NetBSD/sys/kern/exec_subr.c
joerg 5f391f4ae2 Export the guard size of the main thread via vm.guard_size. Add a
complementary writable sysctl for the initial guard size of threads
created via pthread_create. Let the existing attribut accessors do the
right thing. Raise the default guard size for threads to 64KB.
2017-07-02 16:41:32 +00:00

468 lines
12 KiB
C

/* $NetBSD: exec_subr.c,v 1.82 2017/07/02 16:41:33 joerg Exp $ */
/*
* Copyright (c) 1993, 1994, 1996 Christopher G. Demetriou
* All rights reserved.
*
* 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:
* This product includes software developed by Christopher G. Demetriou.
* 4. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
*/
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: exec_subr.c,v 1.82 2017/07/02 16:41:33 joerg Exp $");
#include "opt_pax.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/kmem.h>
#include <sys/vnode.h>
#include <sys/filedesc.h>
#include <sys/exec.h>
#include <sys/mman.h>
#include <sys/resourcevar.h>
#include <sys/device.h>
#include <sys/pax.h>
#include <uvm/uvm_extern.h>
#define VMCMD_EVCNT_DECL(name) \
static struct evcnt vmcmd_ev_##name = \
EVCNT_INITIALIZER(EVCNT_TYPE_MISC, NULL, "vmcmd", #name); \
EVCNT_ATTACH_STATIC(vmcmd_ev_##name)
#define VMCMD_EVCNT_INCR(name) \
vmcmd_ev_##name.ev_count++
VMCMD_EVCNT_DECL(calls);
VMCMD_EVCNT_DECL(extends);
VMCMD_EVCNT_DECL(kills);
#ifdef DEBUG_STACK
#define DPRINTF(a) uprintf a
#else
#define DPRINTF(a)
#endif
unsigned int user_stack_guard_size = 1024 * 1024;
unsigned int user_thread_stack_guard_size = 64 * 1024;
/*
* new_vmcmd():
* create a new vmcmd structure and fill in its fields based
* on function call arguments. make sure objects ref'd by
* the vmcmd are 'held'.
*/
void
new_vmcmd(struct exec_vmcmd_set *evsp,
int (*proc)(struct lwp * l, struct exec_vmcmd *),
vsize_t len, vaddr_t addr, struct vnode *vp, u_long offset,
u_int prot, int flags)
{
struct exec_vmcmd *vcp;
VMCMD_EVCNT_INCR(calls);
KASSERT(proc != vmcmd_map_pagedvn || (vp->v_iflag & VI_TEXT));
KASSERT(vp == NULL || vp->v_usecount > 0);
if (evsp->evs_used >= evsp->evs_cnt)
vmcmdset_extend(evsp);
vcp = &evsp->evs_cmds[evsp->evs_used++];
vcp->ev_proc = proc;
vcp->ev_len = len;
vcp->ev_addr = addr;
if ((vcp->ev_vp = vp) != NULL)
vref(vp);
vcp->ev_offset = offset;
vcp->ev_prot = prot;
vcp->ev_flags = flags;
}
void
vmcmdset_extend(struct exec_vmcmd_set *evsp)
{
struct exec_vmcmd *nvcp;
u_int ocnt;
#ifdef DIAGNOSTIC
if (evsp->evs_used < evsp->evs_cnt)
panic("vmcmdset_extend: not necessary");
#endif
/* figure out number of entries in new set */
if ((ocnt = evsp->evs_cnt) != 0) {
evsp->evs_cnt += ocnt;
VMCMD_EVCNT_INCR(extends);
} else
evsp->evs_cnt = EXEC_DEFAULT_VMCMD_SETSIZE;
/* allocate it */
nvcp = kmem_alloc(evsp->evs_cnt * sizeof(struct exec_vmcmd), KM_SLEEP);
/* free the old struct, if there was one, and record the new one */
if (ocnt) {
memcpy(nvcp, evsp->evs_cmds,
(ocnt * sizeof(struct exec_vmcmd)));
kmem_free(evsp->evs_cmds, ocnt * sizeof(struct exec_vmcmd));
}
evsp->evs_cmds = nvcp;
}
void
kill_vmcmds(struct exec_vmcmd_set *evsp)
{
struct exec_vmcmd *vcp;
u_int i;
VMCMD_EVCNT_INCR(kills);
if (evsp->evs_cnt == 0)
return;
for (i = 0; i < evsp->evs_used; i++) {
vcp = &evsp->evs_cmds[i];
if (vcp->ev_vp != NULL)
vrele(vcp->ev_vp);
}
kmem_free(evsp->evs_cmds, evsp->evs_cnt * sizeof(struct exec_vmcmd));
evsp->evs_used = evsp->evs_cnt = 0;
}
/*
* vmcmd_map_pagedvn():
* handle vmcmd which specifies that a vnode should be mmap'd.
