NetBSD/usr.bin/vmstat/vmstat.c
rmind 40cf6f3659 Remove uarea swap-out functionality:
- Addresses the issue described in PR/38828.
- Some simplification in threading and sleepq subsystems.
- Eliminates pmap_collect() and, as a side note, allows pmap optimisations.
- Eliminates XS_CTL_DATA_ONSTACK in scsipi code.
- Avoids few scans on LWP list and thus potentially long holds of proc_lock.
- Cuts ~1.5k lines of code.  Reduces amd64 kernel size by ~4k.
- Removes __SWAP_BROKEN cases.

Tested on x86, mips, acorn32 (thanks <mpumford>) and partly tested on
acorn26 (thanks to <bjh21>).

Discussed on <tech-kern>, reviewed by <ad>.
2009-10-21 21:11:57 +00:00

1730 lines
46 KiB
C

/* $NetBSD: vmstat.c,v 1.166 2009/10/21 21:12:07 rmind Exp $ */
/*-
* Copyright (c) 1998, 2000, 2001, 2007 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation by:
* - Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
* NASA Ames Research Center.
* - Simon Burge and Luke Mewburn of Wasabi Systems, Inc.
*
* 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.
*
* 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.
*/
/*
* Copyright (c) 1980, 1986, 1991, 1993
* The Regents of the University of California. 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. Neither the name of the University 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 REGENTS 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 REGENTS 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.
*/
#include <sys/cdefs.h>
#ifndef lint
__COPYRIGHT("@(#) Copyright (c) 1980, 1986, 1991, 1993\
The Regents of the University of California. All rights reserved.");
#endif /* not lint */
#ifndef lint
#if 0
static char sccsid[] = "@(#)vmstat.c 8.2 (Berkeley) 3/1/95";
#else
__RCSID("$NetBSD: vmstat.c,v 1.166 2009/10/21 21:12:07 rmind Exp $");
#endif
#endif /* not lint */
#define __POOL_EXPOSE
#include <sys/param.h>
#include <sys/mount.h>
#include <sys/uio.h>
#include <sys/buf.h>
#include <sys/evcnt.h>
#include <sys/ioctl.h>
#include <sys/malloc.h>
#include <sys/mallocvar.h>
#include <sys/namei.h>
#include <sys/pool.h>
#include <sys/proc.h>
#include <sys/sched.h>
#include <sys/socket.h>
#include <sys/sysctl.h>
#include <sys/time.h>
#include <sys/user.h>
#include <uvm/uvm_extern.h>
#include <uvm/uvm_stat.h>
#include <net/if.h>
#include <netinet/in.h>
#include <netinet/in_var.h>
#include <ufs/ufs/inode.h>
#include <nfs/rpcv2.h>
#include <nfs/nfsproto.h>
#include <nfs/nfsnode.h>
#include <ctype.h>
#include <err.h>
#include <errno.h>
#include <fcntl.h>
#include <kvm.h>
#include <limits.h>
#include <nlist.h>
#undef n_hash
#include <paths.h>
#include <signal.h>
#include <stdio.h>
#include <stddef.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
#include <util.h>
#include "drvstats.h"
/*
* General namelist
*/
struct nlist namelist[] =
{
#define X_BOOTTIME 0
{ .n_name = "_boottime" },
#define X_HZ 1
{ .n_name = "_hz" },
#define X_STATHZ 2
{ .n_name = "_stathz" },
#define X_NCHSTATS 3
{ .n_name = "_nchstats" },
#define X_KMEMSTAT 4
{ .n_name = "_kmemstatistics" },
#define X_KMEMBUCKETS 5
{ .n_name = "_kmembuckets" },
#define X_ALLEVENTS 6
{ .n_name = "_allevents" },
#define X_POOLHEAD 7
{ .n_name = "_pool_head" },
#define X_UVMEXP 8
{ .n_name = "_uvmexp" },
#define X_TIME_SECOND 9
{ .n_name = "_time_second" },
#define X_TIME 10
{ .n_name = "_time" },
#define X_NL_SIZE 11
{ .n_name = NULL },
};
/*
* Namelist for pre-evcnt interrupt counters.
*/
struct nlist intrnl[] =
{
#define X_INTRNAMES 0
{ .n_name = "_intrnames" },
#define X_EINTRNAMES 1
{ .n_name = "_eintrnames" },
#define X_INTRCNT 2
{ .n_name = "_intrcnt" },
#define X_EINTRCNT 3
{ .n_name = "_eintrcnt" },
#define X_INTRNL_SIZE 4
{ .n_name = NULL },
};
/*
* Namelist for hash statistics
*/
struct nlist hashnl[] =
{
#define X_NFSNODE 0
{ .n_name = "_nfsnodehash" },
#define X_NFSNODETBL 1
{ .n_name = "_nfsnodehashtbl" },
#define X_IHASH 2
{ .n_name = "_ihash" },
#define X_IHASHTBL 3
{ .n_name = "_ihashtbl" },
#define X_BUFHASH 4
{ .n_name = "_bufhash" },
#define X_BUFHASHTBL 5
{ .n_name = "_bufhashtbl" },
#define X_UIHASH 6
{ .n_name = "_uihash" },
#define X_UIHASHTBL 7
{ .n_name = "_uihashtbl" },
#define X_IFADDRHASH 8
{ .n_name = "_in_ifaddrhash" },
#define X_IFADDRHASHTBL 9
{ .n_name = "_in_ifaddrhashtbl" },
#define X_NCHASH 10
{ .n_name = "_nchash" },
#define X_NCHASHTBL 11
{ .n_name = "_nchashtbl" },
#define X_NCVHASH 12
{ .n_name = "_ncvhash" },
#define X_NCVHASHTBL 13
{ .n_name = "_ncvhashtbl" },
#define X_HASHNL_SIZE 14 /* must be last */
{ .n_name = NULL },
};
/*
* Namelist for UVM histories
*/
struct nlist histnl[] =
{
{ .n_name = "_uvm_histories" },
#define X_UVM_HISTORIES 0
{ .n_name = NULL },
};
#define KILO 1024
struct uvmexp uvmexp, ouvmexp;
int ndrives;
int winlines = 20;
kvm_t *kd;
#define FORKSTAT 1<<0
#define INTRSTAT 1<<1
#define MEMSTAT 1<<2
#define SUMSTAT 1<<3
#define EVCNTSTAT 1<<4
#define VMSTAT 1<<5
#define HISTLIST 1<<6
#define HISTDUMP 1<<7
#define HASHSTAT 1<<8
#define HASHLIST 1<<9
#define VMTOTAL 1<<10
#define POOLCACHESTAT 1<<11
/*
* Print single word. `ovflow' is number of characters didn't fit
* on the last word. `fmt' is a format string to print this word.
* It must contain asterisk for field width. `width' is a width
* occupied by this word. `fixed' is a number of constant chars in
* `fmt'. `val' is a value to be printed using format string `fmt'.
