSCTLFN_PROTO);
static int sysctl_hw_usermem(SYSCTLFN_PROTO);
static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
static u_int sysctl_map_flags(const u_int *, u_int);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
static void fill_kproc2(struct proc *, struct kinfo_proc2 *);
static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
static void fill_file(struct kinfo_file *, const struct file *, struct proc *,
int);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
* ********************************************************************
* section 1: setup routines
* ********************************************************************
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* These functions are stuffed into a link set for sysctl setup
* functions. They're never called or referenced from anywhere else.
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* ********************************************************************
*/
/*
* sets up the base nodes...
*/
SYSCTL_SETUP(sysctl_root_setup, "sysctl base setup")
{
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "kern",
SYSCTL_DESCR("High kernel"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_KERN, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "vm",
SYSCTL_DESCR("Virtual memory"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_VM, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "vfs",
SYSCTL_DESCR("Filesystem"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_VFS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "net",
SYSCTL_DESCR("Networking"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_NET, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "debug",
SYSCTL_DESCR("Debugging"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_DEBUG, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "hw",
SYSCTL_DESCR("Generic CPU, I/O"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_HW, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "machdep",
SYSCTL_DESCR("Machine dependent"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_MACHDEP, CTL_EOL);
/*
* this node is inserted so that the sysctl nodes in libc can
* operate.
*/
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "user",
SYSCTL_DESCR("User-level"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_USER, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "ddb",
SYSCTL_DESCR("In-kernel debugger"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_DDB, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "proc",
SYSCTL_DESCR("Per-process"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_PROC, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "vendor",
SYSCTL_DESCR("Vendor specific"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_VENDOR, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "emul",
SYSCTL_DESCR("Emulation settings"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, NULL, 0,
CTL_EMUL, CTL_EOL);
Introduce ``security.curtain'', new node for security features and settings, and new variable for controlling access to objects based on user-id.
2005-09-07 20:26:15 +04:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
CTLTYPE_NODE, "security",
SYSCTL_DESCR("Security"),
NULL, 0, NULL, 0,
CTL_SECURITY, CTL_EOL);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
/*
* this setup routine is a replacement for kern_sysctl()
*/
SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
{
extern int kern_logsigexit; /* defined in kern/kern_sig.c */
extern fixpt_t ccpu; /* defined in kern/kern_synch.c */
extern int dumponpanic; /* defined in kern/subr_prf.c */
move security.setid_core.* to kern.coredump.setid.*, as requested by yamt@.
2006-07-15 01:55:19 +04:00
const struct sysctlnode *rnode;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTLTYPE_NODE, "kern", NULL,
NULL, 0, NULL, 0,
CTL_KERN, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "ostype",
SYSCTL_DESCR("Operating system type"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, &ostype, 0,
CTL_KERN, KERN_OSTYPE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "osrelease",
SYSCTL_DESCR("Operating system release"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, &osrelease, 0,
CTL_KERN, KERN_OSRELEASE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "osrevision",
SYSCTL_DESCR("Operating system revision"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, __NetBSD_Version__, NULL, 0,
CTL_KERN, KERN_OSREV, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "version",
SYSCTL_DESCR("Kernel version"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, &version, 0,
CTL_KERN, KERN_VERSION, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "maxvnodes",
SYSCTL_DESCR("Maximum number of vnodes"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_maxvnodes, 0, NULL, 0,
CTL_KERN, KERN_MAXVNODES, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "maxproc",
SYSCTL_DESCR("Maximum number of simultaneous processes"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_maxproc, 0, NULL, 0,
CTL_KERN, KERN_MAXPROC, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "maxfiles",
SYSCTL_DESCR("Maximum number of open files"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, &maxfiles, 0,
CTL_KERN, KERN_MAXFILES, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "argmax",
SYSCTL_DESCR("Maximum number of bytes of arguments to "
"execve(2)"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, ARG_MAX, NULL, 0,
CTL_KERN, KERN_ARGMAX, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "hostname",
SYSCTL_DESCR("System hostname"),
Rename sysctl_kern_hostname() to sysctl_setlen() and use it also for domainname. Note that there's no need to copy rnode since we're not changing any of it, nor protecting anything from change. Thanks to martin for initial work.
2003-12-29 01:19:59 +03:00
sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTL_KERN, KERN_HOSTNAME, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
Prefer that kern.hostid is printed in hex, not as a signed decimal, and avoid accidental sign-extension when setting it.
2004-04-16 17:25:40 +04:00
CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "hostid",
SYSCTL_DESCR("System host ID number"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_hostid, 0, NULL, 0,
CTL_KERN, KERN_HOSTID, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRUCT, "clockrate",
SYSCTL_DESCR("Kernel clock rates"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_clockrate, 0, NULL,
sizeof(struct clockinfo),
CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
Implement kern.hardclock_ticks.
2005-09-06 06:36:17 +04:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
CTLTYPE_INT, "hardclock_ticks",
SYSCTL_DESCR("Number of hardclock ticks"),
NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRUCT, "vnode",
SYSCTL_DESCR("System vnode table"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_vnode, 0, NULL, 0,
CTL_KERN, KERN_VNODE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRUCT, "file",
SYSCTL_DESCR("System open file table"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_file, 0, NULL, 0,
CTL_KERN, KERN_FILE, CTL_EOL);
#ifndef GPROF
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "profiling",
SYSCTL_DESCR("Profiling information (not available)"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_notavail, 0, NULL, 0,
CTL_KERN, KERN_PROF, CTL_EOL);
#endif
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "posix1version",
SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
"with which the operating system attempts "
"to comply"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, _POSIX_VERSION, NULL, 0,
CTL_KERN, KERN_POSIX1, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "ngroups",
SYSCTL_DESCR("Maximum number of supplemental groups"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, NGROUPS_MAX, NULL, 0,
CTL_KERN, KERN_NGROUPS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "job_control",
SYSCTL_DESCR("Whether job control is available"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 1, NULL, 0,
CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "saved_ids",
be explicit in the description for POSIX saved set-id that this is for POSIX-style, not sane-style. (ie, add "POSIX " to the description.)
2005-04-18 14:46:39 +04:00
SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
"available"), NULL,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#ifdef _POSIX_SAVED_IDS
1,
#else /* _POSIX_SAVED_IDS */
0,
#endif /* _POSIX_SAVED_IDS */
NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRUCT, "boottime",
SYSCTL_DESCR("System boot time"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, &boottime, sizeof(boottime),
CTL_KERN, KERN_BOOTTIME, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "domainname",
SYSCTL_DESCR("YP domain name"),
Rename sysctl_kern_hostname() to sysctl_setlen() and use it also for domainname. Note that there's no need to copy rnode since we're not changing any of it, nor protecting anything from change. Thanks to martin for initial work.
2003-12-29 01:19:59 +03:00
sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "maxpartitions",
SYSCTL_DESCR("Maximum number of partitions allowed per "
"disk"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, MAXPARTITIONS, NULL, 0,
CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "rawpartition",
SYSCTL_DESCR("Raw partition of a disk"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, RAW_PART, NULL, 0,
CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTLTYPE_STRUCT, "timex", NULL,
sysctl_notavail, 0, NULL, 0,
CTL_KERN, KERN_TIMEX, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "rtc_offset",
SYSCTL_DESCR("Offset of real time clock from UTC in "
"minutes"),
Make kern.rtc_offset writable at securelevel <= 0. This allows boot-time adjustment when a machine runs other OSes with RTC == localtime.
2003-12-27 02:49:39 +03:00
sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "root_device",
SYSCTL_DESCR("Name of the root device"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_root_device, 0, NULL, 0,
CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "msgbufsize",
SYSCTL_DESCR("Size of the kernel message buffer"),
The message buffer datum instrumented by KERN_MSGBUFSIZE is actually a long, not an int, and this causes "problems" on LP64be machines (sparc64, etc). Assign the value to a temporary int and instrument that instead. Should be fine until someone wants a message buffer larger than two gigabytes.
2004-07-27 16:46:18 +04:00
sysctl_msgbuf, 0, NULL, 0,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "fsync",
SYSCTL_DESCR("Whether the POSIX 1003.1b File "
"Synchronization Option is available on "
"this system"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 1, NULL, 0,
CTL_KERN, KERN_FSYNC, CTL_EOL);
PR/34837: Mindaguas: Add SysV SHM dynamic reallocation and locking to the physical memory
2006-11-26 00:40:04 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
CTLTYPE_NODE, "ipc",
SYSCTL_DESCR("SysV IPC options"),
NULL, 0, NULL, 0,
CTL_KERN, KERN_SYSVIPC, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "sysvmsg",
SYSCTL_DESCR("System V style message support available"),
NULL,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#ifdef SYSVMSG
1,
#else /* SYSVMSG */
0,
#endif /* SYSVMSG */
PR/34837: Mindaguas: Add SysV SHM dynamic reallocation and locking to the physical memory
2006-11-26 00:40:04 +03:00
NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "sysvsem",
SYSCTL_DESCR("System V style semaphore support "
"available"), NULL,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#ifdef SYSVSEM
1,
#else /* SYSVSEM */
0,
#endif /* SYSVSEM */
PR/34837: Mindaguas: Add SysV SHM dynamic reallocation and locking to the physical memory
2006-11-26 00:40:04 +03:00
NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "sysvshm",
SYSCTL_DESCR("System V style shared memory support "
"available"), NULL,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#ifdef SYSVSHM
1,
#else /* SYSVSHM */
0,
#endif /* SYSVSHM */
PR/34837: Mindaguas: Add SysV SHM dynamic reallocation and locking to the physical memory
2006-11-26 00:40:04 +03:00
NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "synchronized_io",
SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
"I/O Option is available on this system"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 1, NULL, 0,
CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "iov_max",
SYSCTL_DESCR("Maximum number of iovec structures per "
"process"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, IOV_MAX, NULL, 0,
CTL_KERN, KERN_IOV_MAX, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "mapped_files",
SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
"Files Option is available on this system"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 1, NULL, 0,
CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "memlock",
SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
"Locking Option is available on this "
"system"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 1, NULL, 0,
CTL_KERN, KERN_MEMLOCK, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "memlock_range",
SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
"Locking Option is available on this "
"system"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 1, NULL, 0,
CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "memory_protection",
SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
"Protection Option is available on this "
"system"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 1, NULL, 0,
CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "login_name_max",
SYSCTL_DESCR("Maximum login name length"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, LOGIN_NAME_MAX, NULL, 0,
CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "defcorename",
SYSCTL_DESCR("Default core file name"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "logsigexit",
SYSCTL_DESCR("Log process exit when caused by signals"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, &kern_logsigexit, 0,
CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "fscale",
SYSCTL_DESCR("Kernel fixed-point scale factor"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, FSCALE, NULL, 0,
CTL_KERN, KERN_FSCALE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "ccpu",
SYSCTL_DESCR("Scheduler exponential decay value"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, &ccpu, 0,
CTL_KERN, KERN_CCPU, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRUCT, "cp_time",
SYSCTL_DESCR("Clock ticks spent in different CPU states"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_cptime, 0, NULL, 0,
CTL_KERN, KERN_CP_TIME, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "msgbuf",
SYSCTL_DESCR("Kernel message buffer"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_msgbuf, 0, NULL, 0,
CTL_KERN, KERN_MSGBUF, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRUCT, "consdev",
SYSCTL_DESCR("Console device"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_consdev, 0, NULL, sizeof(dev_t),
CTL_KERN, KERN_CONSDEV, CTL_EOL);
#if NPTY > 0
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "maxptys",
SYSCTL_DESCR("Maximum number of pseudo-ttys"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_maxptys, 0, NULL, 0,
CTL_KERN, KERN_MAXPTYS, CTL_EOL);
#endif /* NPTY > 0 */
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "maxphys",
SYSCTL_DESCR("Maximum raw I/O transfer size"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, MAXPHYS, NULL, 0,
CTL_KERN, KERN_MAXPHYS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "sbmax",
SYSCTL_DESCR("Maximum socket buffer size"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_sbmax, 0, NULL, 0,
CTL_KERN, KERN_SBMAX, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "monotonic_clock",
SYSCTL_DESCR("Implementation version of the POSIX "
"1003.1b Monotonic Clock Option"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/* XXX _POSIX_VERSION */
NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "urandom",
SYSCTL_DESCR("Random integer value"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_urnd, 0, NULL, 0,
CTL_KERN, KERN_URND, CTL_EOL);
implement kern.arandom properly, instead of lying about it and only filling the first 4 bytes of the array with random data.
