qemu-e2k/util/oslib-posix.c

801 lines
22 KiB
C
Raw Normal View History

/*
* os-posix-lib.c
*
* Copyright (c) 2003-2008 Fabrice Bellard
* Copyright (c) 2010 Red Hat, Inc.
*
* QEMU library functions on POSIX which are shared between QEMU and
* the QEMU tools.
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
#include "qemu/osdep.h"
#include <termios.h>
#include <glib/gprintf.h>
#include "sysemu/sysemu.h"
#include "trace.h"
2016-03-14 09:01:28 +01:00
#include "qapi/error.h"
#include "qemu/error-report.h"
#include "qemu/madvise.h"
#include "qemu/sockets.h"
#include "qemu/thread.h"
#include <libgen.h>
#include "qemu/cutils.h"
cfi: Initial support for cfi-icall in QEMU LLVM/Clang, supports runtime checks for forward-edge Control-Flow Integrity (CFI). CFI on indirect function calls (cfi-icall) ensures that, in indirect function calls, the function called is of the right signature for the pointer type defined at compile time. For this check to work, the code must always respect the function signature when using function pointer, the function must be defined at compile time, and be compiled with link-time optimization. This rules out, for example, shared libraries that are dynamically loaded (given that functions are not known at compile time), and code that is dynamically generated at run-time. This patch: 1) Introduces the CONFIG_CFI flag to support cfi in QEMU 2) Introduces a decorator to allow the definition of "sensitive" functions, where a non-instrumented function may be called at runtime through a pointer. The decorator will take care of disabling cfi-icall checks on such functions, when cfi is enabled. 3) Marks functions currently in QEMU that exhibit such behavior, in particular: - The function in TCG that calls pre-compiled TBs - The function in TCI that interprets instructions - Functions in the plugin infrastructures that jump to callbacks - Functions in util that directly call a signal handler Signed-off-by: Daniele Buono <dbuono@linux.vnet.ibm.com> Acked-by: Alex Bennée <alex.bennee@linaro.org Message-Id: <20201204230615.2392-3-dbuono@linux.vnet.ibm.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2020-12-05 00:06:12 +01:00
#include "qemu/compiler.h"
#include "qemu/units.h"
#ifdef CONFIG_LINUX
#include <sys/syscall.h>
#endif
#ifdef __FreeBSD__
#include <sys/thr.h>
#include <sys/types.h>
#include <sys/user.h>
#include <libutil.h>
#endif
#ifdef __NetBSD__
#include <lwp.h>
#endif
#ifdef __APPLE__
#include <mach-o/dyld.h>
#endif
#ifdef __HAIKU__
#include <kernel/image.h>
#endif
#include "qemu/mmap-alloc.h"
#ifdef CONFIG_DEBUG_STACK_USAGE
#include "qemu/error-report.h"
#endif
#define MAX_MEM_PREALLOC_THREAD_COUNT 16
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
struct MemsetThread;
typedef struct MemsetContext {
bool all_threads_created;
bool any_thread_failed;
struct MemsetThread *threads;
int num_threads;
} MemsetContext;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
struct MemsetThread {
char *addr;
size_t numpages;
size_t hpagesize;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
QemuThread pgthread;
sigjmp_buf env;
MemsetContext *context;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
};
typedef struct MemsetThread MemsetThread;
/* used by sigbus_handler() */
static MemsetContext *sigbus_memset_context;
struct sigaction sigbus_oldact;
static QemuMutex sigbus_mutex;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
static QemuMutex page_mutex;
static QemuCond page_cond;
int qemu_get_thread_id(void)
{
#if defined(__linux__)
return syscall(SYS_gettid);
#elif defined(__FreeBSD__)
/* thread id is up to INT_MAX */
long tid;
thr_self(&tid);
return (int)tid;
#elif defined(__NetBSD__)
return _lwp_self();
#elif defined(__OpenBSD__)
return getthrid();
#else
return getpid();
#endif
}
int qemu_daemon(int nochdir, int noclose)
{
return daemon(nochdir, noclose);
}