*/
#define PRWORD(ovflw, fmt, width, fixed, val) do { \
(ovflw) += printf((fmt), \
(width) - (fixed) - (ovflw) > 0 ? \
(width) - (fixed) - (ovflw) : 0, \
(val)) - (width); \
if ((ovflw) < 0) \
(ovflw) = 0; \
} while (/* CONSTCOND */0)
void cpustats(int *);
void deref_kptr(const void *, void *, size_t, const char *);
void drvstats(int *);
void doevcnt(int verbose);
void dohashstat(int, int, const char *);
void dointr(int verbose);
void domem(void);
void dopool(int, int);
void dopoolcache(void);
void dosum(void);
void dovmstat(struct timespec *, int);
void print_total_hdr(void);
void dovmtotal(struct timespec *, int);
void kread(struct nlist *, int, void *, size_t);
int kreadc(struct nlist *, int, void *, size_t);
void needhdr(int);
long getuptime(void);
void printhdr(void);
long pct(long, long);
void usage(void);
void doforkst(void);
void hist_traverse(int, const char *);
void hist_dodump(struct uvm_history *);
int main(int, char **);
char **choosedrives(char **);
/* Namelist and memory file names. */
char *nlistf, *memf;
/* allow old usage [vmstat 1] */
#define BACKWARD_COMPATIBILITY
int
main(int argc, char *argv[])
{
int c, todo, verbose, wide;
struct timespec interval;
int reps;
char errbuf[_POSIX2_LINE_MAX];
gid_t egid = getegid();
const char *histname, *hashname;
size_t i;
histname = hashname = NULL;
(void)setegid(getgid());
memf = nlistf = NULL;
reps = todo = verbose = wide = 0;
interval.tv_sec = 0;
interval.tv_nsec = 0;
while ((c = getopt(argc, argv, "Cc:efh:HilLM:mN:stu:UvWw:")) != -1) {
switch (c) {
case 'c':
reps = atoi(optarg);
break;
case 'C':
todo |= POOLCACHESTAT;
break;
case 'e':
todo |= EVCNTSTAT;
break;
case 'f':
todo |= FORKSTAT;
break;
case 'h':
hashname = optarg;
/* FALLTHROUGH */
case 'H':
todo |= HASHSTAT;
break;
case 'i':
todo |= INTRSTAT;
break;
case 'l':
todo |= HISTLIST;
break;
case 'L':
todo |= HASHLIST;
break;
case 'M':
memf = optarg;
break;
case 'm':
todo |= MEMSTAT;
break;
case 'N':
nlistf = optarg;
break;
case 's':
todo |= SUMSTAT;
break;
case 't':
todo |= VMTOTAL;
break;
case 'u':
histname = optarg;
/* FALLTHROUGH */
case 'U':
todo |= HISTDUMP;
break;
case 'v':
verbose++;
break;
case 'W':
wide++;
break;
case 'w':
interval.tv_sec = atol(optarg);
break;
case '?':
default:
usage();
}
}
argc -= optind;
argv += optind;
if (todo == 0)
todo = VMSTAT;
/*
* Discard setgid privileges. If not the running kernel, we toss
* them away totally so that bad guys can't print interesting stuff
* from kernel memory, otherwise switch back to kmem for the
* duration of the kvm_openfiles() call.
*/
if (nlistf != NULL || memf != NULL)
(void)setgid(getgid());
else
(void)setegid(egid);
kd = kvm_openfiles(nlistf, memf, NULL, O_RDONLY, errbuf);
if (kd == NULL)
errx(1, "kvm_openfiles: %s", errbuf);
if (nlistf == NULL && memf == NULL)
(void)setgid(getgid());
if ((c = kvm_nlist(kd, namelist)) != 0) {
int doexit = 0;
if (c == -1)
errx(1, "kvm_nlist: %s %s", "namelist", kvm_geterr(kd));
for (i = 0; i < sizeof(namelist) / sizeof(namelist[0])-1; i++)
if (namelist[i].n_type == 0 &&
i != X_TIME_SECOND &&
i != X_TIME) {
if (doexit++ == 0)
(void)fprintf(stderr, "vmstat: undefined symbols:");
(void)fprintf(stderr, " %s",
namelist[i].n_name);
}
if (doexit) {
(void)fputc('\n', stderr);
exit(1);
}
}
if (todo & INTRSTAT)
(void) kvm_nlist(kd, intrnl);
if ((c = kvm_nlist(kd, hashnl)) == -1 || c == X_HASHNL_SIZE)
errx(1, "kvm_nlist: %s %s", "hashnl", kvm_geterr(kd));
if (kvm_nlist(kd, histnl) == -1)
errx(1, "kvm_nlist: %s %s", "histnl", kvm_geterr(kd));
if (todo & VMSTAT) {
struct winsize winsize;
(void)drvinit(0);/* Initialize disk stats, no disks selected. */
(void)setgid(getgid()); /* don't need privs anymore */
argv = choosedrives(argv); /* Select disks. */
winsize.ws_row = 0;
(void)ioctl(STDOUT_FILENO, TIOCGWINSZ, &winsize);
if (winsize.ws_row > 0)
winlines = winsize.ws_row;
}
#ifdef BACKWARD_COMPATIBILITY
if (*argv) {
interval.tv_sec = atol(*argv);
if (*++argv)
reps = atoi(*argv);
}
#endif
if (interval.tv_sec) {
if (!reps)
reps = -1;
} else if (reps)
interval.tv_sec = 1;
/*
* Statistics dumping is incompatible with the default
* VMSTAT/dovmstat() output. So perform the interval/reps handling
* for it here.
*/
if ((todo & (VMSTAT|VMTOTAL)) == 0) {
for (;;) {
if (todo & (HISTLIST|HISTDUMP)) {
if ((todo & (HISTLIST|HISTDUMP)) ==
(HISTLIST|HISTDUMP))
errx(1, "you may list or dump,"
" but not both!");
hist_traverse(todo, histname);
(void)putchar('\n');
}
if (todo & FORKSTAT) {
doforkst();
(void)putchar('\n');
}
if (todo & MEMSTAT) {
domem();
dopool(verbose, wide);
(void)putchar('\n');
}
if (todo & POOLCACHESTAT) {
dopoolcache();
(void)putchar('\n');
}
if (todo & SUMSTAT) {
dosum();
(void)putchar('\n');
}
if (todo & INTRSTAT) {
dointr(verbose);
(void)putchar('\n');
}
if (todo & EVCNTSTAT) {
doevcnt(verbose);
(void)putchar('\n');
}
if (todo & (HASHLIST|HASHSTAT)) {
if ((todo & (HASHLIST|HASHSTAT)) ==
(HASHLIST|HASHSTAT))
errx(1, "you may list or display,"
" but not both!");
dohashstat(verbose, todo, hashname);
(void)putchar('\n');
}
fflush(stdout);
if (reps >= 0 && --reps <=0)
break;
(void)nanosleep(&interval, NULL);
}
} else {
if ((todo & (VMSTAT|VMTOTAL)) == (VMSTAT|VMTOTAL)) {
errx(1, "you may not both do vmstat and vmtotal");
}
if (todo & VMSTAT)
dovmstat(&interval, reps);
if (todo & VMTOTAL)
dovmtotal(&interval, reps);
}
return 0;
}
char **
choosedrives(char **argv)
{
size_t i;
/*
* Choose drives to be displayed. Priority goes to (in order) drives
* supplied as arguments, default drives. If everything isn't filled
* in and there are drives not taken care of, display the first few
* that fit.