2006-11-02 01:27:43 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
CTLTYPE_INT, "arandom",
SYSCTL_DESCR("n bytes of random data"),
sysctl_kern_arnd, 0, NULL, 0,
CTL_KERN, KERN_ARND, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "labelsector",
SYSCTL_DESCR("Sector number containing the disklabel"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, LABELSECTOR, NULL, 0,
CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "labeloffset",
SYSCTL_DESCR("Offset of the disklabel within the "
"sector"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, LABELOFFSET, NULL, 0,
CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "lwp",
SYSCTL_DESCR("System-wide LWP information"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_lwp, 0, NULL, 0,
CTL_KERN, KERN_LWP, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "forkfsleep",
SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
"to process limits"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_forkfsleep, 0, NULL, 0,
CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "posix_threads",
SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
"Threads option to which the system "
"attempts to conform"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/* XXX _POSIX_VERSION */
NULL, _POSIX_THREADS, NULL, 0,
CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "posix_semaphores",
SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
"Semaphores option to which the system "
"attempts to conform"), NULL,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#ifdef P1003_1B_SEMAPHORE
200112,
#else /* P1003_1B_SEMAPHORE */
0,
#endif /* P1003_1B_SEMAPHORE */
NULL, 0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "posix_barriers",
SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
"Barriers option to which the system "
"attempts to conform"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/* XXX _POSIX_VERSION */
NULL, _POSIX_BARRIERS, NULL, 0,
CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "posix_timers",
SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
"Timers option to which the system "
"attempts to conform"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/* XXX _POSIX_VERSION */
NULL, _POSIX_TIMERS, NULL, 0,
CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "posix_spin_locks",
SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
"Locks option to which the system attempts "
"to conform"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/* XXX _POSIX_VERSION */
NULL, _POSIX_SPIN_LOCKS, NULL, 0,
CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "posix_reader_writer_locks",
SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
"Read-Write Locks option to which the "
"system attempts to conform"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/* XXX _POSIX_VERSION */
NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "dump_on_panic",
SYSCTL_DESCR("Perform a crash dump on system panic"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, &dumponpanic, 0,
CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
When DIAGNOSTIC is defined, provide a kern.panic_now sysctl to conviniently and reliably panic the system
2006-03-27 00:07:21 +04:00
#ifdef DIAGNOSTIC
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_INT, "panic_now",
SYSCTL_DESCR("Trigger a panic"),
sysctl_kern_trigger_panic, 0, NULL, 0,
CTL_KERN, CTL_CREATE, CTL_EOL);
#endif
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "root_partition",
SYSCTL_DESCR("Root partition on the root device"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_root_partition, 0, NULL, 0,
CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRUCT, "drivers",
SYSCTL_DESCR("List of all drivers with block and "
"character device numbers"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_drivers, 0, NULL, 0,
CTL_KERN, KERN_DRIVERS, CTL_EOL);
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Revert the change that made kern.file2 and net.*.*.pcblist into nodes instead of structs. It had other deleterious side-effects that are rather nasty. Another solution must be found.
2005-03-11 09:16:15 +03:00
CTLTYPE_STRUCT, "file2",
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
SYSCTL_DESCR("System open file table"),
sysctl_kern_file2, 0, NULL, 0,
CTL_KERN, KERN_FILE2, CTL_EOL);
Add a new sysctl 'cp_id' that returns the array of cpu id values. Requested by me, implemented by atatat.
2005-06-16 18:55:58 +04:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
CTLTYPE_STRUCT, "cp_id",
SYSCTL_DESCR("Mapping of CPU number to CPU id"),
sysctl_kern_cpid, 0, NULL, 0,
CTL_KERN, KERN_CP_ID, CTL_EOL);
move security.setid_core.* to kern.coredump.setid.*, as requested by yamt@.
2006-07-15 01:55:19 +04:00
sysctl_createv(clog, 0, NULL, &rnode,
CTLFLAG_PERMANENT,
CTLTYPE_NODE, "coredump",
SYSCTL_DESCR("Coredump settings."),
NULL, 0, NULL, 0,
CTL_KERN, CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, &rnode, &rnode,
CTLFLAG_PERMANENT,
CTLTYPE_NODE, "setid",
SYSCTL_DESCR("Set-id processes' coredump settings."),
NULL, 0, NULL, 0,
CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, &rnode, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_INT, "dump",
SYSCTL_DESCR("Allow set-id processes to dump core."),
sysctl_security_setidcore, 0, &security_setidcore_dump,
sizeof(security_setidcore_dump),
CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, &rnode, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_STRING, "path",
SYSCTL_DESCR("Path pattern for set-id coredumps."),
sysctl_security_setidcorename, 0,
&security_setidcore_path,
sizeof(security_setidcore_path),
CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, &rnode, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_INT, "owner",
SYSCTL_DESCR("Owner id for set-id processes' cores."),
sysctl_security_setidcore, 0, &security_setidcore_owner,
0,
CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, &rnode, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_INT, "group",
SYSCTL_DESCR("Group id for set-id processes' cores."),
sysctl_security_setidcore, 0, &security_setidcore_group,
0,
CTL_CREATE, CTL_EOL);
sysctl_createv(clog, 0, &rnode, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
CTLTYPE_INT, "mode",
SYSCTL_DESCR("Mode for set-id processes' cores."),
sysctl_security_setidcore, 0, &security_setidcore_mode,
0,
CTL_CREATE, CTL_EOL);
Add kern.no_sa_support to easily detect whether a kernel supports Scheduler Activation or not. This is a negative name as ld.so.conf conditionals threat undefined sysctls like 0.
2007-09-29 01:25:21 +04:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
CTLTYPE_INT, "no_sa_support",
SYSCTL_DESCR("0 if the kernel supports SA, otherwise it doesn't"),
NULL, 1, NULL, 0,
CTL_KERN, CTL_CREATE, CTL_EOL);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
SYSCTL_SETUP(sysctl_kern_proc_setup,
"sysctl kern.proc/proc2/proc_args subtree setup")
{
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTLTYPE_NODE, "kern", NULL,
NULL, 0, NULL, 0,
CTL_KERN, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "proc",
SYSCTL_DESCR("System-wide process information"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_doeproc, 0, NULL, 0,
CTL_KERN, KERN_PROC, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "proc2",
SYSCTL_DESCR("Machine-independent process information"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_doeproc, 0, NULL, 0,
CTL_KERN, KERN_PROC2, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_NODE, "proc_args",
SYSCTL_DESCR("Process argument information"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_proc_args, 0, NULL, 0,
CTL_KERN, KERN_PROC_ARGS, CTL_EOL);
/*
"nodes" under these:
KERN_PROC_ALL
KERN_PROC_PID pid
KERN_PROC_PGRP pgrp
KERN_PROC_SESSION sess
KERN_PROC_TTY tty
KERN_PROC_UID uid
KERN_PROC_RUID uid
KERN_PROC_GID gid
KERN_PROC_RGID gid
all in all, probably not worth the effort...
*/
}
SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
{
u_int u;
u_quad_t q;
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTLTYPE_NODE, "hw", NULL,
NULL, 0, NULL, 0,
CTL_HW, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "machine",
SYSCTL_DESCR("Machine class"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, machine, 0,
CTL_HW, HW_MACHINE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "model",
SYSCTL_DESCR("Machine model"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, cpu_model, 0,
CTL_HW, HW_MODEL, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "ncpu",
Add _SC_NPROCESSORS_ONLN and _SC_NPROCESSORS_CONF for sysconf(). These are extensions but are provided by many Unix systems.
2007-10-15 18:12:54 +04:00
SYSCTL_DESCR("Number of CPUs configured"),
NULL, 0, &ncpu, 0,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTL_HW, HW_NCPU, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "byteorder",
SYSCTL_DESCR("System byte order"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, BYTE_ORDER, NULL, 0,
CTL_HW, HW_BYTEORDER, CTL_EOL);
u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
UINT_MAX : physmem * PAGE_SIZE;
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "physmem",
SYSCTL_DESCR("Bytes of physical memory"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, u, NULL, 0,
CTL_HW, HW_PHYSMEM, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "usermem",
SYSCTL_DESCR("Bytes of non-kernel memory"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_hw_usermem, 0, NULL, 0,
CTL_HW, HW_USERMEM, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "pagesize",
SYSCTL_DESCR("Software page size"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, PAGE_SIZE, NULL, 0,
CTL_HW, HW_PAGESIZE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "machine_arch",
SYSCTL_DESCR("Machine CPU class"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, 0, machine_arch, 0,
CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_INT, "alignbytes",
SYSCTL_DESCR("Alignment constraint for all possible "
"data types"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, ALIGNBYTES, NULL, 0,
CTL_HW, HW_ALIGNBYTES, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_STRING, "cnmagic",
SYSCTL_DESCR("Console magic key sequence"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
CTL_HW, HW_CNMAGIC, CTL_EOL);
q = (u_quad_t)physmem * PAGE_SIZE;
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_QUAD, "physmem64",
SYSCTL_DESCR("Bytes of physical memory"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
NULL, q, NULL, 0,
CTL_HW, HW_PHYSMEM64, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Lots of sysctl descriptions (if someone wants to help out here, that would be good) mostly copied from sysctl(3). This takes care of the top-level, most of kern.* and hw.* (modulo the ath and bge stuff), and all of proc.*. If you don't want the added rodata in your kernel, use "options SYSCTL_NO_DESCR" in your kernel config.
2004-04-08 10:20:29 +04:00
CTLTYPE_QUAD, "usermem64",
SYSCTL_DESCR("Bytes of non-kernel memory"),
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_hw_usermem, 0, NULL, 0,
CTL_HW, HW_USERMEM64, CTL_EOL);
Add _SC_NPROCESSORS_ONLN and _SC_NPROCESSORS_CONF for sysconf(). These are extensions but are provided by many Unix systems.
2007-10-15 18:12:54 +04:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
CTLTYPE_INT, "ncpuonline",
SYSCTL_DESCR("Number of CPUs online"),
NULL, 0, &ncpuonline, 0,
CTL_HW, HW_NCPUONLINE, CTL_EOL);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
#ifdef DEBUG
/*
* Debugging related system variables.
*/
struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
struct ctldebug debug5, debug6, debug7, debug8, debug9;
struct ctldebug debug10, debug11, debug12, debug13, debug14;
struct ctldebug debug15, debug16, debug17, debug18, debug19;
static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
&debug0, &debug1, &debug2, &debug3, &debug4,
&debug5, &debug6, &debug7, &debug8, &debug9,
&debug10, &debug11, &debug12, &debug13, &debug14,
&debug15, &debug16, &debug17, &debug18, &debug19,
};
/*
* this setup routine is a replacement for debug_sysctl()
*
* note that it creates several nodes per defined debug variable
*/
SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
{
struct ctldebug *cdp;
char nodename[20];
int i;
/*
* two ways here:
*
* the "old" way (debug.name -> value) which was emulated by
* the sysctl(8) binary
*
* the new way, which the sysctl(8) binary was actually using
node debug
node debug.0
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
string debug.0.name
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
int debug.0.value
int debug.name
*/
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTLTYPE_NODE, "debug", NULL,
NULL, 0, NULL, 0,
CTL_DEBUG, CTL_EOL);
for (i = 0; i < CTL_DEBUG_MAXID; i++) {
cdp = debugvars[i];
if (cdp->debugname == NULL || cdp->debugvar == NULL)
continue;
snprintf(nodename, sizeof(nodename), "debug%d", i);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTLTYPE_NODE, nodename, NULL,
NULL, 0, NULL, 0,
CTL_DEBUG, i, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTLTYPE_STRING, "name", NULL,
- add const. - remove unnecessary casts. - add __UNCONST casts and mark them with XXXUNCONST as necessary.
2005-05-30 02:24:14 +04:00
/*XXXUNCONST*/
NULL, 0, __UNCONST(cdp->debugname), 0,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTLTYPE_INT, "value", NULL,
NULL, 0, cdp->debugvar, 0,
CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
Tango on sysctl_createv() and flags. The flags have all been renamed, and sysctl_createv() now uses more arguments.
2004-03-24 18:34:46 +03:00
sysctl_createv(clog, 0, NULL, NULL,
CTLFLAG_PERMANENT,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
CTLTYPE_INT, cdp->debugname, NULL,
NULL, 0, cdp->debugvar, 0,
CTL_DEBUG, CTL_CREATE, CTL_EOL);
}
}
#endif /* DEBUG */
/*
* ********************************************************************
* section 2: private node-specific helper routines.
* ********************************************************************
*/
When DIAGNOSTIC is defined, provide a kern.panic_now sysctl to conviniently and reliably panic the system
2006-03-27 00:07:21 +04:00
#ifdef DIAGNOSTIC
static int
sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
{
int newtrig, error;
struct sysctlnode node;
newtrig = 0;
node = *rnode;
node.sysctl_data = &newtrig;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
if (newtrig != 0)
panic("Panic triggered");
return (error);
}
#endif
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.maxvnodes. Drain vnodes if
update a comment to match with the previous change (rev.1.12).
2003-12-20 10:33:03 +03:00
* new value is lower than desiredvnodes and then calls reinit
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* routines that needs to adjust to the new value.
*/
static int
sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
{
restore functionality to decrease kern.maxvnodes which has been backed out during sysctl rework.
2003-12-20 10:26:27 +03:00
int error, new_vnodes, old_vnodes;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
struct sysctlnode node;
new_vnodes = desiredvnodes;
node = *rnode;
node.sysctl_data = &new_vnodes;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
restore functionality to decrease kern.maxvnodes which has been backed out during sysctl rework.
2003-12-20 10:26:27 +03:00
old_vnodes = desiredvnodes;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
desiredvnodes = new_vnodes;
restore functionality to decrease kern.maxvnodes which has been backed out during sysctl rework.
2003-12-20 10:26:27 +03:00
if (new_vnodes < old_vnodes) {
merge ktrace-lwp.
2005-12-11 15:16:03 +03:00
error = vfs_drainvnodes(new_vnodes, l);
restore functionality to decrease kern.maxvnodes which has been backed out during sysctl rework.
2003-12-20 10:26:27 +03:00
if (error) {
desiredvnodes = old_vnodes;
return (error);
}
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
vfs_reinit();
nchreinit();
return (0);
}
Make kern.rtc_offset writable at securelevel <= 0. This allows boot-time adjustment when a machine runs other OSes with RTC == localtime.