* appropriate for handling demand-paged text and data segments.
*/
static int
vmcmd_get_prot(struct lwp *l, const struct exec_vmcmd *cmd, vm_prot_t *prot,
vm_prot_t *maxprot)
{
*prot = cmd->ev_prot;
*maxprot = PAX_MPROTECT_MAXPROTECT(l, *prot, 0, UVM_PROT_ALL);
if ((*prot & *maxprot) != *prot)
return EACCES;
return PAX_MPROTECT_VALIDATE(l, *prot);
}
int
vmcmd_map_pagedvn(struct lwp *l, struct exec_vmcmd *cmd)
{
struct uvm_object *uobj;
struct vnode *vp = cmd->ev_vp;
struct proc *p = l->l_proc;
int error;
vm_prot_t prot, maxprot;
KASSERT(vp->v_iflag & VI_TEXT);
/*
* map the vnode in using uvm_map.
*/
if (cmd->ev_len == 0)
return 0;
if (cmd->ev_offset & PAGE_MASK)
return EINVAL;
if (cmd->ev_addr & PAGE_MASK)
return EINVAL;
if (cmd->ev_len & PAGE_MASK)
return EINVAL;
if ((error = vmcmd_get_prot(l, cmd, &prot, &maxprot)) != 0)
return error;
/*
* check the file system's opinion about mmapping the file
*/
error = VOP_MMAP(vp, prot, l->l_cred);
if (error)
return error;
if ((vp->v_vflag & VV_MAPPED) == 0) {
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
vp->v_vflag |= VV_MAPPED;
VOP_UNLOCK(vp);
}
/*
* do the map, reference the object for this map entry
*/
uobj = &vp->v_uobj;
vref(vp);
error = uvm_map(&p->p_vmspace->vm_map, &cmd->ev_addr, cmd->ev_len,
uobj, cmd->ev_offset, 0,
UVM_MAPFLAG(prot, maxprot, UVM_INH_COPY,
UVM_ADV_NORMAL, UVM_FLAG_COPYONW|UVM_FLAG_FIXED));
if (error) {
uobj->pgops->pgo_detach(uobj);
}
return error;
}
/*
* vmcmd_map_readvn():
* handle vmcmd which specifies that a vnode should be read from.
* appropriate for non-demand-paged text/data segments, i.e. impure
* objects (a la OMAGIC and NMAGIC).