*/
#define BACKWARD_COMPATIBILITY
for (ndrives = 0; *argv; ++argv) {
#ifdef BACKWARD_COMPATIBILITY
if (isdigit((unsigned char)**argv))
break;
#endif
for (i = 0; i < ndrive; i++) {
if (strcmp(dr_name[i], *argv))
continue;
drv_select[i] = 1;
++ndrives;
break;
}
}
for (i = 0; i < ndrive && ndrives < 2; i++) {
if (drv_select[i])
continue;
drv_select[i] = 1;
++ndrives;
}
return (argv);
}
long
getuptime(void)
{
static struct timeval boottime;
struct timeval now;
time_t uptime, nowsec;
if (boottime.tv_sec == 0)
kread(namelist, X_BOOTTIME, &boottime, sizeof(boottime));
if (kreadc(namelist, X_TIME_SECOND, &nowsec, sizeof(nowsec))) {
/*
* XXX this assignment dance can be removed once timeval tv_sec
* is SUS mandated time_t
*/
now.tv_sec = nowsec;
now.tv_usec = 0;
} else {
kread(namelist, X_TIME, &now, sizeof(now));
}
uptime = now.tv_sec - boottime.tv_sec;
if (uptime <= 0 || uptime > 60*60*24*365*10)
errx(1, "time makes no sense; namelist must be wrong.");
return (uptime);
}
int hz, hdrcnt;
void
print_total_hdr()
{
(void)printf("procs memory\n");
(void)printf("ru dw pw sl");
(void)printf(" total-v active-v active-r");
(void)printf(" vm-sh avm-sh rm-sh arm-sh free\n");
hdrcnt = winlines - 2;
}
void
dovmtotal(struct timespec *interval, int reps)
{
struct vmtotal total;
int mib[2];
size_t size;
(void)signal(SIGCONT, needhdr);
for (hdrcnt = 1;;) {
if (!--hdrcnt)
print_total_hdr();
if (memf != NULL) {
(void)printf(
"Unable to get vmtotals from crash dump.\n");
(void)memset(&total, 0, sizeof(total));
} else {
size = sizeof(total);
mib[0] = CTL_VM;
mib[1] = VM_METER;
if (sysctl(mib, 2, &total, &size, NULL, 0) < 0) {
(void)printf("Can't get vmtotals: %s\n",
strerror(errno));
(void)memset(&total, 0, sizeof(total));
}
}
(void)printf("%2d ", total.t_rq);
(void)printf("%2d ", total.t_dw);
(void)printf("%2d ", total.t_pw);
(void)printf("%2d ", total.t_sl);
(void)printf("%9d ", total.t_vm);
(void)printf("%9d ", total.t_avm);
(void)printf("%9d ", total.t_arm);
(void)printf("%5d ", total.t_vmshr);
(void)printf("%6d ", total.t_avmshr);
(void)printf("%5d ", total.t_rmshr);
(void)printf("%6d ", total.t_armshr);
(void)printf("%5d", total.t_free);
(void)putchar('\n');
(void)fflush(stdout);
if (reps >= 0 && --reps <= 0)
break;
(void)nanosleep(interval, NULL);
}
}
void
dovmstat(struct timespec *interval, int reps)
{
struct vmtotal total;
time_t uptime, halfuptime;
int mib[2];
size_t size;
int pagesize = getpagesize();
int ovflw;
uptime = getuptime();
halfuptime = uptime / 2;
(void)signal(SIGCONT, needhdr);
if (namelist[X_STATHZ].n_type != 0 && namelist[X_STATHZ].n_value != 0)
kread(namelist, X_STATHZ, &hz, sizeof(hz));
if (!hz)
kread(namelist, X_HZ, &hz, sizeof(hz));
for (hdrcnt = 1;;) {
if (!--hdrcnt)
printhdr();
/* Read new disk statistics */
cpureadstats();
drvreadstats();
tkreadstats();
kread(namelist, X_UVMEXP, &uvmexp, sizeof(uvmexp));
if (memf != NULL) {
/*
* XXX Can't do this if we're reading a crash
* XXX dump because they're lazily-calculated.
*/
(void)printf(
"Unable to get vmtotals from crash dump.\n");
(void)memset(&total, 0, sizeof(total));
} else {
size = sizeof(total);
mib[0] = CTL_VM;
mib[1] = VM_METER;
if (sysctl(mib, 2, &total, &size, NULL, 0) < 0) {
(void)printf("Can't get vmtotals: %s\n",
strerror(errno));
(void)memset(&total, 0, sizeof(total));
}
}
ovflw = 0;
PRWORD(ovflw, " %*d", 2, 1, total.t_rq - 1);
PRWORD(ovflw, " %*d", 2, 1, total.t_dw + total.t_pw);
#define pgtok(a) (long)((a) * ((uint32_t)pagesize >> 10))
#define rate(x) (u_long)(((x) + halfuptime) / uptime) /* round */
PRWORD(ovflw, " %*ld", 9, 1, pgtok(total.t_avm));
PRWORD(ovflw, " %*ld", 7, 1, pgtok(total.t_free));
PRWORD(ovflw, " %*ld", 5, 1,
rate(uvmexp.faults - ouvmexp.faults));
PRWORD(ovflw, " %*ld", 4, 1,
rate(uvmexp.pdreact - ouvmexp.pdreact));
PRWORD(ovflw, " %*ld", 4, 1,
rate(uvmexp.pageins - ouvmexp.pageins));
PRWORD(ovflw, " %*ld", 5, 1,
rate(uvmexp.pgswapout - ouvmexp.pgswapout));
PRWORD(ovflw, " %*ld", 5, 1,
rate(uvmexp.pdfreed - ouvmexp.pdfreed));
PRWORD(ovflw, " %*ld", 6, 2,
rate(uvmexp.pdscans - ouvmexp.pdscans));
drvstats(&ovflw);
PRWORD(ovflw, " %*ld", 5, 1,
rate(uvmexp.intrs - ouvmexp.intrs));
PRWORD(ovflw, " %*ld", 5, 1,
rate(uvmexp.syscalls - ouvmexp.syscalls));
PRWORD(ovflw, " %*ld", 4, 1,
rate(uvmexp.swtch - ouvmexp.swtch));
cpustats(&ovflw);
(void)putchar('\n');
(void)fflush(stdout);
if (reps >= 0 && --reps <= 0)
break;
ouvmexp = uvmexp;
uptime = interval->tv_sec;
/*
* We round upward to avoid losing low-frequency events
* (i.e., >= 1 per interval but < 1 per second).
*/
halfuptime = uptime == 1 ? 0 : (uptime + 1) / 2;
(void)nanosleep(interval, NULL);
}
}
void
printhdr(void)
{
size_t i;
(void)printf(" procs memory page%*s", 23, "");
if (ndrives > 0)
(void)printf("%s %*sfaults cpu\n",
((ndrives > 1) ? "disks" : "disk"),
((ndrives > 1) ? ndrives * 3 - 4 : 0), "");
else
(void)printf("%*s faults cpu\n",
ndrives * 3, "");
(void)printf(" r b avm fre flt re pi po fr sr ");
for (i = 0; i < ndrive; i++)
if (drv_select[i])
(void)printf("%c%c ", dr_name[i][0],
dr_name[i][strlen(dr_name[i]) - 1]);
(void)printf(" in sy cs us sy id\n");
hdrcnt = winlines - 2;
}
/*
* Force a header to be prepended to the next output.