2003-12-27 02:49:39 +03:00
/*
* sysctl helper routine for rtc_offset - set time after changes
*/
static int
sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
{
- make settime take timespec. - avoid wrapping of time in settime. - pass struct proc down so that we can log a detailed message.
2005-12-05 03:16:33 +03:00
struct timespec ts, delta;
int error, new_rtc_offset;
Make kern.rtc_offset writable at securelevel <= 0. This allows boot-time adjustment when a machine runs other OSes with RTC == localtime.
2003-12-27 02:49:39 +03:00
struct sysctlnode node;
new_rtc_offset = rtc_offset;
node = *rnode;
node.sysctl_data = &new_rtc_offset;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
First take at security model abstraction. - Add a few scopes to the kernel: system, network, and machdep. - Add a few more actions/sub-actions (requests), and start using them as opposed to the KAUTH_GENERIC_ISSUSER place-holders. - Introduce a basic set of listeners that implement our "traditional" security model, called "bsd44". This is the default (and only) model we have at the moment. - Update all relevant documentation. - Add some code and docs to help folks who want to actually use this stuff: * There's a sample overlay model, sitting on-top of "bsd44", for fast experimenting with tweaking just a subset of an existing model. This is pretty cool because it's *really* straightforward to do stuff you had to use ugly hacks for until now... * And of course, documentation describing how to do the above for quick reference, including code samples. All of these changes were tested for regressions using a Python-based testsuite that will be (I hope) available soon via pkgsrc. Information about the tests, and how to write new ones, can be found on: http://kauth.linbsd.org/kauthwiki NOTE FOR DEVELOPERS: *PLEASE* don't add any code that does any of the following: - Uses a KAUTH_GENERIC_ISSUSER kauth(9) request, - Checks 'securelevel' directly, - Checks a uid/gid directly. (or if you feel you have to, contact me first) This is still work in progress; It's far from being done, but now it'll be a lot easier. Relevant mailing list threads: http://mail-index.netbsd.org/tech-security/2006/01/25/0011.html http://mail-index.netbsd.org/tech-security/2006/03/24/0001.html http://mail-index.netbsd.org/tech-security/2006/04/18/0000.html http://mail-index.netbsd.org/tech-security/2006/05/15/0000.html http://mail-index.netbsd.org/tech-security/2006/08/01/0000.html http://mail-index.netbsd.org/tech-security/2006/08/25/0000.html Many thanks to YAMAMOTO Takashi, Matt Thomas, and Christos Zoulas for help stablizing kauth(9). Full credit for the regression tests, making sure these changes didn't break anything, goes to Matt Fleming and Jaime Fournier. Happy birthday Randi! :)
2006-09-09 00:58:56 +04:00
if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
Don't use KAUTH_RESULT_* where it's not applicable. Prompted by yamt@.
2006-09-13 14:07:42 +04:00
(void *)(u_long)new_rtc_offset, NULL, NULL))
Make kern.rtc_offset writable at securelevel <= 0. This allows boot-time adjustment when a machine runs other OSes with RTC == localtime.
2003-12-27 02:49:39 +03:00
return (EPERM);
if (rtc_offset == new_rtc_offset)
return (0);
/* if we change the offset, adjust the time */
- make settime take timespec. - avoid wrapping of time in settime. - pass struct proc down so that we can log a detailed message.
2005-12-05 03:16:33 +03:00
nanotime(&ts);
delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
delta.tv_nsec = 0;
timespecadd(&ts, &delta, &ts);
Make kern.rtc_offset writable at securelevel <= 0. This allows boot-time adjustment when a machine runs other OSes with RTC == localtime.
2003-12-27 02:49:39 +03:00
rtc_offset = new_rtc_offset;
- make settime take timespec. - avoid wrapping of time in settime. - pass struct proc down so that we can log a detailed message.
2005-12-05 03:16:33 +03:00
settime(l->l_proc, &ts);
Make kern.rtc_offset writable at securelevel <= 0. This allows boot-time adjustment when a machine runs other OSes with RTC == localtime.
2003-12-27 02:49:39 +03:00
return (0);
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.maxproc. Ensures that the new
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* values are not too low or too high.
*/
static int
sysctl_kern_maxproc(SYSCTLFN_ARGS)
{
int error, nmaxproc;
struct sysctlnode node;
nmaxproc = maxproc;
node = *rnode;
node.sysctl_data = &nmaxproc;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
if (nmaxproc < 0 || nmaxproc >= PID_MAX)
return (EINVAL);
#ifdef __HAVE_CPU_MAXPROC
if (nmaxproc > cpu_maxproc())
return (EINVAL);
#endif
maxproc = nmaxproc;
return (0);
}
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper function for kern.hostid. The hostid is a long, but
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* we export it as an int, so we need to give it a little help.
*/
static int
sysctl_kern_hostid(SYSCTLFN_ARGS)
{
int error, inthostid;
struct sysctlnode node;
Fix a comment. Approved by Andrew Brown.
2004-05-03 17:39:50 +04:00
inthostid = hostid; /* XXX assumes sizeof int <= sizeof long */
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
node = *rnode;
node.sysctl_data = &inthostid;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
Prefer that kern.hostid is printed in hex, not as a signed decimal, and avoid accidental sign-extension when setting it.
2004-04-16 17:25:40 +04:00
hostid = (unsigned)inthostid;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
return (0);
}
After changing hostname, adjust hostnamelen. This closes PR kern/23907.
2003-12-28 17:39:36 +03:00
/*
Rename sysctl_kern_hostname() to sysctl_setlen() and use it also for domainname. Note that there's no need to copy rnode since we're not changing any of it, nor protecting anything from change. Thanks to martin for initial work.
2003-12-29 01:19:59 +03:00
* sysctl helper function for kern.hostname and kern.domainnname.
* resets the relevant recorded length when the underlying name is
* changed.
After changing hostname, adjust hostnamelen. This closes PR kern/23907.
2003-12-28 17:39:36 +03:00
*/
static int
Rename sysctl_kern_hostname() to sysctl_setlen() and use it also for domainname. Note that there's no need to copy rnode since we're not changing any of it, nor protecting anything from change. Thanks to martin for initial work.
2003-12-29 01:19:59 +03:00
sysctl_setlen(SYSCTLFN_ARGS)
After changing hostname, adjust hostnamelen. This closes PR kern/23907.
2003-12-28 17:39:36 +03:00
{
int error;
Properly fix the constipated lossage wrt -Wcast-qual and the sysctl code. I know it's not the prettiest code, but it seems to work rather well in spite of itself.
2005-06-09 06:19:59 +04:00
error = sysctl_lookup(SYSCTLFN_CALL(rnode));
After changing hostname, adjust hostnamelen. This closes PR kern/23907.
2003-12-28 17:39:36 +03:00
if (error || newp == NULL)
return (error);
Rename sysctl_kern_hostname() to sysctl_setlen() and use it also for domainname. Note that there's no need to copy rnode since we're not changing any of it, nor protecting anything from change. Thanks to martin for initial work.
2003-12-29 01:19:59 +03:00
switch (rnode->sysctl_num) {
case KERN_HOSTNAME:
hostnamelen = strlen((const char*)rnode->sysctl_data);
break;
case KERN_DOMAINNAME:
domainnamelen = strlen((const char*)rnode->sysctl_data);
break;
}
After changing hostname, adjust hostnamelen. This closes PR kern/23907.
2003-12-28 17:39:36 +03:00
return (0);
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.clockrate. Assembles a struct on
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* the fly to be returned to the caller.
*/
static int
sysctl_kern_clockrate(SYSCTLFN_ARGS)
{
struct clockinfo clkinfo;
struct sysctlnode node;
clkinfo.tick = tick;
clkinfo.tickadj = tickadj;
clkinfo.hz = hz;
clkinfo.profhz = profhz;
clkinfo.stathz = stathz ? stathz : hz;
node = *rnode;
node.sysctl_data = &clkinfo;
return (sysctl_lookup(SYSCTLFN_CALL(&node)));
}
/*
* sysctl helper routine for kern.file pseudo-subtree.
*/
static int
sysctl_kern_file(SYSCTLFN_ARGS)
{
int error;
size_t buflen;
struct file *fp;
char *start, *where;
start = where = oldp;
buflen = *oldlenp;
if (where == NULL) {
/*
* overestimate by 10 files
*/
*oldlenp = sizeof(filehead) + (nfiles + 10) * sizeof(struct file);
return (0);
}
/*
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
* first dcopyout filehead
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
*/
if (buflen < sizeof(filehead)) {
*oldlenp = 0;
return (0);
}
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, &filehead, where, sizeof(filehead));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (error)
return (error);
buflen -= sizeof(filehead);
where += sizeof(filehead);
/*
* followed by an array of file structures
*/
LIST_FOREACH(fp, &filehead, f_list) {
Use the LWP cached credentials where sane.
2006-07-24 02:06:03 +04:00
if (kauth_authorize_generic(l->l_cred,
CURTAIN() -> KAUTH_GENERIC_CANSEE.
2006-07-17 00:21:42 +04:00
KAUTH_GENERIC_CANSEE, fp->f_cred) != 0)
Implement curtain in KERN_{PROC,PROC2,FILE,FILE2,PROC_ARGS}. While I'm here, disable curtain by default.
2005-09-07 21:30:07 +04:00
continue;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (buflen < sizeof(struct file)) {
*oldlenp = where - start;
return (ENOMEM);
}
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, fp, where, sizeof(struct file));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (error)
return (error);
buflen -= sizeof(struct file);
where += sizeof(struct file);
}
*oldlenp = where - start;
return (0);
}
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
* former it merely checks the message buffer is set up. For the latter,
s/the the/the/ (only in sources that aren't regularly imported from elsewhere).
2004-04-23 06:58:27 +04:00
* it also copies out the data if necessary.
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
*/
static int
sysctl_msgbuf(SYSCTLFN_ARGS)
{
char *where = oldp;
size_t len, maxlen;
long beg, end;
int error;
if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
msgbufenabled = 0;
return (ENXIO);
}
switch (rnode->sysctl_num) {
The message buffer datum instrumented by KERN_MSGBUFSIZE is actually a long, not an int, and this causes "problems" on LP64be machines (sparc64, etc). Assign the value to a temporary int and instrument that instead. Should be fine until someone wants a message buffer larger than two gigabytes.
2004-07-27 16:46:18 +04:00
case KERN_MSGBUFSIZE: {
struct sysctlnode node = *rnode;
int msg_bufs = (int)msgbufp->msg_bufs;
node.sysctl_data = &msg_bufs;
return (sysctl_lookup(SYSCTLFN_CALL(&node)));
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
case KERN_MSGBUF:
break;
default:
return (EOPNOTSUPP);
}
if (newp != NULL)
return (EPERM);
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
if (oldp == NULL) {
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/* always return full buffer size */
*oldlenp = msgbufp->msg_bufs;
return (0);
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
error = 0;
maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
/*
* First, copy from the write pointer to the end of
* message buffer.
*/
beg = msgbufp->msg_bufx;
end = msgbufp->msg_bufs;
while (maxlen > 0) {
len = MIN(end - beg, maxlen);
if (len == 0)
break;
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (error)
break;
where += len;
maxlen -= len;
/*
* ... then, copy from the beginning of message buffer to
* the write pointer.
*/
beg = 0;
end = msgbufp->msg_bufx;
}
return (error);
}
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.defcorename. In the case of a new
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* string being assigned, check that it's not a zero-length string.
* (XXX the check in -current doesn't work, but do we really care?)
*/
static int
sysctl_kern_defcorename(SYSCTLFN_ARGS)
{
int error;
for some random places, use PNBUF_GET/PUT rather than - on-stack buffer - malloc(MAXPATHLEN)
2006-02-04 15:09:50 +03:00
char *newcorename;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
struct sysctlnode node;
for some random places, use PNBUF_GET/PUT rather than - on-stack buffer - malloc(MAXPATHLEN)
2006-02-04 15:09:50 +03:00
newcorename = PNBUF_GET();
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
node = *rnode;
node.sysctl_data = &newcorename[0];
memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
error = sysctl_lookup(SYSCTLFN_CALL(&node));
for some random places, use PNBUF_GET/PUT rather than - on-stack buffer - malloc(MAXPATHLEN)
2006-02-04 15:09:50 +03:00
if (error || newp == NULL) {
goto done;
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
* when sysctl_lookup() deals with a string, it's guaranteed
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* to come back nul terminated. So there. :)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
*/
for some random places, use PNBUF_GET/PUT rather than - on-stack buffer - malloc(MAXPATHLEN)
2006-02-04 15:09:50 +03:00
if (strlen(newcorename) == 0) {
error = EINVAL;
} else {
memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
error = 0;
}
done:
PNBUF_PUT(newcorename);
return error;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.cp_time node. Adds up cpu time
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* across all cpus.
*/
static int
sysctl_kern_cptime(SYSCTLFN_ARGS)
{
struct sysctlnode node = *rnode;
Fix kern.cp_time for MULTIPROCESSOR kernels: calculate size of result correctly, free original instead of incremented pointer, copy results for n = -2 case too, so top shows correct stats. Additionaly, rearange code for better readability (from Andrew).
2003-12-07 00:33:51 +03:00
#ifndef MULTIPROCESSOR
Adjust error returns in kern.cp_time when a specific processor is being requested so that (1) the uniprocessor case and the multiprocessor case are more similar and (2) so that we return ENOENT when a non-existent processor is requested (which is both more sensible and follows the general order of things anyway).
