Files
openssh-portable/openbsd-compat/port-linux.c
T
Daniil Tatianin 6c49e5f7dc Add support for locking memory on Linux
Linux wakes up kcompactd threads in order to make more contiguous memory
available on the system, it does this by migrating live movable pages
(actively modifying live processes' page tables and constantly flooding
them with page invalidation IPIs, which can be up to millions per
second), which causes the process to become unresponsive for up to
seconds or even minutes in some severe cases. In case of sshd, we want
to always be able to connect to the system, even if it's under heavy
kcompactd load.

Introduce an option to protect sshd and its children sessions from being
compacted by kcompactd (this works in cojunction with
compact_unevictable_allowed = 0). Note that we depend on MCL_ONFAULT
being available, which was introduced in linux 4.4. MCL_ONFAULT allows
the system to lock pages lazily, thus drastically reducing memory usage
of a locked process (without MCL_ONFAULT, every existing mapping in the
process is instantly write-faulted).
2025-03-23 11:24:03 +11:00

421 lines
10 KiB
C

/*
* Copyright (c) 2005 Daniel Walsh <[email protected]>
* Copyright (c) 2006 Damien Miller <[email protected]>
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
/*
* Linux-specific portability code - just SELinux support at present
*/
#include "includes.h"
#if defined(WITH_SELINUX) || defined(LINUX_OOM_ADJUST) || \
defined(SYSTEMD_NOTIFY)
#include <sys/socket.h>
#include <sys/un.h>
#include <errno.h>
#include <inttypes.h>
#include <stdarg.h>
#include <string.h>
#include <stdio.h>
#include <stdlib.h>
#include <time.h>
#include <unistd.h>
#include "log.h"
#include "xmalloc.h"
#include "port-linux.h"
#include "misc.h"
#ifdef WITH_SELINUX
#include <selinux/selinux.h>
#include <selinux/label.h>
#include <selinux/get_context_list.h>
#ifndef SSH_SELINUX_UNCONFINED_TYPE
# define SSH_SELINUX_UNCONFINED_TYPE ":unconfined_t:"
#endif
/* Wrapper around is_selinux_enabled() to log its return value once only */
int
ssh_selinux_enabled(void)
{
static int enabled = -1;
if (enabled == -1) {
enabled = (is_selinux_enabled() == 1);
debug("SELinux support %s", enabled ? "enabled" : "disabled");
}
return (enabled);
}
/* Return the default security context for the given username */
static char *
ssh_selinux_getctxbyname(char *pwname)
{
char *sc = NULL, *sename = NULL, *lvl = NULL;
int r;
#ifdef HAVE_GETSEUSERBYNAME
if (getseuserbyname(pwname, &sename, &lvl) != 0)
return NULL;
#else
sename = pwname;
lvl = NULL;
#endif
#ifdef HAVE_GET_DEFAULT_CONTEXT_WITH_LEVEL
r = get_default_context_with_level(sename, lvl, NULL, &sc);
#else
r = get_default_context(sename, NULL, &sc);
#endif
if (r != 0) {
switch (security_getenforce()) {
case -1:
fatal("%s: ssh_selinux_getctxbyname: "
"security_getenforce() failed", __func__);
case 0:
error("%s: Failed to get default SELinux security "
"context for %s", __func__, pwname);
sc = NULL;
break;
default:
fatal("%s: Failed to get default SELinux security "
"context for %s (in enforcing mode)",
__func__, pwname);
}
}
#ifdef HAVE_GETSEUSERBYNAME
free(sename);
free(lvl);
#endif
return sc;
}
/* Set the execution context to the default for the specified user */
void
ssh_selinux_setup_exec_context(char *pwname)
{
char *user_ctx = NULL;
if (!ssh_selinux_enabled())
return;
debug3("%s: setting execution context", __func__);
user_ctx = ssh_selinux_getctxbyname(pwname);
if (setexeccon(user_ctx) != 0) {