*/
void
/*ARGSUSED*/
needhdr(int dummy)
{
hdrcnt = 1;
}
long
pct(long top, long bot)
{
long ans;
if (bot == 0)
return (0);
ans = (long)((quad_t)top * 100 / bot);
return (ans);
}
#define PCT(top, bot) (int)pct((long)(top), (long)(bot))
void
dosum(void)
{
struct nchstats nchstats;
u_long nchtotal;
int mib[2];
struct uvmexp_sysctl uvmexp2;
size_t ssize;
int active_kernel;
/*
* The "active" and "inactive" variables
* are now estimated by the kernel and sadly
* can not easily be dug out of a crash dump.
*/
ssize = sizeof(uvmexp2);
memset(&uvmexp2, 0, ssize);
active_kernel = (memf == NULL);
if (active_kernel) {
/* only on active kernel */
mib[0] = CTL_VM;
mib[1] = VM_UVMEXP2;
if (sysctl(mib, 2, &uvmexp2, &ssize, NULL, 0) < 0)
fprintf(stderr, "%s: sysctl vm.uvmexp2 failed: %s",
getprogname(), strerror(errno));
}
kread(namelist, X_UVMEXP, &uvmexp, sizeof(uvmexp));
(void)printf("%9u bytes per page\n", uvmexp.pagesize);
(void)printf("%9u page color%s\n",
uvmexp.ncolors, uvmexp.ncolors == 1 ? "" : "s");
(void)printf("%9u pages managed\n", uvmexp.npages);
(void)printf("%9u pages free\n", uvmexp.free);
if (active_kernel) {
(void)printf("%9" PRIu64 " pages active\n", uvmexp2.active);
(void)printf("%9" PRIu64 " pages inactive\n", uvmexp2.inactive);
}
(void)printf("%9u pages paging\n", uvmexp.paging);
(void)printf("%9u pages wired\n", uvmexp.wired);
(void)printf("%9u zero pages\n", uvmexp.zeropages);
(void)printf("%9u reserve pagedaemon pages\n",
uvmexp.reserve_pagedaemon);
(void)printf("%9u reserve kernel pages\n", uvmexp.reserve_kernel);
(void)printf("%9u anonymous pages\n", uvmexp.anonpages);
(void)printf("%9u cached file pages\n", uvmexp.filepages);
(void)printf("%9u cached executable pages\n", uvmexp.execpages);
(void)printf("%9u minimum free pages\n", uvmexp.freemin);
(void)printf("%9u target free pages\n", uvmexp.freetarg);
(void)printf("%9u maximum wired pages\n", uvmexp.wiredmax);
(void)printf("%9u swap devices\n", uvmexp.nswapdev);
(void)printf("%9u swap pages\n", uvmexp.swpages);
(void)printf("%9u swap pages in use\n", uvmexp.swpginuse);
(void)printf("%9u swap allocations\n", uvmexp.nswget);
(void)printf("%9u total faults taken\n", uvmexp.faults);
(void)printf("%9u traps\n", uvmexp.traps);
(void)printf("%9u device interrupts\n", uvmexp.intrs);
(void)printf("%9u CPU context switches\n", uvmexp.swtch);
(void)printf("%9u software interrupts\n", uvmexp.softs);
(void)printf("%9u system calls\n", uvmexp.syscalls);
(void)printf("%9u pagein requests\n", uvmexp.pageins);
(void)printf("%9u pageout requests\n", uvmexp.pdpageouts);
(void)printf("%9u pages swapped in\n", uvmexp.pgswapin);
(void)printf("%9u pages swapped out\n", uvmexp.pgswapout);
(void)printf("%9u forks total\n", uvmexp.forks);
(void)printf("%9u forks blocked parent\n", uvmexp.forks_ppwait);
(void)printf("%9u forks shared address space with parent\n",
uvmexp.forks_sharevm);
(void)printf("%9u pagealloc zero wanted and avail\n",
uvmexp.pga_zerohit);
(void)printf("%9u pagealloc zero wanted and not avail\n",
uvmexp.pga_zeromiss);
(void)printf("%9u aborts of idle page zeroing\n",
uvmexp.zeroaborts);
(void)printf("%9u pagealloc desired color avail\n",
uvmexp.colorhit);
(void)printf("%9u pagealloc desired color not avail\n",
uvmexp.colormiss);
(void)printf("%9u pagealloc local cpu avail\n",
uvmexp.cpuhit);
(void)printf("%9u pagealloc local cpu not avail\n",
uvmexp.cpumiss);
(void)printf("%9u faults with no memory\n", uvmexp.fltnoram);
(void)printf("%9u faults with no anons\n", uvmexp.fltnoanon);
(void)printf("%9u faults had to wait on pages\n", uvmexp.fltpgwait);
(void)printf("%9u faults found released page\n", uvmexp.fltpgrele);
(void)printf("%9u faults relock (%u ok)\n", uvmexp.fltrelck,
uvmexp.fltrelckok);
(void)printf("%9u anon page faults\n", uvmexp.fltanget);
(void)printf("%9u anon retry faults\n", uvmexp.fltanretry);
(void)printf("%9u amap copy faults\n", uvmexp.fltamcopy);
(void)printf("%9u neighbour anon page faults\n", uvmexp.fltnamap);
(void)printf("%9u neighbour object page faults\n", uvmexp.fltnomap);
(void)printf("%9u locked pager get faults\n", uvmexp.fltlget);
(void)printf("%9u unlocked pager get faults\n", uvmexp.fltget);
(void)printf("%9u anon faults\n", uvmexp.flt_anon);
(void)printf("%9u anon copy on write faults\n", uvmexp.flt_acow);
(void)printf("%9u object faults\n", uvmexp.flt_obj);
(void)printf("%9u promote copy faults\n", uvmexp.flt_prcopy);
(void)printf("%9u promote zero fill faults\n", uvmexp.flt_przero);
(void)printf("%9u times daemon wokeup\n",uvmexp.pdwoke);
(void)printf("%9u revolutions of the clock hand\n", uvmexp.pdrevs);
(void)printf("%9u pages freed by daemon\n", uvmexp.pdfreed);
(void)printf("%9u pages scanned by daemon\n", uvmexp.pdscans);
(void)printf("%9u anonymous pages scanned by daemon\n",
uvmexp.pdanscan);
(void)printf("%9u object pages scanned by daemon\n", uvmexp.pdobscan);
(void)printf("%9u pages reactivated\n", uvmexp.pdreact);
(void)printf("%9u pages found busy by daemon\n", uvmexp.pdbusy);
(void)printf("%9u total pending pageouts\n", uvmexp.pdpending);
(void)printf("%9u pages deactivated\n", uvmexp.pddeact);
kread(namelist, X_NCHSTATS, &nchstats, sizeof(nchstats));
nchtotal = nchstats.ncs_goodhits + nchstats.ncs_neghits +
nchstats.ncs_badhits + nchstats.ncs_falsehits +
nchstats.ncs_miss + nchstats.ncs_long;
(void)printf("%9lu total name lookups\n", nchtotal);
(void)printf("%9lu good hits\n", nchstats.ncs_goodhits);
(void)printf("%9lu negative hits\n", nchstats.ncs_neghits);
(void)printf("%9lu bad hits\n", nchstats.ncs_badhits);
(void)printf("%9lu false hits\n", nchstats.ncs_falsehits);