2003-12-29 01:24:12 +03:00
if (namelen == 1) {
if (name[0] != 0)
return (ENOENT);
Fix kern.cp_time for MULTIPROCESSOR kernels: calculate size of result correctly, free original instead of incremented pointer, copy results for n = -2 case too, so top shows correct stats. Additionaly, rearange code for better readability (from Andrew).
2003-12-07 00:33:51 +03:00
/*
* you're allowed to ask for the zero'th processor
*/
name++;
namelen--;
}
node.sysctl_data = curcpu()->ci_schedstate.spc_cp_time;
node.sysctl_size = sizeof(curcpu()->ci_schedstate.spc_cp_time);
return (sysctl_lookup(SYSCTLFN_CALL(&node)));
#else /* MULTIPROCESSOR */
u_intN_t -> uintN_t
2005-12-26 21:41:36 +03:00
uint64_t *cp_time = NULL;
Count the number of CPUs at boot and stash in 'ncpu'. Eventually should have each CPU register at attach, so we can figure out the topology for the scheduler.
2007-02-15 23:32:47 +03:00
int error, n = ncpu, i;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
struct cpu_info *ci;
CPU_INFO_ITERATOR cii;
/*
* if you specifically pass a buffer that is the size of the
* sum, or if you are probing for the size, you get the "sum"
* of cp_time (and the size thereof) across all processors.
*
* alternately, you can pass an additional mib number and get
* cp_time for that particular processor.
*/
switch (namelen) {
case 0:
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
u_intN_t -> uintN_t
2005-12-26 21:41:36 +03:00
node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
n = -1; /* SUM */
}
else {
u_intN_t -> uintN_t
2005-12-26 21:41:36 +03:00
node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
n = -2; /* ALL */
}
break;
case 1:
if (name[0] < 0 || name[0] >= n)
Adjust error returns in kern.cp_time when a specific processor is being requested so that (1) the uniprocessor case and the multiprocessor case are more similar and (2) so that we return ENOENT when a non-existent processor is requested (which is both more sensible and follows the general order of things anyway).
2003-12-29 01:24:12 +03:00
return (ENOENT); /* ENOSUCHPROCESSOR */
u_intN_t -> uintN_t
2005-12-26 21:41:36 +03:00
node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
n = name[0];
/*
* adjust these so that sysctl_lookup() will be happy
*/
name++;
namelen--;
Add missing break.
2003-12-07 13:33:03 +03:00
break;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
default:
return (EINVAL);
}
Fix kern.cp_time for MULTIPROCESSOR kernels: calculate size of result correctly, free original instead of incremented pointer, copy results for n = -2 case too, so top shows correct stats. Additionaly, rearange code for better readability (from Andrew).
2003-12-07 00:33:51 +03:00
cp_time = malloc(node.sysctl_size, M_TEMP, M_WAITOK|M_CANFAIL);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (cp_time == NULL)
return (ENOMEM);
node.sysctl_data = cp_time;
memset(cp_time, 0, node.sysctl_size);
for (CPU_INFO_FOREACH(cii, ci)) {
Fix kern.cp_time for MULTIPROCESSOR kernels: calculate size of result correctly, free original instead of incremented pointer, copy results for n = -2 case too, so top shows correct stats. Additionaly, rearange code for better readability (from Andrew).
2003-12-07 00:33:51 +03:00
if (n <= 0)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
for (i = 0; i < CPUSTATES; i++)
cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
/*
* if a specific processor was requested and we just
* did it, we're done here
*/
if (n == 0)
break;
/*
* if doing "all", skip to next cp_time set for next processor
*/
if (n == -2)
cp_time += CPUSTATES;
/*
* if we're doing a specific processor, we're one
* processor closer
*/
if (n > 0)
n--;
}
error = sysctl_lookup(SYSCTLFN_CALL(&node));
Fix kern.cp_time for MULTIPROCESSOR kernels: calculate size of result correctly, free original instead of incremented pointer, copy results for n = -2 case too, so top shows correct stats. Additionaly, rearange code for better readability (from Andrew).
2003-12-07 00:33:51 +03:00
free(node.sysctl_data, M_TEMP);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
return (error);
Fix kern.cp_time for MULTIPROCESSOR kernels: calculate size of result correctly, free original instead of incremented pointer, copy results for n = -2 case too, so top shows correct stats. Additionaly, rearange code for better readability (from Andrew).
2003-12-07 00:33:51 +03:00
#endif /* MULTIPROCESSOR */
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
#if NPTY > 0
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.maxptys. Ensures that any new value
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* is acceptable to the pty subsystem.
*/
static int
sysctl_kern_maxptys(SYSCTLFN_ARGS)
{
int pty_maxptys(int, int); /* defined in kern/tty_pty.c */
- add const. - remove unnecessary casts. - add __UNCONST casts and mark them with XXXUNCONST as necessary.
2005-05-30 02:24:14 +04:00
int error, xmax;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
struct sysctlnode node;
/* get current value of maxptys */
- add const. - remove unnecessary casts. - add __UNCONST casts and mark them with XXXUNCONST as necessary.
2005-05-30 02:24:14 +04:00
xmax = pty_maxptys(0, 0);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
node = *rnode;
- add const. - remove unnecessary casts. - add __UNCONST casts and mark them with XXXUNCONST as necessary.
2005-05-30 02:24:14 +04:00
node.sysctl_data = &xmax;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
nuke trailing whitespace
2005-02-27 00:34:55 +03:00
- add const. - remove unnecessary casts. - add __UNCONST casts and mark them with XXXUNCONST as necessary.
2005-05-30 02:24:14 +04:00
if (xmax != pty_maxptys(xmax, 1))
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
return (EINVAL);
return (0);
}
#endif /* NPTY > 0 */
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.sbmax. Basically just ensures that
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* any new value is not too small.
*/
static int
sysctl_kern_sbmax(SYSCTLFN_ARGS)
{
int error, new_sbmax;
struct sysctlnode node;
new_sbmax = sb_max;
node = *rnode;
node.sysctl_data = &new_sbmax;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
error = sb_max_set(new_sbmax);
return (error);
}
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.urandom node. Picks a random number
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* for you.
*/
static int
sysctl_kern_urnd(SYSCTLFN_ARGS)
{
#if NRND > 0
int v;
if (rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY) == sizeof(v)) {
struct sysctlnode node = *rnode;
node.sysctl_data = &v;
return (sysctl_lookup(SYSCTLFN_CALL(&node)));
}
else
return (EIO); /*XXX*/
#else
return (EOPNOTSUPP);
#endif
}
implement kern.arandom properly, instead of lying about it and only filling the first 4 bytes of the array with random data.
2006-11-02 01:27:43 +03:00
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.arandom node. Picks a random number
implement kern.arandom properly, instead of lying about it and only filling the first 4 bytes of the array with random data.
2006-11-02 01:27:43 +03:00
* for you.
*/
static int
sysctl_kern_arnd(SYSCTLFN_ARGS)
{
#if NRND > 0
int error;
void *v;
struct sysctlnode node = *rnode;
if (*oldlenp == 0)
return 0;
if (*oldlenp > 8192)
return E2BIG;
v = malloc(*oldlenp, M_TEMP, M_WAITOK);
arc4randbytes(v, *oldlenp);
node.sysctl_data = v;
node.sysctl_size = *oldlenp;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
free(v, M_TEMP);
return error;
#else
return (EOPNOTSUPP);
#endif
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
* sysctl helper routine to do kern.lwp.* work.
*/
static int
sysctl_kern_lwp(SYSCTLFN_ARGS)
{
struct kinfo_lwp klwp;
struct proc *p;
struct lwp *l2;
char *where, *dp;
int pid, elem_size, elem_count;
int buflen, needed, error;
Sysctl functions called for "generic" nodes should forward "query" requests (where possible), rather than returning errors.
2003-12-29 01:36:37 +03:00
if (namelen == 1 && name[0] == CTL_QUERY)
Properly fix the constipated lossage wrt -Wcast-qual and the sysctl code. I know it's not the prettiest code, but it seems to work rather well in spite of itself.
2005-06-09 06:19:59 +04:00
return (sysctl_query(SYSCTLFN_CALL(rnode)));
Sysctl functions called for "generic" nodes should forward "query" requests (where possible), rather than returning errors.
2003-12-29 01:36:37 +03:00
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
dp = where = oldp;
buflen = where != NULL ? *oldlenp : 0;
error = needed = 0;
In sysctl_kern_lwp adjust offsets into the mib entries so that they are now correct. Fixes problems with "ps -s" not working. Also use KERN_LWPSLOP instead of KERN_PROCSLOP. Both changes from Andrew Brown.
2003-12-13 02:21:44 +03:00
if (newp != NULL || namelen != 3)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
return (EINVAL);
In sysctl_kern_lwp adjust offsets into the mib entries so that they are now correct. Fixes problems with "ps -s" not working. Also use KERN_LWPSLOP instead of KERN_PROCSLOP. Both changes from Andrew Brown.
2003-12-13 02:21:44 +03:00
pid = name[0];
elem_size = name[1];
elem_count = name[2];
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_enter(&proclist_lock);
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
if (pid == -1) {
LIST_FOREACH(l2, &alllwp, l_list) {
if (buflen >= elem_size && elem_count > 0) {
lwp_lock(l2);
fill_lwp(l2, &klwp);
lwp_unlock(l2);
/*
* Copy out elem_size, but not larger than
* the size of a struct kinfo_proc2.
*/
error = dcopyout(l, &klwp, dp,
min(sizeof(klwp), elem_size));
if (error)
goto cleanup;
dp += elem_size;
buflen -= elem_size;
elem_count--;
}
needed += elem_size;
}
} else {
p = p_find(pid, PFIND_LOCKED);
if (p == NULL) {
mutex_exit(&proclist_lock);
return (ESRCH);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
mutex_enter(&p->p_smutex);
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
if (buflen >= elem_size && elem_count > 0) {
Merge from vmlocking: don't hold scheduler locks across copyout().
2007-10-08 22:09:37 +04:00
struct lwp *l3;
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
lwp_lock(l2);
fill_lwp(l2, &klwp);
lwp_unlock(l2);
Merge from vmlocking: don't hold scheduler locks across copyout().
2007-10-08 22:09:37 +04:00
mutex_exit(&p->p_smutex);
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
/*
* Copy out elem_size, but not larger than
* the size of a struct kinfo_proc2.
*/
error = dcopyout(l, &klwp, dp,
min(sizeof(klwp), elem_size));
Merge from vmlocking: don't hold scheduler locks across copyout().
2007-10-08 22:09:37 +04:00
if (error) {
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
goto cleanup;
Merge from vmlocking: don't hold scheduler locks across copyout().
2007-10-08 22:09:37 +04:00
}
mutex_enter(&p->p_smutex);
LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
if (l2 == l3)
break;
}
sysctl_kern_lwp: Use a correct variable when rechecking if LWP still exists after relocking. Found via CID: 4689. OK by <dsl>.
2007-10-13 14:04:08 +04:00
if (l3 == NULL) {
Merge from vmlocking: don't hold scheduler locks across copyout().
2007-10-08 22:09:37 +04:00
mutex_exit(&p->p_smutex);
error = EAGAIN;
goto cleanup;
}
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
dp += elem_size;
buflen -= elem_size;
elem_count--;
}
needed += elem_size;
}
mutex_exit(&p->p_smutex);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_exit(&proclist_lock);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (where != NULL) {
*oldlenp = dp - where;
if (needed > *oldlenp)
return (ENOMEM);
} else {
Use KERN_PROCSLOP for struct kinfo_proc and KERN_LWPSLOP for struct kinfo_lwp, and not vice versa. Should solve the issue with top dying because it's unable to "allocate memory".
2004-02-21 06:27:57 +03:00
needed += KERN_LWPSLOP;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
*oldlenp = needed;
}
return (0);
cleanup:
Merge from vmlocking: don't hold scheduler locks across copyout().
2007-10-08 22:09:37 +04:00
mutex_exit(&proclist_lock);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
return (error);
}
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.forkfsleep node. Ensures that the
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* given value is not too large or two small, and is at least one
* timer tick if not zero.
*/
static int
sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
{
/* userland sees value in ms, internally is in ticks */
extern int forkfsleep; /* defined in kern/kern_fork.c */
int error, timo, lsleep;
struct sysctlnode node;
lsleep = forkfsleep * 1000 / hz;
node = *rnode;
node.sysctl_data = &lsleep;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
/* refuse negative values, and overly 'long time' */
if (lsleep < 0 || lsleep > MAXSLP * 1000)
return (EINVAL);
timo = mstohz(lsleep);
/* if the interval is >0 ms && <1 tick, use 1 tick */
if (lsleep != 0 && timo == 0)
forkfsleep = 1;
else
forkfsleep = timo;
return (0);
}
/*
* sysctl helper routine for kern.root_partition
*/
static int
sysctl_kern_root_partition(SYSCTLFN_ARGS)
{
int rootpart = DISKPART(rootdev);
struct sysctlnode node = *rnode;
node.sysctl_data = &rootpart;
return (sysctl_lookup(SYSCTLFN_CALL(&node)));
}
/*
* sysctl helper function for kern.drivers
*/
static int
sysctl_kern_drivers(SYSCTLFN_ARGS)
{
int error;
size_t buflen;
struct kinfo_drivers kd;
char *start, *where;
const char *dname;
int i;
extern struct devsw_conv *devsw_conv;
extern int max_devsw_convs;
if (newp != NULL || namelen != 0)
return (EINVAL);
start = where = oldp;
buflen = *oldlenp;
if (where == NULL) {
*oldlenp = max_devsw_convs * sizeof kd;
return 0;
}
/*
* An array of kinfo_drivers structures
*/
error = 0;
for (i = 0; i < max_devsw_convs; i++) {
dname = devsw_conv[i].d_name;
if (dname == NULL)
continue;
if (buflen < sizeof kd) {
error = ENOMEM;
break;
}
Clear out the struct kinfo_drivers before stuffing things into it. Avoids leaking garbage from the stack (left over from the earlier call to sysctl_locate()).