switch (security_getenforce()) {
case -1:
fatal("%s: security_getenforce() failed", __func__);
case 0:
error("%s: Failed to set SELinux execution "
"context for %s", __func__, pwname);
break;
default:
fatal("%s: Failed to set SELinux execution context "
"for %s (in enforcing mode)", __func__, pwname);
}
}
if (user_ctx != NULL)
freecon(user_ctx);
debug3("%s: done", __func__);
}
/* Set the TTY context for the specified user */
void
ssh_selinux_setup_pty(char *pwname, const char *tty)
{
char *new_tty_ctx = NULL, *user_ctx = NULL, *old_tty_ctx = NULL;
security_class_t chrclass;
if (!ssh_selinux_enabled())
return;
debug3("%s: setting TTY context on %s", __func__, tty);
user_ctx = ssh_selinux_getctxbyname(pwname);
/* XXX: should these calls fatal() upon failure in enforcing mode? */
if (getfilecon(tty, &old_tty_ctx) == -1) {
error("%s: getfilecon: %s", __func__, strerror(errno));
goto out;
}
if ((chrclass = string_to_security_class("chr_file")) == 0) {
error("%s: couldn't get security class for chr_file", __func__);
goto out;
}
if (security_compute_relabel(user_ctx, old_tty_ctx,
chrclass, &new_tty_ctx) != 0) {
error("%s: security_compute_relabel: %s",
__func__, strerror(errno));
goto out;
}
if (setfilecon(tty, new_tty_ctx) != 0)
error("%s: setfilecon: %s", __func__, strerror(errno));
out:
if (new_tty_ctx != NULL)
freecon(new_tty_ctx);
if (old_tty_ctx != NULL)
freecon(old_tty_ctx);
if (user_ctx != NULL)
freecon(user_ctx);
debug3("%s: done", __func__);
}
void
ssh_selinux_change_context(const char *newname)
{
char *oldctx, *newctx, *cx, *cx2;
LogLevel log_level = SYSLOG_LEVEL_INFO;
if (!ssh_selinux_enabled())
return;
if (getcon(&oldctx) < 0) {
logit_f("getcon failed with %s", strerror(errno));
return;
}
if ((cx = strchr(oldctx, ':')) == NULL ||
(cx = strchr(cx + 1, ':')) == NULL ||
(cx - oldctx) >= INT_MAX) {
logit_f("unparsable context %s", oldctx);
return;
}
/*
* Check whether we are attempting to switch away from an unconfined
* security context.
*/
if (strncmp(cx, SSH_SELINUX_UNCONFINED_TYPE,
sizeof(SSH_SELINUX_UNCONFINED_TYPE) - 1) == 0)
log_level = SYSLOG_LEVEL_DEBUG3;
cx2 = strchr(cx + 1, ':');
xasprintf(&newctx, "%.*s%s%s", (int)(cx - oldctx + 1), oldctx,
newname, cx2 == NULL ? "" : cx2);
debug3_f("setting context from '%s' to '%s'", oldctx, newctx);
if (setcon(newctx) < 0)
do_log2_f(log_level, "setcon %s from %s failed with %s",
newctx, oldctx, strerror(errno));
free(oldctx);
free(newctx);
}
void
ssh_selinux_setfscreatecon(const char *path)
{
char *context;
struct selabel_handle *shandle = NULL;
if (!ssh_selinux_enabled())
return;
if (path == NULL) {
setfscreatecon(NULL);
return;
}
if ((shandle = selabel_open(SELABEL_CTX_FILE, NULL, 0)) == NULL) {
debug_f("selabel_open failed");
return;
}
if (selabel_lookup(shandle, &context, path, 0700) == 0)
setfscreatecon(context);
selabel_close(shandle);
}
#endif /* WITH_SELINUX */
#ifdef LINUX_OOM_ADJUST
/*
* The magic "don't kill me" values, old and new, as documented in eg:
* http://lxr.linux.no/#linux+v2.6.32/Documentation/filesystems/proc.txt
* http://lxr.linux.no/#linux+v2.6.36/Documentation/filesystems/proc.txt
*/
static int oom_adj_save = INT_MIN;
static char *oom_adj_path = NULL;
struct {
char *path;
int value;
} oom_adjust[] = {
{"/proc/self/oom_score_adj", -1000}, /* kernels >= 2.6.36 */
{"/proc/self/oom_adj", -17}, /* kernels <= 2.6.35 */
{NULL, 0},
};
/*
* Tell the kernel's out-of-memory killer to avoid sshd.
* Returns the previous oom_adj value or zero.