(void)printf("%9lu miss\n", nchstats.ncs_miss);
(void)printf("%9lu too long\n", nchstats.ncs_long);
(void)printf("%9lu pass2 hits\n", nchstats.ncs_pass2);
(void)printf("%9lu 2passes\n", nchstats.ncs_2passes);
(void)printf(
"%9s cache hits (%d%% pos + %d%% neg) system %d%% per-process\n",
"", PCT(nchstats.ncs_goodhits, nchtotal),
PCT(nchstats.ncs_neghits, nchtotal),
PCT(nchstats.ncs_pass2, nchtotal));
(void)printf("%9s deletions %d%%, falsehits %d%%, toolong %d%%\n", "",
PCT(nchstats.ncs_badhits, nchtotal),
PCT(nchstats.ncs_falsehits, nchtotal),
PCT(nchstats.ncs_long, nchtotal));
}
void
doforkst(void)
{
kread(namelist, X_UVMEXP, &uvmexp, sizeof(uvmexp));
(void)printf("%u forks total\n", uvmexp.forks);
(void)printf("%u forks blocked parent\n", uvmexp.forks_ppwait);
(void)printf("%u forks shared address space with parent\n",
uvmexp.forks_sharevm);
}
void
drvstats(int *ovflwp)
{
size_t dn;
double etime;
int ovflw = *ovflwp;
/* Calculate disk stat deltas. */
cpuswap();
drvswap();
tkswap();
etime = cur.cp_etime;
for (dn = 0; dn < ndrive; ++dn) {
if (!drv_select[dn])
continue;
PRWORD(ovflw, " %*.0f", 3, 1,
(cur.rxfer[dn] + cur.wxfer[dn]) / etime);
}
*ovflwp = ovflw;
}
void
cpustats(int *ovflwp)
{
int state;
double pcnt, total;
double stat_us, stat_sy, stat_id;
int ovflw = *ovflwp;
total = 0;
for (state = 0; state < CPUSTATES; ++state)
total += cur.cp_time[state];
if (total)
pcnt = 100 / total;
else
pcnt = 0;
stat_us = (cur.cp_time[CP_USER] + cur.cp_time[CP_NICE]) * pcnt;
stat_sy = (cur.cp_time[CP_SYS] + cur.cp_time[CP_INTR]) * pcnt;
stat_id = cur.cp_time[CP_IDLE] * pcnt;
PRWORD(ovflw, " %*.0f", ((stat_sy >= 100) ? 2 : 3), 1, stat_us);
PRWORD(ovflw, " %*.0f", ((stat_us >= 100 || stat_id >= 100) ? 2 : 3), 1,
stat_sy);
PRWORD(ovflw, " %*.0f", 3, 1, stat_id);
*ovflwp = ovflw;
}
void
dointr(int verbose)
{
unsigned long *intrcnt, *ointrcnt;
unsigned long long inttotal, uptime;
int nintr, inamlen;
char *intrname, *ointrname;
struct evcntlist allevents;
struct evcnt evcnt, *evptr;
char evgroup[EVCNT_STRING_MAX], evname[EVCNT_STRING_MAX];
inttotal = 0;
uptime = getuptime();
(void)printf("%-34s %16s %8s\n", "interrupt", "total", "rate");
nintr = intrnl[X_EINTRCNT].n_value - intrnl[X_INTRCNT].n_value;
inamlen = intrnl[X_EINTRNAMES].n_value - intrnl[X_INTRNAMES].n_value;
if (nintr != 0 && inamlen != 0) {
ointrcnt = intrcnt = malloc((size_t)nintr);
ointrname = intrname = malloc((size_t)inamlen);
if (intrcnt == NULL || intrname == NULL)
errx(1, "%s", "");
kread(intrnl, X_INTRCNT, intrcnt, (size_t)nintr);
kread(intrnl, X_INTRNAMES, intrname, (size_t)inamlen);
nintr /= sizeof(long);
while (--nintr >= 0) {
if (*intrcnt || verbose)
(void)printf("%-34s %16llu %8llu\n", intrname,
(unsigned long long)*intrcnt,
(unsigned long long)
(*intrcnt / uptime));
intrname += strlen(intrname) + 1;
inttotal += *intrcnt++;
}
free(ointrcnt);
free(ointrname);
}
kread(namelist, X_ALLEVENTS, &allevents, sizeof allevents);
evptr = TAILQ_FIRST(&allevents);
while (evptr) {
deref_kptr(evptr, &evcnt, sizeof(evcnt), "event chain trashed");
evptr = TAILQ_NEXT(&evcnt, ev_list);
if (evcnt.ev_type != EVCNT_TYPE_INTR)
continue;
if (evcnt.ev_count == 0 && !verbose)
continue;
deref_kptr(evcnt.ev_group, evgroup,
(size_t)evcnt.ev_grouplen + 1, "event chain trashed");
deref_kptr(evcnt.ev_name, evname,
(size_t)evcnt.ev_namelen + 1, "event chain trashed");
(void)printf("%s %s%*s %16llu %8llu\n", evgroup, evname,
34 - (evcnt.ev_grouplen + 1 + evcnt.ev_namelen), "",
(unsigned long long)evcnt.ev_count,
(unsigned long long)(evcnt.ev_count / uptime));
inttotal += evcnt.ev_count++;
}
(void)printf("%-34s %16llu %8llu\n", "Total", inttotal,
(unsigned long long)(inttotal / uptime));
}
void
doevcnt(int verbose)
{
static const char * evtypes [] = { "misc", "intr", "trap" };
unsigned long long uptime;
struct evcntlist allevents;
struct evcnt evcnt, *evptr;
char evgroup[EVCNT_STRING_MAX], evname[EVCNT_STRING_MAX];
/* XXX should print type! */
uptime = getuptime();
(void)printf("%-34s %16s %8s %s\n", "event", "total", "rate", "type");
kread(namelist, X_ALLEVENTS, &allevents, sizeof allevents);
evptr = TAILQ_FIRST(&allevents);
while (evptr) {
deref_kptr(evptr, &evcnt, sizeof(evcnt), "event chain trashed");
evptr = TAILQ_NEXT(&evcnt, ev_list);
if (evcnt.ev_count == 0 && !verbose)
continue;
deref_kptr(evcnt.ev_group, evgroup,
(size_t)evcnt.ev_grouplen + 1, "event chain trashed");
deref_kptr(evcnt.ev_name, evname,
(size_t)evcnt.ev_namelen + 1, "event chain trashed");
(void)printf("%s %s%*s %16llu %8llu %s\n", evgroup, evname,
34 - (evcnt.ev_grouplen + 1 + evcnt.ev_namelen), "",
(unsigned long long)evcnt.ev_count,
(unsigned long long)(evcnt.ev_count / uptime),
(evcnt.ev_type < sizeof(evtypes)/sizeof(evtypes[0]) ?
evtypes[evcnt.ev_type] : "?"));
}
}
static char memname[64];
void
domem(void)
{
struct kmembuckets *kp;
struct malloc_type ks, *ksp;
int i, j;
int len, size, first;
long totuse = 0, totfree = 0, totreq = 0;
struct kmembuckets buckets[MINBUCKET + 16];
kread(namelist, X_KMEMBUCKETS, buckets, sizeof(buckets));
for (first = 1, i = MINBUCKET, kp = &buckets[i]; i < MINBUCKET + 16;
i++, kp++) {
if (kp->kb_calls == 0)
continue;
if (first) {
(void)printf("Memory statistics by bucket size\n");
(void)printf(
" Size In Use Free Requests HighWater Couldfree\n");
first = 0;
}
size = 1 << i;
(void)printf("%8d %8ld %6ld %10ld %7ld %10ld\n", size,
kp->kb_total - kp->kb_totalfree,
kp->kb_totalfree, kp->kb_calls,
kp->kb_highwat, kp->kb_couldfree);
totfree += size * kp->kb_totalfree;
}
/*
* If kmem statistics are not being gathered by the kernel,
* first will still be 1.