2004-04-08 07:35:10 +04:00
memset(&kd, 0, sizeof(kd));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
kd.d_bmajor = devsw_conv[i].d_bmajor;
kd.d_cmajor = devsw_conv[i].d_cmajor;
strlcpy(kd.d_name, dname, sizeof kd.d_name);
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, &kd, where, sizeof kd);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (error != 0)
break;
buflen -= sizeof kd;
where += sizeof kd;
}
*oldlenp = where - start;
return error;
}
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
/*
* sysctl helper function for kern.file2
*/
static int
sysctl_kern_file2(SYSCTLFN_ARGS)
{
struct proc *p;
struct file *fp;
struct filedesc *fd;
struct kinfo_file kf;
char *dp;
u_int i, op;
size_t len, needed, elem_size, out_size;
int error, arg, elem_count;
if (namelen == 1 && name[0] == CTL_QUERY)
Properly fix the constipated lossage wrt -Wcast-qual and the sysctl code. I know it's not the prettiest code, but it seems to work rather well in spite of itself.
2005-06-09 06:19:59 +04:00
return (sysctl_query(SYSCTLFN_CALL(rnode)));
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
if (namelen != 4)
return (EINVAL);
error = 0;
dp = oldp;
len = (oldp != NULL) ? *oldlenp : 0;
op = name[0];
arg = name[1];
elem_size = name[2];
elem_count = name[3];
out_size = MIN(sizeof(kf), elem_size);
needed = 0;
if (elem_size < 1 || elem_count < 0)
return (EINVAL);
switch (op) {
case KERN_FILE_BYFILE:
/*
* doesn't use arg so it must be zero
*/
if (arg != 0)
return (EINVAL);
LIST_FOREACH(fp, &filehead, f_list) {
Use the LWP cached credentials where sane.
2006-07-24 02:06:03 +04:00
if (kauth_authorize_generic(l->l_cred,
CURTAIN() -> KAUTH_GENERIC_CANSEE.
2006-07-17 00:21:42 +04:00
KAUTH_GENERIC_CANSEE, fp->f_cred) != 0)
Implement curtain in KERN_{PROC,PROC2,FILE,FILE2,PROC_ARGS}. While I'm here, disable curtain by default.
2005-09-07 21:30:07 +04:00
continue;
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
if (len >= elem_size && elem_count > 0) {
fill_file(&kf, fp, NULL, 0);
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, &kf, dp, out_size);
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
if (error)
break;
dp += elem_size;
len -= elem_size;
}
if (elem_count > 0) {
needed += elem_size;
if (elem_count != INT_MAX)
elem_count--;
}
}
break;
sysctl_kern_file2: fix an indent.
2006-06-13 17:23:03 +04:00
case KERN_FILE_BYPID:
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
if (arg < -1)
/* -1 means all processes */
return (EINVAL);
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_enter(&proclist_lock);
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
PROCLIST_FOREACH(p, &allproc) {
if (p->p_stat == SIDL)
/* skip embryonic processes */
continue;
Use the LWP cached credentials where sane.
2006-07-24 02:06:03 +04:00
if (kauth_authorize_process(l->l_cred,
remove unnecessary arguments from kauth_authorize_process. ie. make it similar to the one found in apple TN.
2006-06-13 17:56:50 +04:00
KAUTH_PROCESS_CANSEE, p, NULL, NULL, NULL) != 0)
Implement curtain in KERN_{PROC,PROC2,FILE,FILE2,PROC_ARGS}. While I'm here, disable curtain by default.
2005-09-07 21:30:07 +04:00
continue;
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
if (arg > 0 && p->p_pid != arg)
/* pick only the one we want */
/* XXX want 0 to mean "kernel files" */
continue;
fd = p->p_fd;
for (i = 0; i < fd->fd_nfiles; i++) {
fp = fd->fd_ofiles[i];
if (fp == NULL || !FILE_IS_USABLE(fp))
continue;
if (len >= elem_size && elem_count > 0) {
fill_file(&kf, fd->fd_ofiles[i],
p, i);
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, &kf, dp, out_size);
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
if (error)
break;
dp += elem_size;
len -= elem_size;
}
if (elem_count > 0) {
needed += elem_size;
if (elem_count != INT_MAX)
elem_count--;
}
}
}
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_exit(&proclist_lock);
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
break;
default:
return (EINVAL);
}
if (oldp == NULL)
needed += KERN_FILESLOP * elem_size;
*oldlenp = needed;
return (error);
}
static void
fill_file(struct kinfo_file *kp, const struct file *fp, struct proc *p, int i)
{
memset(kp, 0, sizeof(*kp));
kp->ki_fileaddr = PTRTOUINT64(fp);
kp->ki_flag = fp->f_flag;
kp->ki_iflags = fp->f_iflags;
kp->ki_ftype = fp->f_type;
kp->ki_count = fp->f_count;
kp->ki_msgcount = fp->f_msgcount;
kp->ki_usecount = fp->f_usecount;
kp->ki_fucred = PTRTOUINT64(fp->f_cred);
integrate kauth.
2006-05-15 01:15:11 +04:00
kp->ki_fuid = kauth_cred_geteuid(fp->f_cred);
kp->ki_fgid = kauth_cred_getegid(fp->f_cred);
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
kp->ki_fops = PTRTOUINT64(fp->f_ops);
kp->ki_foffset = fp->f_offset;
kp->ki_fdata = PTRTOUINT64(fp->f_data);
/* vnode information to glue this file to something */
if (fp->f_type == DTYPE_VNODE) {
struct vnode *vp = (struct vnode *)fp->f_data;
kp->ki_vun = PTRTOUINT64(vp->v_un.vu_socket);
kp->ki_vsize = vp->v_size;
kp->ki_vtype = vp->v_type;
kp->ki_vtag = vp->v_tag;
kp->ki_vdata = PTRTOUINT64(vp->v_data);
}
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
/* process information when retrieved via KERN_FILE_BYPID */
Add kern.file2. As kern.proc2 is to kern.proc, so is kern.file2 to kern.file, namely a 32/64 bit clean sysctl interface to the same data. It also borrows a few things from struct vnode (if applicable) and from struct proc, just to tie things together a bit more. You can walk this list "by file" or "by pid". The former method is similar to kern.file but omits the filehead, and the latter can give you duplicates if multiple processes have the same struct file open, but tells you which process it is.
2005-03-09 08:02:17 +03:00
if (p) {
kp->ki_pid = p->p_pid;
kp->ki_fd = i;
kp->ki_ofileflags = p->p_fd->fd_ofileflags[i];
}
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
static int
sysctl_doeproc(SYSCTLFN_ARGS)
{
don't allocate too much stuff on the stack.
2006-06-20 07:20:44 +04:00
struct eproc *eproc;
struct kinfo_proc2 *kproc2;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
struct kinfo_proc *dp;
struct proc *p;
const struct proclist_desc *pd;
char *where, *dp2;
int type, op, arg;
u_int elem_size, elem_count;
size_t buflen, needed;
int error;
Sysctl functions called for "generic" nodes should forward "query" requests (where possible), rather than returning errors.
2003-12-29 01:36:37 +03:00
if (namelen == 1 && name[0] == CTL_QUERY)
Properly fix the constipated lossage wrt -Wcast-qual and the sysctl code. I know it's not the prettiest code, but it seems to work rather well in spite of itself.
2005-06-09 06:19:59 +04:00
return (sysctl_query(SYSCTLFN_CALL(rnode)));
Sysctl functions called for "generic" nodes should forward "query" requests (where possible), rather than returning errors.
2003-12-29 01:36:37 +03:00
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
dp = oldp;
dp2 = where = oldp;
buflen = where != NULL ? *oldlenp : 0;
error = 0;
needed = 0;
type = rnode->sysctl_num;
if (type == KERN_PROC) {
if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
return (EINVAL);
op = name[0];
if (op != KERN_PROC_ALL)
arg = name[1];
else
arg = 0; /* Quell compiler warning */
elem_size = elem_count = 0; /* Ditto */
} else {
if (namelen != 4)
return (EINVAL);
op = name[0];
arg = name[1];
elem_size = name[2];
elem_count = name[3];
}
Don't leak memory on success. Allocate only the type of struct that we'll need for efficiency.
2006-06-21 17:46:17 +04:00
if (type == KERN_PROC) {
eproc = malloc(sizeof(*eproc), M_TEMP, M_WAITOK);
kproc2 = NULL;
} else {
eproc = NULL;
kproc2 = malloc(sizeof(*kproc2), M_TEMP, M_WAITOK);
}
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_enter(&proclist_lock);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
pd = proclists;
again:
introduce a function, proclist_foreach_call, to iterate all procs on a proclist and call the specified function for each of them. primarily to fix a procfs locking problem, but i think that it's useful for others as well. while i'm here, introduce PROCLIST_FOREACH macro, which is similar to LIST_FOREACH but skips marker entries which are used by proclist_foreach_call.
2004-10-01 20:30:52 +04:00
PROCLIST_FOREACH(p, pd->pd_list) {
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
* Skip embryonic processes.
*/
if (p->p_stat == SIDL)
continue;
Implement curtain in KERN_{PROC,PROC2,FILE,FILE2,PROC_ARGS}. While I'm here, disable curtain by default.
2005-09-07 21:30:07 +04:00
Use the LWP cached credentials where sane.
2006-07-24 02:06:03 +04:00
if (kauth_authorize_process(l->l_cred,
remove unnecessary arguments from kauth_authorize_process. ie. make it similar to the one found in apple TN.
2006-06-13 17:56:50 +04:00
KAUTH_PROCESS_CANSEE, p, NULL, NULL, NULL) != 0)
Implement curtain in KERN_{PROC,PROC2,FILE,FILE2,PROC_ARGS}. While I'm here, disable curtain by default.
2005-09-07 21:30:07 +04:00
continue;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
* TODO - make more efficient (see notes below).
* do by session.
*/
switch (op) {
case KERN_PROC_PID:
/* could do this with just a lookup */
if (p->p_pid != (pid_t)arg)
continue;
break;
case KERN_PROC_PGRP:
/* could do this by traversing pgrp */
if (p->p_pgrp->pg_id != (pid_t)arg)
continue;
break;
case KERN_PROC_SESSION:
if (p->p_session->s_sid != (pid_t)arg)
continue;
break;
case KERN_PROC_TTY:
if (arg == (int) KERN_PROC_TTY_REVOKE) {
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
if ((p->p_lflag & PL_CONTROLT) == 0 ||
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
p->p_session->s_ttyp == NULL ||
p->p_session->s_ttyvp != NULL)
continue;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
} else if ((p->p_lflag & PL_CONTROLT) == 0 ||
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
p->p_session->s_ttyp == NULL) {
if ((dev_t)arg != KERN_PROC_TTY_NODEV)
continue;
} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg)
continue;
break;
case KERN_PROC_UID:
integrate kauth.
2006-05-15 01:15:11 +04:00
if (kauth_cred_geteuid(p->p_cred) != (uid_t)arg)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
continue;
break;
case KERN_PROC_RUID:
integrate kauth.
2006-05-15 01:15:11 +04:00
if (kauth_cred_getuid(p->p_cred) != (uid_t)arg)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
continue;
break;
case KERN_PROC_GID:
integrate kauth.
2006-05-15 01:15:11 +04:00
if (kauth_cred_getegid(p->p_cred) != (uid_t)arg)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
continue;
break;
case KERN_PROC_RGID:
integrate kauth.
2006-05-15 01:15:11 +04:00
if (kauth_cred_getgid(p->p_cred) != (uid_t)arg)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
continue;
break;
case KERN_PROC_ALL:
/* allow everything */
break;
default:
error = EINVAL;
goto cleanup;
}
if (type == KERN_PROC) {
if (buflen >= sizeof(struct kinfo_proc)) {
don't allocate too much stuff on the stack.
2006-06-20 07:20:44 +04:00
fill_eproc(p, eproc);
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, p, &dp->kp_proc,
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sizeof(struct proc));
if (error)
goto cleanup;
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, eproc, &dp->kp_eproc,
don't allocate too much stuff on the stack.
2006-06-20 07:20:44 +04:00
sizeof(*eproc));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (error)
goto cleanup;
dp++;
buflen -= sizeof(struct kinfo_proc);
}
needed += sizeof(struct kinfo_proc);
} else { /* KERN_PROC2 */
if (buflen >= elem_size && elem_count > 0) {
don't allocate too much stuff on the stack.
2006-06-20 07:20:44 +04:00
fill_kproc2(p, kproc2);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
* Copy out elem_size, but not larger than
* the size of a struct kinfo_proc2.
*/
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, kproc2, dp2,
don't allocate too much stuff on the stack.