*/
void
oom_adjust_setup(void)
{
int i, value;
FILE *fp;
debug3("%s", __func__);
for (i = 0; oom_adjust[i].path != NULL; i++) {
oom_adj_path = oom_adjust[i].path;
value = oom_adjust[i].value;
if ((fp = fopen(oom_adj_path, "r+")) != NULL) {
if (fscanf(fp, "%d", &oom_adj_save) != 1)
verbose("error reading %s: %s", oom_adj_path,
strerror(errno));
else {
rewind(fp);
if (fprintf(fp, "%d\n", value) <= 0)
verbose("error writing %s: %s",
oom_adj_path, strerror(errno));
else
debug("Set %s from %d to %d",
oom_adj_path, oom_adj_save, value);
}
fclose(fp);
return;
}
}
oom_adj_path = NULL;
}
/* Restore the saved OOM adjustment */
void
oom_adjust_restore(void)
{
FILE *fp;
debug3("%s", __func__);
if (oom_adj_save == INT_MIN || oom_adj_path == NULL ||
(fp = fopen(oom_adj_path, "w")) == NULL)
return;
if (fprintf(fp, "%d\n", oom_adj_save) <= 0)
verbose("error writing %s: %s", oom_adj_path, strerror(errno));
else
debug("Set %s to %d", oom_adj_path, oom_adj_save);
fclose(fp);
return;
}
#endif /* LINUX_OOM_ADJUST */
#ifdef LINUX_MEMLOCK_ONFAULT
#include <sys/mman.h>
void
memlock_onfault_setup(void)
{
if (mlockall(MCL_CURRENT | MCL_FUTURE | MCL_ONFAULT) < 0)
verbose("unable to lock memory: %s", strerror(errno));
else
debug("memory locked");
}
#endif /* LINUX_MEMLOCK_ONFAULT */
#ifdef SYSTEMD_NOTIFY
static void ssh_systemd_notify(const char *, ...)
__attribute__((__format__ (printf, 1, 2))) __attribute__((__nonnull__ (1)));
static void
ssh_systemd_notify(const char *fmt, ...)
{
char *s = NULL;
const char *path;
struct stat sb;
struct sockaddr_un addr;
int fd = -1;
va_list ap;
if ((path = getenv("NOTIFY_SOCKET")) == NULL || strlen(path) == 0)
return;
va_start(ap, fmt);
xvasprintf(&s, fmt, ap);
va_end(ap);
/* Only AF_UNIX is supported, with path or abstract sockets */
if (path[0] != '/' && path[0] != '@') {
error_f("socket \"%s\" is not compatible with AF_UNIX", path);
goto out;
}
if (path[0] == '/' && stat(path, &sb) != 0) {
error_f("socket \"%s\" stat: %s", path, strerror(errno));
goto out;
}
memset(&addr, 0, sizeof(addr));
addr.sun_family = AF_UNIX;
if (strlcpy(addr.sun_path, path,
sizeof(addr.sun_path)) >= sizeof(addr.sun_path)) {
error_f("socket path \"%s\" too long", path);
goto out;
}
/* Support for abstract socket */
if (addr.sun_path[0] == '@')
addr.sun_path[0] = 0;
if ((fd = socket(PF_UNIX, SOCK_DGRAM, 0)) == -1) {
error_f("socket \"%s\": %s", path, strerror(errno));
goto out;
}
if (connect(fd, (struct sockaddr *)&addr, sizeof(addr)) != 0) {
error_f("socket \"%s\" connect: %s", path, strerror(errno));
goto out;
}
if (write(fd, s, strlen(s)) != (ssize_t)strlen(s)) {
error_f("socket \"%s\" write: %s", path, strerror(errno));
goto out;
}
debug_f("socket \"%s\" notified %s", path, s);
out:
if (fd != -1)
close(fd);
free(s);
}
void
ssh_systemd_notify_ready(void)
{
ssh_systemd_notify("READY=1");
}
void
ssh_systemd_notify_reload(void)
{
struct timespec now;
monotime_ts(&now);
if (now.tv_sec < 0 || now.tv_nsec < 0) {
error_f("monotime returned negative value");
ssh_systemd_notify("RELOADING=1");
} else {
ssh_systemd_notify("RELOADING=1\nMONOTONIC_USEC=%llu",
((uint64_t)now.tv_sec * 1000000ULL) +
((uint64_t)now.tv_nsec / 1000ULL));
}
}
#endif /* SYSTEMD_NOTIFY */
#endif /* WITH_SELINUX || LINUX_OOM_ADJUST || SYSTEMD_NOTIFY */