*/
if (first) {
warnx("Kmem statistics are not being gathered by the kernel.");
return;
}
(void)printf("\nMemory usage type by bucket size\n");
(void)printf(" Size Type(s)\n");
kp = &buckets[MINBUCKET];
for (j = 1 << MINBUCKET; j < 1 << (MINBUCKET + 16); j <<= 1, kp++) {
if (kp->kb_calls == 0)
continue;
first = 1;
len = 8;
for (kread(namelist, X_KMEMSTAT, &ksp, sizeof(ksp));
ksp != NULL; ksp = ks.ks_next) {
deref_kptr(ksp, &ks, sizeof(ks), "malloc type");
if (ks.ks_calls == 0)
continue;
if ((ks.ks_size & j) == 0)
continue;
deref_kptr(ks.ks_shortdesc, memname,
sizeof(memname), "malloc type name");
len += 2 + strlen(memname);
if (first)
(void)printf("%8d %s", j, memname);
else
(void)printf(",");
if (len >= 80) {
(void)printf("\n\t ");
len = 10 + strlen(memname);
}
if (!first)
(void)printf(" %s", memname);
first = 0;
}
(void)putchar('\n');
}
(void)printf(
"\nMemory statistics by type Type Kern\n");
(void)printf(
" Type InUse MemUse HighUse Limit Requests Limit Limit Size(s)\n");
for (kread(namelist, X_KMEMSTAT, &ksp, sizeof(ksp));
ksp != NULL; ksp = ks.ks_next) {
deref_kptr(ksp, &ks, sizeof(ks), "malloc type");
if (ks.ks_calls == 0)
continue;
deref_kptr(ks.ks_shortdesc, memname,
sizeof(memname), "malloc type name");
(void)printf("%15s %5ld %6ldK %6ldK %6ldK %10ld %5u %5u",
memname,
ks.ks_inuse, howmany(ks.ks_memuse, KILO),
howmany(ks.ks_maxused, KILO),
howmany(ks.ks_limit, KILO), ks.ks_calls,
ks.ks_limblocks, ks.ks_mapblocks);
first = 1;
for (j = 1 << MINBUCKET; j < 1 << (MINBUCKET + 16); j <<= 1) {
if ((ks.ks_size & j) == 0)
continue;
if (first)
(void)printf(" %d", j);
else
(void)printf(",%d", j);
first = 0;
}
(void)printf("\n");
totuse += ks.ks_memuse;
totreq += ks.ks_calls;
}
(void)printf("\nMemory totals: In Use Free Requests\n");
(void)printf(" %7ldK %6ldK %8ld\n\n",
howmany(totuse, KILO), howmany(totfree, KILO), totreq);
}
void
dopool(int verbose, int wide)
{
int first, ovflw;
void *addr;
long total, inuse, this_total, this_inuse;
TAILQ_HEAD(,pool) pool_head;
struct pool pool, *pp = &pool;
struct pool_allocator pa;
char name[32], maxp[32];
kread(namelist, X_POOLHEAD, &pool_head, sizeof(pool_head));
addr = TAILQ_FIRST(&pool_head);
total = inuse = 0;
for (first = 1; addr != NULL; addr = TAILQ_NEXT(pp, pr_poollist) ) {
deref_kptr(addr, pp, sizeof(*pp), "pool chain trashed");
deref_kptr(pp->pr_alloc, &pa, sizeof(pa),
"pool allocator trashed");
deref_kptr(pp->pr_wchan, name, sizeof(name),
"pool wait channel trashed");
name[sizeof(name)-1] = '\0';
if (first) {
(void)printf("Memory resource pool statistics\n");
(void)printf(
"%-*s%*s%*s%5s%*s%s%s%*s%*s%6s%s%6s%6s%6s%5s%s%s\n",
wide ? 16 : 11, "Name",
wide ? 6 : 5, "Size",
wide ? 12 : 9, "Requests",
"Fail",
wide ? 12 : 9, "Releases",
wide ? " InUse" : "",
wide ? " Avail" : "",
wide ? 7 : 6, "Pgreq",
wide ? 7 : 6, "Pgrel",
"Npage",
wide ? " PageSz" : "",
"Hiwat",
"Minpg",
"Maxpg",
"Idle",
wide ? " Flags" : "",
wide ? " Util" : "");
first = 0;
}
if (pp->pr_nget == 0 && !verbose)
continue;
if (pp->pr_maxpages == UINT_MAX)
(void)snprintf(maxp, sizeof(maxp), "inf");
else
(void)snprintf(maxp, sizeof(maxp), "%u",
pp->pr_maxpages);
ovflw = 0;
PRWORD(ovflw, "%-*s", wide ? 16 : 11, 0, name);
PRWORD(ovflw, " %*u", wide ? 6 : 5, 1, pp->pr_size);
PRWORD(ovflw, " %*lu", wide ? 12 : 9, 1, pp->pr_nget);
PRWORD(ovflw, " %*lu", 5, 1, pp->pr_nfail);
PRWORD(ovflw, " %*lu", wide ? 12 : 9, 1, pp->pr_nput);
if (wide)
PRWORD(ovflw, " %*u", 7, 1, pp->pr_nout);
if (wide)
PRWORD(ovflw, " %*u", 6, 1, pp->pr_nitems);
PRWORD(ovflw, " %*lu", wide ? 7 : 6, 1, pp->pr_npagealloc);
PRWORD(ovflw, " %*lu", wide ? 7 : 6, 1, pp->pr_npagefree);
PRWORD(ovflw, " %*u", 6, 1, pp->pr_npages);
if (wide)
PRWORD(ovflw, " %*u", 7, 1, pa.pa_pagesz);
PRWORD(ovflw, " %*u", 6, 1, pp->pr_hiwat);
PRWORD(ovflw, " %*u", 6, 1, pp->pr_minpages);
PRWORD(ovflw, " %*s", 6, 1, maxp);
PRWORD(ovflw, " %*lu", 5, 1, pp->pr_nidle);
if (wide)
PRWORD(ovflw, " 0x%0*x", 4, 1,
pp->pr_flags | pp->pr_roflags);
this_inuse = pp->pr_nout * pp->pr_size;
this_total = pp->pr_npages * pa.pa_pagesz;
if (pp->pr_roflags & PR_RECURSIVE) {
/*
* Don't count in-use memory, since it's part
* of another pool and will be accounted for
* there.