2006-06-20 07:20:44 +04:00
min(sizeof(*kproc2), elem_size));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (error)
goto cleanup;
dp2 += elem_size;
buflen -= elem_size;
elem_count--;
}
needed += elem_size;
}
}
pd++;
if (pd->pd_list != NULL)
goto again;
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_exit(&proclist_lock);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (where != NULL) {
if (type == KERN_PROC)
*oldlenp = (char *)dp - where;
else
*oldlenp = dp2 - where;
don't allocate too much stuff on the stack.
2006-06-20 07:20:44 +04:00
if (needed > *oldlenp) {
error = ENOMEM;
goto out;
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
} else {
Use KERN_PROCSLOP for struct kinfo_proc and KERN_LWPSLOP for struct kinfo_lwp, and not vice versa. Should solve the issue with top dying because it's unable to "allocate memory".
2004-02-21 06:27:57 +03:00
needed += KERN_PROCSLOP;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
*oldlenp = needed;
}
Don't leak memory on success. Allocate only the type of struct that we'll need for efficiency.
2006-06-21 17:46:17 +04:00
if (kproc2)
free(kproc2, M_TEMP);
if (eproc)
free(eproc, M_TEMP);
return 0;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
cleanup:
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_exit(&proclist_lock);
don't allocate too much stuff on the stack.
2006-06-20 07:20:44 +04:00
out:
Don't leak memory on success. Allocate only the type of struct that we'll need for efficiency.
2006-06-21 17:46:17 +04:00
if (kproc2)
free(kproc2, M_TEMP);
if (eproc)
free(eproc, M_TEMP);
return error;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
/*
* sysctl helper routine for kern.proc_args pseudo-subtree.
*/
static int
sysctl_kern_proc_args(SYSCTLFN_ARGS)
{
struct ps_strings pss;
Use the LWP cached credentials where sane.
2006-07-24 02:06:03 +04:00
struct proc *p;
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
size_t len, i;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
struct uio auio;
struct iovec aiov;
pid_t pid;
int nargv, type, error;
char *arg;
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
char **argv = NULL;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
char *tmp;
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
struct vmspace *vmspace;
vaddr_t psstr_addr;
vaddr_t offsetn;
vaddr_t offsetv;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
Sysctl functions called for "generic" nodes should forward "query" requests (where possible), rather than returning errors.
2003-12-29 01:36:37 +03:00
if (namelen == 1 && name[0] == CTL_QUERY)
Properly fix the constipated lossage wrt -Wcast-qual and the sysctl code. I know it's not the prettiest code, but it seems to work rather well in spite of itself.
2005-06-09 06:19:59 +04:00
return (sysctl_query(SYSCTLFN_CALL(rnode)));
Sysctl functions called for "generic" nodes should forward "query" requests (where possible), rather than returning errors.
2003-12-29 01:36:37 +03:00
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (newp != NULL || namelen != 2)
return (EINVAL);
pid = name[0];
type = name[1];
switch (type) {
case KERN_PROC_ARGV:
case KERN_PROC_NARGV:
case KERN_PROC_ENV:
case KERN_PROC_NENV:
/* ok */
break;
default:
return (EINVAL);
}
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_enter(&proclist_lock);
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/* check pid */
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
if ((p = p_find(pid, PFIND_LOCKED)) == NULL) {
error = EINVAL;
goto out_locked;
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
Use the LWP cached credentials where sane.
2006-07-24 02:06:03 +04:00
error = kauth_authorize_process(l->l_cred,
remove unnecessary arguments from kauth_authorize_process. ie. make it similar to the one found in apple TN.
2006-06-13 17:56:50 +04:00
KAUTH_PROCESS_CANSEE, p, NULL, NULL, NULL);
if (error) {
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
goto out_locked;
}
Implement curtain in KERN_{PROC,PROC2,FILE,FILE2,PROC_ARGS}. While I'm here, disable curtain by default.
2005-09-07 21:30:07 +04:00
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/* only root or same user change look at the environment */
if (type == KERN_PROC_ENV || type == KERN_PROC_NENV) {
Don't rely on KAUTH_PROCESS_CANSEE for environment just yet, otherwise we're allowing anyone to read the environment unless curtain is enabled. From yamt@.
2007-01-22 18:11:52 +03:00
if (kauth_authorize_generic(l->l_cred, KAUTH_GENERIC_ISSUSER,
NULL) != 0) {
if (kauth_cred_getuid(l->l_cred) !=
kauth_cred_getuid(p->p_cred) ||
kauth_cred_getuid(l->l_cred) !=
kauth_cred_getsvuid(p->p_cred)) {
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
error = EPERM;
goto out_locked;
Don't rely on KAUTH_PROCESS_CANSEE for environment just yet, otherwise we're allowing anyone to read the environment unless curtain is enabled. From yamt@.
2007-01-22 18:11:52 +03:00
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
}
if (oldp == NULL) {
if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
*oldlenp = sizeof (int);
else
*oldlenp = ARG_MAX; /* XXX XXX XXX */
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
error = 0;
goto out_locked;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
/*
* Zombies don't have a stack, so we can't read their psstrings.
* System processes also don't have a user stack.
*/
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) {
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
error = EINVAL;
goto out_locked;
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
* Lock the process down in memory.
*/
/* XXXCDC: how should locking work here? */
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
if ((l->l_flag & LW_WEXIT) || (p->p_vmspace->vm_refcnt < 1)) {
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
error = EFAULT;
goto out_locked;
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
psstr_addr = (vaddr_t)p->p_psstr;
if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) {
offsetn = p->p_psnargv;
offsetv = p->p_psargv;
} else {
offsetn = p->p_psnenv;
offsetv = p->p_psenv;
}
vmspace = p->p_vmspace;
vmspace->vm_refcnt++; /* XXX */
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_exit(&proclist_lock);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
* Allocate a temporary buffer to hold the arguments.
*/
arg = malloc(PAGE_SIZE, M_TEMP, M_WAITOK);
/*
* Read in the ps_strings structure.
*/
aiov.iov_base = &pss;
aiov.iov_len = sizeof(pss);
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
auio.uio_offset = psstr_addr;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
auio.uio_resid = sizeof(pss);
auio.uio_rw = UIO_READ;
merge yamt-uio_vmspace branch. - use vmspace rather than proc or lwp where appropriate. the latter is more natural to specify an address space. (and less likely to be abused for random purposes.) - fix a swdmover race.
2006-03-01 15:38:10 +03:00
UIO_SETUP_SYSSPACE(&auio);
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
error = uvm_io(&vmspace->vm_map, &auio);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (error)
goto done;
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
memcpy(&nargv, (char *)&pss + offsetn, sizeof(nargv));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, &nargv, oldp, sizeof(nargv));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
*oldlenp = sizeof(nargv);
goto done;
}
/*
* Now read the address of the argument vector.
*/
switch (type) {
case KERN_PROC_ARGV:
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
/* FALLTHROUGH */
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
case KERN_PROC_ENV:
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
memcpy(&tmp, (char *)&pss + offsetv, sizeof(tmp));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
break;
default:
return (EINVAL);
}
When colecting a 32 bit process' argument or environement vector, we need to convert 32 bits pointers to the 64 bit environement
2006-09-08 15:59:52 +04:00
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
#ifdef COMPAT_NETBSD32
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
if (p->p_flag & PK_32)
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
len = sizeof(netbsd32_charp) * nargv;
else
#endif
len = sizeof(char *) * nargv;
argv = malloc(len, M_TEMP, M_WAITOK);
aiov.iov_base = argv;
aiov.iov_len = len;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
auio.uio_offset = (off_t)(unsigned long)tmp;
auio.uio_resid = len;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
auio.uio_rw = UIO_READ;
merge yamt-uio_vmspace branch. - use vmspace rather than proc or lwp where appropriate. the latter is more natural to specify an address space. (and less likely to be abused for random purposes.) - fix a swdmover race.
2006-03-01 15:38:10 +03:00
UIO_SETUP_SYSSPACE(&auio);
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
error = uvm_io(&vmspace->vm_map, &auio);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (error)
goto done;
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
/*
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
* Now copy each string.
When colecting a 32 bit process' argument or environement vector, we need to convert 32 bits pointers to the 64 bit environement
2006-09-08 15:59:52 +04:00
*/
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
len = 0; /* bytes written to user buffer */
for (i = 0; i < nargv; i++) {
int finished = 0;
vaddr_t base;
size_t xlen;
int j;
When colecting a 32 bit process' argument or environement vector, we need to convert 32 bits pointers to the 64 bit environement
2006-09-08 15:59:52 +04:00
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
#ifdef COMPAT_NETBSD32
Change the process/lwp flags seen by userland via sysctl back to the P_*/L_* naming convention, and rename the in-kernel flags to avoid conflict. (P_ -> PK_, L_ -> LW_ ). Add back the (now unused) LSDEAD constant. Restores source compatibility with pre-newlock2 tools like ps or top. Reviewed by Andrew Doran.
2007-02-18 01:31:36 +03:00
if (p->p_flag & PK_32) {
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
netbsd32_charp *argv32;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
argv32 = (netbsd32_charp *)argv;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
base = (vaddr_t)NETBSD32PTR64(argv32[i]);
} else
#endif
base = (vaddr_t)argv[i];
Don't fail to produce the argument vector if the program has modified it by deleting arguments. This is a popular practice, and failing means that ps(1) prints (programname). For example this is what XtOpenDisplay() with -geometry. This used to work before 2.0H, and the behavior is allowed and hinted by POSIX. Found out by Anon Ymous.
2007-10-16 20:10:59 +04:00
/*
* The program has messed around with its arguments,
* possibly deleting some, and replacing them with
* NULL's. Treat this as the last argument and not
* a failure.
*/
if (base == 0)
break;
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
while (!finished) {
xlen = PAGE_SIZE - (base & PAGE_MASK);
aiov.iov_base = arg;
aiov.iov_len = PAGE_SIZE;
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
auio.uio_offset = base;
auio.uio_resid = xlen;
auio.uio_rw = UIO_READ;
UIO_SETUP_SYSSPACE(&auio);
error = uvm_io(&vmspace->vm_map, &auio);
if (error)
goto done;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
/* Look for the end of the string */
for (j = 0; j < xlen; j++) {
if (arg[j] == '\0') {
xlen = j + 1;
finished = 1;
break;
}
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
/* Check for user buffer overflow */
if (len + xlen > *oldlenp) {
finished = 1;
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
if (len > *oldlenp)
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
xlen = 0;
else
xlen = *oldlenp - len;
}
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
/* Copyout the page */
Add a -t+S flag to ktrace for tracing activity related to sysctl. MIB names will be displayed, with data readen and written as well.
2006-09-24 02:01:04 +04:00
error = dcopyout(l, arg, (char *)oldp + len, xlen);
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
if (error)
goto done;
len += xlen;
base += xlen;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
}
*oldlenp = len;
done:
When getting the program argument or environement string, we previously assumed that all the strings were stored in a row, separated by NUL chars, at the address pointed bu argv[0] (or envp[0]). This was wrong: if the program changed argvs[0], we still read the first string correctly, but the next strings did contain unexpected data. The fix: read the whole argv (or envp) array, then copy the string one by one, using their addresses in argv (or agrp)
2006-09-10 09:46:02 +04:00
if (argv != NULL)
free(argv, M_TEMP);
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
uvmspace_free(vmspace);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
free(arg, M_TEMP);
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
return error;
out_locked:
- Make the proclist_lock a mutex. The write:read ratio is unfavourable, and mutexes are cheaper use than RW locks. - LOCK_ASSERT -> KASSERT in some places. - Hold proclist_lock/kernel_lock longer in a couple of places.
2007-03-09 17:11:22 +03:00
mutex_exit(&proclist_lock);
sysctl_kern_proc_args: don't assume that the process is resident while we are sleeping.
2005-10-08 10:35:56 +04:00
return error;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
static int
sysctl_security_setidcore(SYSCTLFN_ARGS)
{
int newsize, error;
struct sysctlnode node;
node = *rnode;
node.sysctl_data = &newsize;
newsize = *(int *)rnode->sysctl_data;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return error;
First take at security model abstraction. - Add a few scopes to the kernel: system, network, and machdep. - Add a few more actions/sub-actions (requests), and start using them as opposed to the KAUTH_GENERIC_ISSUSER place-holders. - Introduce a basic set of listeners that implement our "traditional" security model, called "bsd44". This is the default (and only) model we have at the moment. - Update all relevant documentation. - Add some code and docs to help folks who want to actually use this stuff: * There's a sample overlay model, sitting on-top of "bsd44", for fast experimenting with tweaking just a subset of an existing model. This is pretty cool because it's *really* straightforward to do stuff you had to use ugly hacks for until now... * And of course, documentation describing how to do the above for quick reference, including code samples. All of these changes were tested for regressions using a Python-based testsuite that will be (I hope) available soon via pkgsrc. Information about the tests, and how to write new ones, can be found on: http://kauth.linbsd.org/kauthwiki NOTE FOR DEVELOPERS: *PLEASE* don't add any code that does any of the following: - Uses a KAUTH_GENERIC_ISSUSER kauth(9) request, - Checks 'securelevel' directly, - Checks a uid/gid directly. (or if you feel you have to, contact me first) This is still work in progress; It's far from being done, but now it'll be a lot easier. Relevant mailing list threads: http://mail-index.netbsd.org/tech-security/2006/01/25/0011.html http://mail-index.netbsd.org/tech-security/2006/03/24/0001.html http://mail-index.netbsd.org/tech-security/2006/04/18/0000.html http://mail-index.netbsd.org/tech-security/2006/05/15/0000.html http://mail-index.netbsd.org/tech-security/2006/08/01/0000.html http://mail-index.netbsd.org/tech-security/2006/08/25/0000.html Many thanks to YAMAMOTO Takashi, Matt Thomas, and Christos Zoulas for help stablizing kauth(9). Full credit for the regression tests, making sure these changes didn't break anything, goes to Matt Fleming and Jaime Fournier. Happy birthday Randi! :)
2006-09-09 00:58:56 +04:00
if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
Don't use KAUTH_RESULT_* where it's not applicable. Prompted by yamt@.