*/
int
vmcmd_map_readvn(struct lwp *l, struct exec_vmcmd *cmd)
{
struct proc *p = l->l_proc;
int error;
long diff;
if (cmd->ev_len == 0)
return 0;
diff = cmd->ev_addr - trunc_page(cmd->ev_addr);
cmd->ev_addr -= diff; /* required by uvm_map */
cmd->ev_offset -= diff;
cmd->ev_len += diff;
error = uvm_map(&p->p_vmspace->vm_map, &cmd->ev_addr,
round_page(cmd->ev_len), NULL, UVM_UNKNOWN_OFFSET, 0,
UVM_MAPFLAG(UVM_PROT_ALL, UVM_PROT_ALL, UVM_INH_COPY,
UVM_ADV_NORMAL,
UVM_FLAG_FIXED|UVM_FLAG_OVERLAY|UVM_FLAG_COPYONW));
if (error)
return error;
return vmcmd_readvn(l, cmd);
}
int
vmcmd_readvn(struct lwp *l, struct exec_vmcmd *cmd)
{
struct proc *p = l->l_proc;
int error;
vm_prot_t prot, maxprot;
error = vn_rdwr(UIO_READ, cmd->ev_vp, (void *)cmd->ev_addr,
cmd->ev_len, cmd->ev_offset, UIO_USERSPACE, IO_UNIT,
l->l_cred, NULL, l);
if (error)
return error;
if ((error = vmcmd_get_prot(l, cmd, &prot, &maxprot)) != 0)
return error;
#ifdef PMAP_NEED_PROCWR
/*
* we had to write the process, make sure the pages are synched
* with the instruction cache.
*/
if (prot & VM_PROT_EXECUTE)
pmap_procwr(p, cmd->ev_addr, cmd->ev_len);
#endif
/*
* we had to map in the area at PROT_ALL so that vn_rdwr()
* could write to it. however, the caller seems to want
* it mapped read-only, so now we are going to have to call
* uvm_map_protect() to fix up the protection. ICK.
*/
if (maxprot != VM_PROT_ALL) {
error = uvm_map_protect(&p->p_vmspace->vm_map,
trunc_page(cmd->ev_addr),
round_page(cmd->ev_addr + cmd->ev_len),
maxprot, true);
if (error)
return error;
}
if (prot != maxprot) {
error = uvm_map_protect(&p->p_vmspace->vm_map,
trunc_page(cmd->ev_addr),
round_page(cmd->ev_addr + cmd->ev_len),
prot, false);
if (error)
return error;
}
return 0;
}
/*
* vmcmd_map_zero():
* handle vmcmd which specifies a zero-filled address space region. The
* address range must be first allocated, then protected appropriately.
*/
int
vmcmd_map_zero(struct lwp *l, struct exec_vmcmd *cmd)
{
struct proc *p = l->l_proc;
int error;
long diff;
vm_prot_t prot, maxprot;
diff = cmd->ev_addr - trunc_page(cmd->ev_addr);
cmd->ev_addr -= diff; /* required by uvm_map */
cmd->ev_len += diff;
if ((error = vmcmd_get_prot(l, cmd, &prot, &maxprot)) != 0)
return error;
error = uvm_map(&p->p_vmspace->vm_map, &cmd->ev_addr,
round_page(cmd->ev_len), NULL, UVM_UNKNOWN_OFFSET, 0,
UVM_MAPFLAG(prot, maxprot, UVM_INH_COPY,
UVM_ADV_NORMAL,
UVM_FLAG_FIXED|UVM_FLAG_COPYONW));
if (cmd->ev_flags & VMCMD_STACK)
curproc->p_vmspace->vm_issize += atop(round_page(cmd->ev_len));
return error;
}
/*
* exec_read_from():
*
* Read from vnode into buffer at offset.
*/
int
exec_read_from(struct lwp *l, struct vnode *vp, u_long off, void *bf,
size_t size)
{
int error;
size_t resid;
if ((error = vn_rdwr(UIO_READ, vp, bf, size, off, UIO_SYSSPACE,
0, l->l_cred, &resid, NULL)) != 0)
return error;
/*
* See if we got all of it
*/
if (resid != 0)
return ENOEXEC;
return 0;
}
/*
* exec_setup_stack(): Set up the stack segment for an elf
* executable.
*
* Note that the ep_ssize parameter must be set to be the current stack
* limit; this is adjusted in the body of execve() to yield the
* appropriate stack segment usage once the argument length is
* calculated.
*
* This function returns an int for uniformity with other (future) formats'
* stack setup functions. They might have errors to return.