*/
total += (this_total - this_inuse);
} else {
inuse += this_inuse;
total += this_total;
}
if (wide) {
if (this_total == 0)
(void)printf(" ---");
else
(void)printf(" %5.1f%%",
(100.0 * this_inuse) / this_total);
}
(void)printf("\n");
}
inuse /= KILO;
total /= KILO;
(void)printf(
"\nIn use %ldK, total allocated %ldK; utilization %.1f%%\n",
inuse, total, (100.0 * inuse) / total);
}
void
dopoolcache(void)
{
struct pool_cache pool_cache, *pc = &pool_cache;
pool_cache_cpu_t cache_cpu, *cc = &cache_cpu;
TAILQ_HEAD(,pool) pool_head;
struct pool pool, *pp = &pool;
char name[32];
uint64_t cpuhit, cpumiss, tot;
void *addr;
int first, ovflw;
size_t i;
double p;
kread(namelist, X_POOLHEAD, &pool_head, sizeof(pool_head));
addr = TAILQ_FIRST(&pool_head);
for (first = 1; addr != NULL; addr = TAILQ_NEXT(pp, pr_poollist) ) {
deref_kptr(addr, pp, sizeof(*pp), "pool chain trashed");
if (pp->pr_cache == NULL)
continue;
deref_kptr(pp->pr_wchan, name, sizeof(name),
"pool wait channel trashed");
deref_kptr(pp->pr_cache, pc, sizeof(*pc), "pool cache trashed");
name[sizeof(name)-1] = '\0';
cpuhit = 0;
cpumiss = 0;
for (i = 0; i < sizeof(pc->pc_cpus) / sizeof(pc->pc_cpus[0]);
i++) {
if ((addr = pc->pc_cpus[i]) == NULL)
continue;
deref_kptr(addr, cc, sizeof(*cc),
"pool cache cpu trashed");
cpuhit += cc->cc_hits;
cpumiss += cc->cc_misses;
}
if (first) {
(void)printf("Pool cache statistics.\n");
(void)printf("%-*s%*s%*s%*s%*s%*s%*s%*s%*s%*s\n",
12, "Name",
6, "Spin",
6, "GrpSz",
5, "Full",
5, "Emty",
10, "PoolLayer",
11, "CacheLayer",
6, "Hit%",
12, "CpuLayer",
6, "Hit%"
);
first = 0;
}
ovflw = 0;
PRWORD(ovflw, "%-*s", 13, 1, name);
PRWORD(ovflw, " %*llu", 6, 1, (long long)pc->pc_contended);
PRWORD(ovflw, " %*u", 6, 1, pc->pc_pcgsize);
PRWORD(ovflw, " %*u", 5, 1, pc->pc_nfull);
PRWORD(ovflw, " %*u", 5, 1, pc->pc_nempty);
PRWORD(ovflw, " %*llu", 10, 1, (long long)pc->pc_misses);
tot = pc->pc_hits + pc->pc_misses;
p = pc->pc_hits * 100.0 / (tot);
PRWORD(ovflw, " %*llu", 11, 1, (long long)tot);
PRWORD(ovflw, " %*.1f", 6, 1, p);
tot = cpuhit + cpumiss;
p = cpuhit * 100.0 / (tot);
PRWORD(ovflw, " %*llu", 12, 1, (long long)tot);
PRWORD(ovflw, " %*.1f", 6, 1, p);
printf("\n");
}
}
enum hashtype { /* from <sys/systm.h> */
HASH_LIST,
HASH_TAILQ
};
struct uidinfo { /* XXX: no kernel header file */
LIST_ENTRY(uidinfo) ui_hash;
uid_t ui_uid;
long ui_proccnt;
};
struct kernel_hash {
const char * description; /* description */
int hashsize; /* nlist index for hash size */
int hashtbl; /* nlist index for hash table */
enum hashtype type; /* type of hash table */
size_t offset; /* offset of {LIST,TAILQ}_NEXT */
} khashes[] =
{
{
"buffer hash",
X_BUFHASH, X_BUFHASHTBL,
HASH_LIST, offsetof(struct buf, b_hash)
}, {
"inode cache (ihash)",
X_IHASH, X_IHASHTBL,
HASH_LIST, offsetof(struct inode, i_hash)
}, {
"ipv4 address -> interface hash",
X_IFADDRHASH, X_IFADDRHASHTBL,
HASH_LIST, offsetof(struct in_ifaddr, ia_hash),
}, {
"name cache hash",
X_NCHASH, X_NCHASHTBL,
HASH_LIST, offsetof(struct namecache, nc_hash),
}, {
"name cache directory hash",
X_NCVHASH, X_NCVHASHTBL,
HASH_LIST, offsetof(struct namecache, nc_vhash),
}, {
"user info (uid -> used processes) hash",
X_UIHASH, X_UIHASHTBL,
HASH_LIST, offsetof(struct uidinfo, ui_hash),
}, {
NULL, -1, -1, 0, 0,
}
};
void
dohashstat(int verbose, int todo, const char *hashname)
{
LIST_HEAD(, generic) *hashtbl_list;
TAILQ_HEAD(, generic) *hashtbl_tailq;
struct kernel_hash *curhash;
void *hashaddr, *hashbuf, *nhashbuf, *nextaddr;
size_t elemsize, hashbufsize, thissize;
u_long hashsize, i;
int used, items, chain, maxchain;
hashbuf = NULL;
hashbufsize = 0;
if (todo & HASHLIST) {
(void)printf("Supported hashes:\n");
for (curhash = khashes; curhash->description; curhash++) {
if (hashnl[curhash->hashsize].n_value == 0 ||
hashnl[curhash->hashtbl].n_value == 0)
continue;
(void)printf("\t%-16s%s\n",
hashnl[curhash->hashsize].n_name + 1,
curhash->description);
}
return;
}
if (hashname != NULL) {
for (curhash = khashes; curhash->description; curhash++) {
if (strcmp(hashnl[curhash->hashsize].n_name + 1,
hashname) == 0 &&
hashnl[curhash->hashsize].n_value != 0 &&
hashnl[curhash->hashtbl].n_value != 0)
break;
}
if (curhash->description == NULL) {
warnx("%s: no such hash", hashname);
return;
}
}
(void)printf(
"%-16s %8s %8s %8s %8s %8s %8s\n"
"%-16s %8s %8s %8s %8s %8s %8s\n",
"", "total", "used", "util", "num", "average", "maximum",
"hash table", "buckets", "buckets", "%", "items", "chain",
"chain");
for (curhash = khashes; curhash->description; curhash++) {
if (hashnl[curhash->hashsize].n_value == 0 ||
hashnl[curhash->hashtbl].n_value == 0)
continue;
if (hashname != NULL &&
strcmp(hashnl[curhash->hashsize].n_name + 1, hashname))
continue;
elemsize = curhash->type == HASH_LIST ?