2006-09-13 14:07:42 +04:00
0, NULL, NULL, NULL))
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
return (EPERM);
*(int *)rnode->sysctl_data = newsize;
return 0;
}
static int
sysctl_security_setidcorename(SYSCTLFN_ARGS)
{
int error;
sysctl_security_setidcorename: don't allocate MAXPATHLEN bytes on stack.
2006-06-17 10:54:58 +04:00
char *newsetidcorename;
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
struct sysctlnode node;
sysctl_security_setidcorename: don't allocate MAXPATHLEN bytes on stack.
2006-06-17 10:54:58 +04:00
newsetidcorename = PNBUF_GET();
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
node = *rnode;
sysctl_security_setidcorename: don't allocate MAXPATHLEN bytes on stack.
2006-06-17 10:54:58 +04:00
node.sysctl_data = newsetidcorename;
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
error = sysctl_lookup(SYSCTLFN_CALL(&node));
sysctl_security_setidcorename: don't allocate MAXPATHLEN bytes on stack.
2006-06-17 10:54:58 +04:00
if (error || newp == NULL) {
goto out;
}
First take at security model abstraction. - Add a few scopes to the kernel: system, network, and machdep. - Add a few more actions/sub-actions (requests), and start using them as opposed to the KAUTH_GENERIC_ISSUSER place-holders. - Introduce a basic set of listeners that implement our "traditional" security model, called "bsd44". This is the default (and only) model we have at the moment. - Update all relevant documentation. - Add some code and docs to help folks who want to actually use this stuff: * There's a sample overlay model, sitting on-top of "bsd44", for fast experimenting with tweaking just a subset of an existing model. This is pretty cool because it's *really* straightforward to do stuff you had to use ugly hacks for until now... * And of course, documentation describing how to do the above for quick reference, including code samples. All of these changes were tested for regressions using a Python-based testsuite that will be (I hope) available soon via pkgsrc. Information about the tests, and how to write new ones, can be found on: http://kauth.linbsd.org/kauthwiki NOTE FOR DEVELOPERS: *PLEASE* don't add any code that does any of the following: - Uses a KAUTH_GENERIC_ISSUSER kauth(9) request, - Checks 'securelevel' directly, - Checks a uid/gid directly. (or if you feel you have to, contact me first) This is still work in progress; It's far from being done, but now it'll be a lot easier. Relevant mailing list threads: http://mail-index.netbsd.org/tech-security/2006/01/25/0011.html http://mail-index.netbsd.org/tech-security/2006/03/24/0001.html http://mail-index.netbsd.org/tech-security/2006/04/18/0000.html http://mail-index.netbsd.org/tech-security/2006/05/15/0000.html http://mail-index.netbsd.org/tech-security/2006/08/01/0000.html http://mail-index.netbsd.org/tech-security/2006/08/25/0000.html Many thanks to YAMAMOTO Takashi, Matt Thomas, and Christos Zoulas for help stablizing kauth(9). Full credit for the regression tests, making sure these changes didn't break anything, goes to Matt Fleming and Jaime Fournier. Happy birthday Randi! :)
2006-09-09 00:58:56 +04:00
if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
Don't use KAUTH_RESULT_* where it's not applicable. Prompted by yamt@.
2006-09-13 14:07:42 +04:00
0, NULL, NULL, NULL)) {
sysctl_security_setidcorename: don't allocate MAXPATHLEN bytes on stack.
2006-06-17 10:54:58 +04:00
error = EPERM;
goto out;
}
if (strlen(newsetidcorename) == 0) {
error = EINVAL;
goto out;
}
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
sysctl_security_setidcorename: don't allocate MAXPATHLEN bytes on stack.
2006-06-17 10:54:58 +04:00
out:
PNBUF_PUT(newsetidcorename);
return error;
implement a security.setid_core node as discussed on tech-kern@ and tech-security@.
2006-02-02 20:48:51 +03:00
}
Add a new sysctl 'cp_id' that returns the array of cpu id values. Requested by me, implemented by atatat.
2005-06-16 18:55:58 +04:00
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.cp_id node. Maps cpus to their
Add a new sysctl 'cp_id' that returns the array of cpu id values. Requested by me, implemented by atatat.
2005-06-16 18:55:58 +04:00
* cpuids.
*/
static int
sysctl_kern_cpid(SYSCTLFN_ARGS)
{
struct sysctlnode node = *rnode;
#ifndef MULTIPROCESSOR
u_intN_t -> uintN_t
2005-12-26 21:41:36 +03:00
uint64_t id;
Add a new sysctl 'cp_id' that returns the array of cpu id values. Requested by me, implemented by atatat.
2005-06-16 18:55:58 +04:00
if (namelen == 1) {
if (name[0] != 0)
return (ENOENT);
/*
* you're allowed to ask for the zero'th processor
*/
name++;
namelen--;
}
node.sysctl_data = &id;
node.sysctl_size = sizeof(id);
id = cpu_number();
return (sysctl_lookup(SYSCTLFN_CALL(&node)));
#else /* MULTIPROCESSOR */
u_intN_t -> uintN_t
2005-12-26 21:41:36 +03:00
uint64_t *cp_id = NULL;
Count the number of CPUs at boot and stash in 'ncpu'. Eventually should have each CPU register at attach, so we can figure out the topology for the scheduler.
2007-02-15 23:32:47 +03:00
int error, n = ncpu;
Add a new sysctl 'cp_id' that returns the array of cpu id values. Requested by me, implemented by atatat.
2005-06-16 18:55:58 +04:00
struct cpu_info *ci;
CPU_INFO_ITERATOR cii;
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* Here you may either retrieve a single cpu id or the whole
* set. The size you get back when probing depends on what
Comment in new cp_id implementation was wrong since I abandoned rewriting it in favor of some testing and then never got back to it. It's better now.
2005-06-18 03:53:21 +04:00
* you ask for.
Add a new sysctl 'cp_id' that returns the array of cpu id values. Requested by me, implemented by atatat.
2005-06-16 18:55:58 +04:00
*/
switch (namelen) {
case 0:
u_intN_t -> uintN_t
2005-12-26 21:41:36 +03:00
node.sysctl_size = n * sizeof(uint64_t);
Add a new sysctl 'cp_id' that returns the array of cpu id values. Requested by me, implemented by atatat.
2005-06-16 18:55:58 +04:00
n = -2; /* ALL */
break;
case 1:
if (name[0] < 0 || name[0] >= n)
return (ENOENT); /* ENOSUCHPROCESSOR */
u_intN_t -> uintN_t
2005-12-26 21:41:36 +03:00
node.sysctl_size = sizeof(uint64_t);
Add a new sysctl 'cp_id' that returns the array of cpu id values. Requested by me, implemented by atatat.
2005-06-16 18:55:58 +04:00
n = name[0];
/*
* adjust these so that sysctl_lookup() will be happy
*/
name++;
namelen--;
break;
default:
return (EINVAL);
}
cp_id = malloc(node.sysctl_size, M_TEMP, M_WAITOK|M_CANFAIL);
if (cp_id == NULL)
return (ENOMEM);
node.sysctl_data = cp_id;
memset(cp_id, 0, node.sysctl_size);
for (CPU_INFO_FOREACH(cii, ci)) {
if (n <= 0)
cp_id[0] = ci->ci_cpuid;
/*
* if a specific processor was requested and we just
* did it, we're done here
*/
if (n == 0)
break;
/*
* if doing "all", skip to next cp_id slot for next processor
*/
if (n == -2)
cp_id++;
/*
* if we're doing a specific processor, we're one
* processor closer
*/
if (n > 0)
n--;
}
error = sysctl_lookup(SYSCTLFN_CALL(&node));
free(node.sysctl_data, M_TEMP);
return (error);
#endif /* MULTIPROCESSOR */
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for hw.usermem and hw.usermem64. Values are
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* calculate on the fly taking into account integer overflow and the
* current wired count.
*/
static int
sysctl_hw_usermem(SYSCTLFN_ARGS)
{
u_int ui;
u_quad_t uq;
struct sysctlnode node;
node = *rnode;
switch (rnode->sysctl_num) {
case HW_USERMEM:
if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
ui = UINT_MAX;
else
ui *= PAGE_SIZE;
node.sysctl_data = &ui;
break;
case HW_USERMEM64:
uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
node.sysctl_data = &uq;
break;
default:
return (EINVAL);
}
return (sysctl_lookup(SYSCTLFN_CALL(&node)));
}
/*
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* sysctl helper routine for kern.cnmagic node. Pulls the old value
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* out, encoded, and stuffs the new value in for decoding.
*/
static int
sysctl_hw_cnmagic(SYSCTLFN_ARGS)
{
char magic[CNS_LEN];
int error;
struct sysctlnode node;
if (oldp)
cn_get_magic(magic, CNS_LEN);
node = *rnode;
node.sysctl_data = &magic[0];
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
nuke trailing whitespace
2005-02-27 00:34:55 +03:00
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
return (cn_set_magic(magic));
}
/*
* ********************************************************************
* section 3: public helper routines that are used for more than one
* node
* ********************************************************************
*/
/*
* sysctl helper routine for the kern.root_device node and some ports'
* machdep.root_device nodes.
*/
int
sysctl_root_device(SYSCTLFN_ARGS)
{
struct sysctlnode node;
node = *rnode;
node.sysctl_data = root_device->dv_xname;
node.sysctl_size = strlen(root_device->dv_xname) + 1;
return (sysctl_lookup(SYSCTLFN_CALL(&node)));
}
/*
* sysctl helper routine for kern.consdev, dependent on the current
- fix comment sentence capitalization. - whitespace cleanup. No functional changes.
2007-10-16 20:05:50 +04:00
* state of the console. Also used for machdep.console_device on some
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
* ports.
*/
int
sysctl_consdev(SYSCTLFN_ARGS)
{
dev_t consdev;
struct sysctlnode node;
if (cn_tab != NULL)
consdev = cn_tab->cn_dev;
else
consdev = NODEV;
node = *rnode;
node.sysctl_data = &consdev;
node.sysctl_size = sizeof(consdev);
return (sysctl_lookup(SYSCTLFN_CALL(&node)));
}
/*
* ********************************************************************
* section 4: support for some helpers
* ********************************************************************
*/
/*
* Fill in a kinfo_proc2 structure for the specified process.
*/
static void
fill_kproc2(struct proc *p, struct kinfo_proc2 *ki)
{
struct tty *tp;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
struct lwp *l, *l2;
struct timeval ut, st, rt;
sigset_t ss1, ss2;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
memset(ki, 0, sizeof(*ki));
Use new PTRTOUINT64() macro instead of local PTRTOINT64() macro.
2004-02-19 06:57:56 +03:00
ki->p_paddr = PTRTOUINT64(p);
ki->p_fd = PTRTOUINT64(p->p_fd);
ki->p_cwdi = PTRTOUINT64(p->p_cwdi);
ki->p_stats = PTRTOUINT64(p->p_stats);
ki->p_limit = PTRTOUINT64(p->p_limit);
ki->p_vmspace = PTRTOUINT64(p->p_vmspace);
ki->p_sigacts = PTRTOUINT64(p->p_sigacts);
ki->p_sess = PTRTOUINT64(p->p_session);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_tsess = 0; /* may be changed if controlling tty below */
Remove proc->p_ru and the 'rusage' pool. I think it existed to cache the numbers in kernel memory of a zombie when proc->p_stats was part of the 'u' area - so got freed earlier and wouldn't (easily) be accessible from a separate process. However since both the p_ru and p_stats fields are freed at the same time it is no longer needed. Ride the recent 4.99.19 version change.
2007-05-01 00:11:41 +04:00
ki->p_ru = PTRTOUINT64(&p->p_stats->p_ru);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_eflag = 0;
ki->p_exitsig = p->p_exitsig;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
ki->p_flag = sysctl_map_flags(sysctl_flagmap, p->p_flag);
ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag);
ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag);
ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag);
ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_pid = p->p_pid;
if (p->p_pptr)
ki->p_ppid = p->p_pptr->p_pid;
else
ki->p_ppid = 0;
ki->p_sid = p->p_session->s_sid;
ki->p__pgid = p->p_pgrp->pg_id;
ki->p_tpgid = NO_PGID; /* may be changed if controlling tty below */
integrate kauth.
2006-05-15 01:15:11 +04:00
ki->p_uid = kauth_cred_geteuid(p->p_cred);
ki->p_ruid = kauth_cred_getuid(p->p_cred);
ki->p_gid = kauth_cred_getegid(p->p_cred);
ki->p_rgid = kauth_cred_getgid(p->p_cred);
ki->p_svuid = kauth_cred_getsvuid(p->p_cred);
ki->p_svgid = kauth_cred_getsvgid(p->p_cred);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
integrate kauth.