*/
int
exec_setup_stack(struct lwp *l, struct exec_package *epp)
{
vsize_t max_stack_size;
vaddr_t access_linear_min;
vsize_t access_size;
vaddr_t noaccess_linear_min;
vsize_t noaccess_size;
#ifndef USRSTACK32
#define USRSTACK32 (0x00000000ffffffffL&~PGOFSET)
#endif
#ifndef MAXSSIZ32
#define MAXSSIZ32 (MAXSSIZ >> 2)
#endif
if (epp->ep_flags & EXEC_32) {
epp->ep_minsaddr = USRSTACK32;
max_stack_size = MAXSSIZ32;
} else {
epp->ep_minsaddr = USRSTACK;
max_stack_size = MAXSSIZ;
}
DPRINTF(("ep_minsaddr=%#jx max_stack_size=%#jx\n",
(uintmax_t)epp->ep_minsaddr, (uintmax_t)max_stack_size));
pax_aslr_stack(epp, &max_stack_size);
DPRINTF(("[RLIMIT_STACK].lim_cur=%#jx max_stack_size=%#jx\n",
(uintmax_t)l->l_proc->p_rlimit[RLIMIT_STACK].rlim_cur,
(uintmax_t)max_stack_size));
epp->ep_ssize = MIN(l->l_proc->p_rlimit[RLIMIT_STACK].rlim_cur,
max_stack_size);
l->l_proc->p_stackbase = epp->ep_minsaddr;
epp->ep_maxsaddr = (vaddr_t)STACK_GROW(epp->ep_minsaddr,
max_stack_size);
DPRINTF(("ep_ssize=%#jx ep_minsaddr=%#jx ep_maxsaddr=%#jx\n",
(uintmax_t)epp->ep_ssize, (uintmax_t)epp->ep_minsaddr,
(uintmax_t)epp->ep_maxsaddr));
/*
* set up commands for stack. note that this takes *two*, one to
* map the part of the stack which we can access, and one to map
* the part which we can't.
*
* arguably, it could be made into one, but that would require the
* addition of another mapping proc, which is unnecessary
*/
access_size = epp->ep_ssize;
access_linear_min = (vaddr_t)STACK_ALLOC(epp->ep_minsaddr, access_size);
noaccess_size = max_stack_size - access_size;
noaccess_linear_min = (vaddr_t)STACK_ALLOC(STACK_GROW(epp->ep_minsaddr,
access_size), noaccess_size);
DPRINTF(("access_size=%#jx, access_linear_min=%#jx, "
"noaccess_size=%#jx, noaccess_linear_min=%#jx\n",
(uintmax_t)access_size, (uintmax_t)access_linear_min,
(uintmax_t)noaccess_size, (uintmax_t)noaccess_linear_min));
if (user_stack_guard_size > 0) {
#ifdef __MACHINE_STACK_GROWS_UP
vsize_t guard_size = MIN(VM_MAXUSER_ADDRESS - epp->ep_maxsaddr, user_stack_guard_size);
if (guard_size > 0)
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero, guard_size,
epp->ep_maxsaddr, NULL, 0, VM_PROT_NONE);
#else
NEW_VMCMD(&epp->ep_vmcmds, vmcmd_map_zero, user_stack_guard_size,
epp->ep_maxsaddr - user_stack_guard_size, NULL, 0, VM_PROT_NONE);
#endif
}
if (noaccess_size > 0 && noaccess_size <= MAXSSIZ) {
NEW_VMCMD2(&epp->ep_vmcmds, vmcmd_map_zero, noaccess_size,
noaccess_linear_min, NULL, 0, VM_PROT_NONE, VMCMD_STACK);
}
KASSERT(access_size > 0 && access_size <= MAXSSIZ);
NEW_VMCMD2(&epp->ep_vmcmds, vmcmd_map_zero, access_size,
access_linear_min, NULL, 0, VM_PROT_READ | VM_PROT_WRITE,
VMCMD_STACK);
return 0;
}