sizeof(*hashtbl_list) : sizeof(*hashtbl_tailq);
deref_kptr((void *)hashnl[curhash->hashsize].n_value,
&hashsize, sizeof(hashsize),
hashnl[curhash->hashsize].n_name);
hashsize++;
deref_kptr((void *)hashnl[curhash->hashtbl].n_value,
&hashaddr, sizeof(hashaddr),
hashnl[curhash->hashtbl].n_name);
if (verbose)
(void)printf(
"%s %lu, %s %p, offset %ld, elemsize %llu\n",
hashnl[curhash->hashsize].n_name + 1, hashsize,
hashnl[curhash->hashtbl].n_name + 1, hashaddr,
(long)curhash->offset,
(unsigned long long)elemsize);
thissize = hashsize * elemsize;
if (hashbuf == NULL || thissize > hashbufsize) {
if ((nhashbuf = realloc(hashbuf, thissize)) == NULL)
errx(1, "malloc hashbuf %llu",
(unsigned long long)hashbufsize);
hashbuf = nhashbuf;
hashbufsize = thissize;
}
deref_kptr(hashaddr, hashbuf, thissize,
hashnl[curhash->hashtbl].n_name);
used = 0;
items = maxchain = 0;
if (curhash->type == HASH_LIST) {
hashtbl_list = hashbuf;
hashtbl_tailq = NULL;
} else {
hashtbl_list = NULL;
hashtbl_tailq = hashbuf;
}
for (i = 0; i < hashsize; i++) {
if (curhash->type == HASH_LIST)
nextaddr = LIST_FIRST(&hashtbl_list[i]);
else
nextaddr = TAILQ_FIRST(&hashtbl_tailq[i]);
if (nextaddr == NULL)
continue;
if (verbose)
(void)printf("%5lu: %p\n", i, nextaddr);
used++;
chain = 0;
do {
chain++;
deref_kptr((char *)nextaddr + curhash->offset,
&nextaddr, sizeof(void *),
"hash chain corrupted");
if (verbose > 1)
(void)printf("got nextaddr as %p\n",
nextaddr);
} while (nextaddr != NULL);
items += chain;
if (verbose && chain > 1)
(void)printf("\tchain = %d\n", chain);
if (chain > maxchain)
maxchain = chain;
}
(void)printf("%-16s %8ld %8d %8.2f %8d %8.2f %8d\n",
hashnl[curhash->hashsize].n_name + 1,
hashsize, used, used * 100.0 / hashsize,
items, used ? (double)items / used : 0.0, maxchain);
}
}
/*
* kreadc like kread but returns 1 if sucessful, 0 otherwise
*/
int
kreadc(struct nlist *nl, int nlx, void *addr, size_t size)
{
const char *sym;
sym = nl[nlx].n_name;
if (*sym == '_')
++sym;
if (nl[nlx].n_type == 0 || nl[nlx].n_value == 0)
return 0;
deref_kptr((void *)nl[nlx].n_value, addr, size, sym);
return 1;
}
/*
* kread reads something from the kernel, given its nlist index in namelist[].
*/
void
kread(struct nlist *nl, int nlx, void *addr, size_t size)
{
const char *sym;
sym = nl[nlx].n_name;
if (*sym == '_')
++sym;
if (nl[nlx].n_type == 0 || nl[nlx].n_value == 0)
errx(1, "symbol %s not defined", sym);
deref_kptr((void *)nl[nlx].n_value, addr, size, sym);
}
/*
* Dereference the kernel pointer `kptr' and fill in the local copy
* pointed to by `ptr'. The storage space must be pre-allocated,
* and the size of the copy passed in `len'.
*/
void
deref_kptr(const void *kptr, void *ptr, size_t len, const char *msg)
{
if (*msg == '_')
msg++;
if ((size_t)kvm_read(kd, (u_long)kptr, (char *)ptr, len) != len)
errx(1, "kptr %lx: %s: %s", (u_long)kptr, msg, kvm_geterr(kd));
}
/*
* Traverse the UVM history buffers, performing the requested action.
*
* Note, we assume that if we're not listing, we're dumping.
*/
void
hist_traverse(int todo, const char *histname)
{
struct uvm_history_head histhead;
struct uvm_history hist, *histkva;
char *name = NULL;
size_t namelen = 0;
if (histnl[0].n_value == 0) {
warnx("UVM history is not compiled into the kernel.");
return;
}
deref_kptr((void *)histnl[X_UVM_HISTORIES].n_value, &histhead,
sizeof(histhead), histnl[X_UVM_HISTORIES].n_name);
if (histhead.lh_first == NULL) {
warnx("No active UVM history logs.");
return;
}
if (todo & HISTLIST)
(void)printf("Active UVM histories:");
for (histkva = LIST_FIRST(&histhead); histkva != NULL;
histkva = LIST_NEXT(&hist, list)) {
deref_kptr(histkva, &hist, sizeof(hist), "histkva");
if (name == NULL || hist.namelen > namelen) {
if (name != NULL)
free(name);
namelen = hist.namelen;
if ((name = malloc(namelen + 1)) == NULL)
err(1, "malloc history name");
}
deref_kptr(hist.name, name, namelen, "history name");
name[namelen] = '\0';
if (todo & HISTLIST)
(void)printf(" %s", name);
else {
/*
* If we're dumping all histories, do it, else
* check to see if this is the one we want.
*/
if (histname == NULL || strcmp(histname, name) == 0) {
if (histname == NULL)
(void)printf(
"\nUVM history `%s':\n", name);
hist_dodump(&hist);
}
}
}
if (todo & HISTLIST)
(void)putchar('\n');
if (name != NULL)
free(name);
}
/*
* Actually dump the history buffer at the specified KVA.
*/
void
hist_dodump(struct uvm_history *histp)
{
struct uvm_history_ent *histents, *e;
size_t histsize;
char *fmt = NULL, *fn = NULL;
size_t fmtlen = 0, fnlen = 0;
int i;
histsize = sizeof(struct uvm_history_ent) * histp->n;
if ((histents = malloc(histsize)) == NULL)
err(1, "malloc history entries");
(void)memset(histents, 0, histsize);
deref_kptr(histp->e, histents, histsize, "history entries");
i = histp->f;
do {
e = &histents[i];
if (e->fmt != NULL) {
if (fmt == NULL || e->fmtlen > fmtlen) {
if (fmt != NULL)
free(fmt);
fmtlen = e->fmtlen;
if ((fmt = malloc(fmtlen + 1)) == NULL)
err(1, "malloc printf format");
}
if (fn == NULL || e->fnlen > fnlen) {
if (fn != NULL)
free(fn);
fnlen = e->fnlen;
if ((fn = malloc(fnlen + 1)) == NULL)
err(1, "malloc function name");
}
deref_kptr(e->fmt, fmt, fmtlen, "printf format");
fmt[fmtlen] = '\0';
deref_kptr(e->fn, fn, fnlen, "function name");
fn[fnlen] = '\0';
(void)printf("%06ld.%06ld ", (long int)e->tv.tv_sec,
(long int)e->tv.tv_usec);
(void)printf("%s#%ld: ", fn, e->call);
(void)printf(fmt, e->v[0], e->v[1], e->v[2], e->v[3]);
(void)putchar('\n');
}
i = (i + 1) % histp->n;
} while (i != histp->f);
free(histents);
if (fmt != NULL)
free(fmt);
if (fn != NULL)
free(fn);
}
void
usage(void)
{
(void)fprintf(stderr,
"usage: %s [-CefHiLlmstUvW] [-c count] [-h hashname] [-M core] [-N system]\n"
"\t\t[-u histname] [-w wait] [disks]\n", getprogname());
exit(1);
}