2006-05-15 01:15:11 +04:00
ki->p_ngroups = kauth_cred_ngroups(p->p_cred);
kauth_cred_getgroups(p->p_cred, ki->p_groups,
Add a flags parameter to kauth_cred_get/setgroups() so that sys_set/setgroups can copy directly to/from userspace. Avoids exposing the implementation of the group list as an array to code outside kern_auth.c. compat code and man page need updating.
2007-06-30 17:32:14 +04:00
min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])),
UIO_SYSSPACE);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_jobc = p->p_pgrp->pg_jobc;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) {
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_tdev = tp->t_dev;
ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
Use new PTRTOUINT64() macro instead of local PTRTOINT64() macro.
2004-02-19 06:57:56 +03:00
ki->p_tsess = PTRTOUINT64(tp->t_session);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
} else {
ki->p_tdev = NODEV;
}
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
mutex_enter(&p->p_smutex);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_uticks = p->p_uticks;
ki->p_sticks = p->p_sticks;
ki->p_iticks = p->p_iticks;
Use new PTRTOUINT64() macro instead of local PTRTOINT64() macro.
2004-02-19 06:57:56 +03:00
ki->p_tracep = PTRTOUINT64(p->p_tracep);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_traceflag = p->p_traceflag;
memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
ki->p_cpticks = 0;
ki->p_pctcpu = p->p_pctcpu;
Merge scheduler changes from the vmlocking branch. All discussed on tech-kern: - Invert priority space so that zero is the lowest priority. Rearrange number and type of priority levels into bands. Add new bands like 'kernel real time'. - Ignore the priority level passed to tsleep. Compute priority for sleep dynamically. - For SCHED_4BSD, make priority adjustment per-LWP, not per-process.
2007-11-06 03:42:39 +03:00
ki->p_estcpu = 0;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
ss1 = p->p_sigpend.sp_set;
LIST_FOREACH(l, &p->p_lwps, l_sibling) {
/* This is hardly correct, but... */
sigplusset(&l->l_sigpend.sp_set, &ss1);
sigplusset(&l->l_sigmask, &ss2);
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
ki->p_cpticks += l->l_cpticks;
ki->p_pctcpu += l->l_pctcpu;
Merge scheduler changes from the vmlocking branch. All discussed on tech-kern: - Invert priority space so that zero is the lowest priority. Rearrange number and type of priority levels into bands. Add new bands like 'kernel real time'. - Ignore the priority level passed to tsleep. Compute priority for sleep dynamically. - For SCHED_4BSD, make priority adjustment per-LWP, not per-process.
2007-11-06 03:42:39 +03:00
ki->p_estcpu += l->l_estcpu;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
}
memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t));
memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
ki->p_realstat = p->p_stat;
ki->p_nice = p->p_nice;
ki->p_xstat = p->p_xstat;
ki->p_acflag = p->p_acflag;
strncpy(ki->p_comm, p->p_comm,
min(sizeof(ki->p_comm), sizeof(p->p_comm)));
strncpy(ki->p_login, p->p_session->s_login,
min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
ki->p_nlwps = p->p_nlwps;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
ki->p_realflag = ki->p_flag;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (p->p_stat == SIDL || P_ZOMBIE(p)) {
ki->p_vm_rssize = 0;
ki->p_vm_tsize = 0;
ki->p_vm_dsize = 0;
ki->p_vm_ssize = 0;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
ki->p_nrlwps = 0;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
l = NULL;
} else {
struct vmspace *vm = p->p_vmspace;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
int tmp;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_vm_rssize = vm_resident_count(vm);
ki->p_vm_tsize = vm->vm_tsize;
ki->p_vm_dsize = vm->vm_dsize;
ki->p_vm_ssize = vm->vm_ssize;
/* Pick a "representative" LWP */
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
l = proc_representative_lwp(p, &tmp, 1);
lwp_lock(l);
ki->p_nrlwps = tmp;
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
ki->p_forw = 0;
ki->p_back = 0;
Use new PTRTOUINT64() macro instead of local PTRTOINT64() macro.
2004-02-19 06:57:56 +03:00
ki->p_addr = PTRTOUINT64(l->l_addr);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_stat = l->l_stat;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_swtime = l->l_swtime;
ki->p_slptime = l->l_slptime;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
if (l->l_stat == LSONPROC)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
else
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_schedflags = 0;
ki->p_holdcnt = l->l_holdcnt;
Merge scheduler changes from the vmlocking branch. All discussed on tech-kern: - Invert priority space so that zero is the lowest priority. Rearrange number and type of priority levels into bands. Add new bands like 'kernel real time'. - Ignore the priority level passed to tsleep. Compute priority for sleep dynamically. - For SCHED_4BSD, make priority adjustment per-LWP, not per-process.
2007-11-06 03:42:39 +03:00
ki->p_priority = lwp_eprio(l);
ki->p_usrpri = l->l_priority;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (l->l_wmesg)
strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
Use new PTRTOUINT64() macro instead of local PTRTOINT64() macro.
2004-02-19 06:57:56 +03:00
ki->p_wchan = PTRTOUINT64(l->l_wchan);
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
lwp_unlock(l);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
if (p->p_session->s_ttyvp)
ki->p_eflag |= EPROC_CTTY;
if (SESS_LEADER(p))
ki->p_eflag |= EPROC_SLEADER;
/* XXX Is this double check necessary? */
if (P_ZOMBIE(p)) {
ki->p_uvalid = 0;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
ki->p_rtime_sec = 0;
ki->p_rtime_usec = 0;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
} else {
ki->p_uvalid = 1;
ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
calcru(p, &ut, &st, NULL, &rt);
ki->p_rtime_sec = rt.tv_sec;
ki->p_rtime_usec = rt.tv_usec;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_uutime_sec = ut.tv_sec;
ki->p_uutime_usec = ut.tv_usec;
ki->p_ustime_sec = st.tv_sec;
ki->p_ustime_usec = st.tv_usec;
ki->p_uru_maxrss = p->p_stats->p_ru.ru_maxrss;
ki->p_uru_ixrss = p->p_stats->p_ru.ru_ixrss;
ki->p_uru_idrss = p->p_stats->p_ru.ru_idrss;
ki->p_uru_isrss = p->p_stats->p_ru.ru_isrss;
ki->p_uru_minflt = p->p_stats->p_ru.ru_minflt;
ki->p_uru_majflt = p->p_stats->p_ru.ru_majflt;
ki->p_uru_nswap = p->p_stats->p_ru.ru_nswap;
ki->p_uru_inblock = p->p_stats->p_ru.ru_inblock;
ki->p_uru_oublock = p->p_stats->p_ru.ru_oublock;
ki->p_uru_msgsnd = p->p_stats->p_ru.ru_msgsnd;
ki->p_uru_msgrcv = p->p_stats->p_ru.ru_msgrcv;
ki->p_uru_nsignals = p->p_stats->p_ru.ru_nsignals;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
ki->p_uru_nvcsw = 0;
ki->p_uru_nivcsw = 0;
LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
ki->p_uru_nvcsw += (l->l_ncsw - l->l_nivcsw);
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
ki->p_uru_nivcsw += l->l_nivcsw;
}
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
timeradd(&p->p_stats->p_cru.ru_utime,
&p->p_stats->p_cru.ru_stime, &ut);
ki->p_uctime_sec = ut.tv_sec;
ki->p_uctime_usec = ut.tv_usec;
}
#ifdef MULTIPROCESSOR
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
if (l != NULL)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
ki->p_cpuid = l->l_cpu->ci_cpuid;
else
#endif
ki->p_cpuid = KI_NOCPU;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
mutex_exit(&p->p_smutex);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
/*
* Fill in a kinfo_lwp structure for the specified lwp.
*/
static void
fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
{
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
struct proc *p = l->l_proc;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
kl->l_forw = 0;
kl->l_back = 0;
Use new PTRTOUINT64() macro instead of local PTRTOINT64() macro.
2004-02-19 06:57:56 +03:00
kl->l_laddr = PTRTOUINT64(l);
kl->l_addr = PTRTOUINT64(l->l_addr);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
kl->l_stat = l->l_stat;
kl->l_lid = l->l_lid;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
kl->l_flag = sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
kl->l_swtime = l->l_swtime;
kl->l_slptime = l->l_slptime;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
if (l->l_stat == LSONPROC)
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
else
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
kl->l_schedflags = 0;
kl->l_holdcnt = l->l_holdcnt;
Merge scheduler changes from the vmlocking branch. All discussed on tech-kern: - Invert priority space so that zero is the lowest priority. Rearrange number and type of priority levels into bands. Add new bands like 'kernel real time'. - Ignore the priority level passed to tsleep. Compute priority for sleep dynamically. - For SCHED_4BSD, make priority adjustment per-LWP, not per-process.
2007-11-06 03:42:39 +03:00
kl->l_priority = lwp_eprio(l);
kl->l_usrpri = l->l_priority;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (l->l_wmesg)
strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
Use new PTRTOUINT64() macro instead of local PTRTOINT64() macro.
2004-02-19 06:57:56 +03:00
kl->l_wchan = PTRTOUINT64(l->l_wchan);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#ifdef MULTIPROCESSOR
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
kl->l_cpuid = l->l_cpu->ci_cpuid;
#else
kl->l_cpuid = KI_NOCPU;
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
#endif
Add the LWP's runtime to kinfo_lwp.
2007-03-12 00:38:38 +03:00
kl->l_rtime_sec = l->l_rtime.tv_sec;
kl->l_rtime_usec = l->l_rtime.tv_usec;
merge yamt-idlelwp branch. asked by core@. some ports still needs work. from doc/BRANCHES: idle lwp, and some changes depending on it. 1. separate context switching and thread scheduling. (cf. gmcgarry_ctxsw) 2. implement idle lwp. 3. clean up related MD/MI interfaces. 4. make scheduler(s) modular.
2007-05-17 18:51:11 +04:00
kl->l_cpticks = l->l_cpticks;
kl->l_pctcpu = l->l_pctcpu;
kl->l_pid = p->p_pid;
if (l->l_name == NULL)
kl->l_name[0] = '\0';
else
strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
/*
* Fill in an eproc structure for the specified process.
*/
void
fill_eproc(struct proc *p, struct eproc *ep)
{
struct tty *tp;
struct lwp *l;
ep->e_paddr = p;
ep->e_sess = p->p_session;
- Don't cast kauth_cred_t to (struct ucred *), just set pc_ucred = NULL. - Fill ucred::cr_ref.
2006-07-17 18:47:02 +04:00
kauth_cred_topcred(p->p_cred, &ep->e_pcred);
kauth_cred_toucred(p->p_cred, &ep->e_ucred);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (p->p_stat == SIDL || P_ZOMBIE(p)) {
ep->e_vm.vm_rssize = 0;
ep->e_vm.vm_tsize = 0;
ep->e_vm.vm_dsize = 0;
ep->e_vm.vm_ssize = 0;
/* ep->e_vm.vm_pmap = XXX; */
} else {
struct vmspace *vm = p->p_vmspace;
ep->e_vm.vm_rssize = vm_resident_count(vm);
ep->e_vm.vm_tsize = vm->vm_tsize;
ep->e_vm.vm_dsize = vm->vm_dsize;
ep->e_vm.vm_ssize = vm->vm_ssize;
/* Pick a "representative" LWP */
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
mutex_enter(&p->p_smutex);
l = proc_representative_lwp(p, NULL, 1);
lwp_lock(l);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
if (l->l_wmesg)
strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
lwp_unlock(l);
mutex_exit(&p->p_smutex);
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
}
if (p->p_pptr)
ep->e_ppid = p->p_pptr->p_pid;
else
ep->e_ppid = 0;
ep->e_pgid = p->p_pgrp->pg_id;
ep->e_sid = ep->e_sess->s_sid;
ep->e_jobc = p->p_pgrp->pg_jobc;
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
if ((p->p_lflag & PL_CONTROLT) &&
Dynamic sysctl. Gone are the old kern_sysctl(), cpu_sysctl(), hw_sysctl(), vfs_sysctl(), etc, routines, along with sysctl_int() et al. Now all nodes are registered with the tree, and nodes can be added (or removed) easily, and I/O to and from the tree is handled generically. Since the nodes are registered with the tree, the mapping from name to number (and back again) can now be discovered, instead of having to be hard coded. Adding new nodes to the tree is likewise much simpler -- the new infrastructure handles almost all the work for simple types, and just about anything else can be done with a small helper function. All existing nodes are where they were before (numerically speaking), so all existing consumers of sysctl information should notice no difference. PS - I'm sorry, but there's a distinct lack of documentation at the moment. I'm working on sysctl(3/8/9) right now, and I promise to watch out for buses.
2003-12-04 22:38:21 +03:00
(tp = ep->e_sess->s_ttyp)) {
ep->e_tdev = tp->t_dev;
ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
ep->e_tsess = tp->t_session;
} else
ep->e_tdev = NODEV;
ep->e_xsize = ep->e_xrssize = 0;
ep->e_xccount = ep->e_xswrss = 0;
ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
if (SESS_LEADER(p))
ep->e_flag |= EPROC_SLEADER;
strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
}
Merge newlock2 to head.
2007-02-10 00:55:00 +03:00
u_int
sysctl_map_flags(const u_int *map, u_int word)
{
u_int rv;
for (rv = 0; *map != 0; map += 2)
if ((word & map[0]) != 0)
rv |= map[1];
return rv;
}