util: add qemu_write_pidfile() There are variants of qemu_create_pidfile() in qemu-pr-helper and qemu-ga. Let's have a common implementation in libqemuutil. The code is initially based from pr-helper write_pidfile(), with various improvements and suggestions from Daniel Berrangé: QEMU will leave the pidfile existing on disk when it exits which initially made me think it avoids the deletion race. The app managing QEMU, however, may well delete the pidfile after it has seen QEMU exit, and even if the app locks the pidfile before deleting it, there is still a race. eg consider the following sequence QEMU 1 libvirtd QEMU 2 1. lock(pidfile) 2. exit() 3. open(pidfile) 4. lock(pidfile) 5. open(pidfile) 6. unlink(pidfile) 7. close(pidfile) 8. lock(pidfile) IOW, at step 8 the new QEMU has successfully acquired the lock, but the pidfile no longer exists on disk because it was deleted after the original QEMU exited. While we could just say no external app should ever delete the pidfile, I don't think that is satisfactory as people don't read docs, and admins don't like stale pidfiles being left around on disk. To make this robust, I think we might want to copy libvirt's approach to pidfile acquisition which runs in a loop and checks that the file on disk /after/ acquiring the lock matches the file that was locked. Then we could in fact safely let QEMU delete its own pidfiles on clean exit.. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20180831145314.14736-2-marcandre.lureau@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-08-31 16:53:12 +02:00
bool qemu_write_pidfile(const char *path, Error **errp)
{
int fd;
char pidstr[32];
while (1) {
struct stat a, b;
struct flock lock = {
.l_type = F_WRLCK,
.l_whence = SEEK_SET,
.l_len = 0,
};
util: add qemu_write_pidfile() There are variants of qemu_create_pidfile() in qemu-pr-helper and qemu-ga. Let's have a common implementation in libqemuutil. The code is initially based from pr-helper write_pidfile(), with various improvements and suggestions from Daniel Berrangé: QEMU will leave the pidfile existing on disk when it exits which initially made me think it avoids the deletion race. The app managing QEMU, however, may well delete the pidfile after it has seen QEMU exit, and even if the app locks the pidfile before deleting it, there is still a race. eg consider the following sequence QEMU 1 libvirtd QEMU 2 1. lock(pidfile) 2. exit() 3. open(pidfile) 4. lock(pidfile) 5. open(pidfile) 6. unlink(pidfile) 7. close(pidfile) 8. lock(pidfile) IOW, at step 8 the new QEMU has successfully acquired the lock, but the pidfile no longer exists on disk because it was deleted after the original QEMU exited. While we could just say no external app should ever delete the pidfile, I don't think that is satisfactory as people don't read docs, and admins don't like stale pidfiles being left around on disk. To make this robust, I think we might want to copy libvirt's approach to pidfile acquisition which runs in a loop and checks that the file on disk /after/ acquiring the lock matches the file that was locked. Then we could in fact safely let QEMU delete its own pidfiles on clean exit.. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20180831145314.14736-2-marcandre.lureau@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-08-31 16:53:12 +02:00
fd = qemu_create(path, O_WRONLY, S_IRUSR | S_IWUSR, errp);
util: add qemu_write_pidfile() There are variants of qemu_create_pidfile() in qemu-pr-helper and qemu-ga. Let's have a common implementation in libqemuutil. The code is initially based from pr-helper write_pidfile(), with various improvements and suggestions from Daniel Berrangé: QEMU will leave the pidfile existing on disk when it exits which initially made me think it avoids the deletion race. The app managing QEMU, however, may well delete the pidfile after it has seen QEMU exit, and even if the app locks the pidfile before deleting it, there is still a race. eg consider the following sequence QEMU 1 libvirtd QEMU 2 1. lock(pidfile) 2. exit() 3. open(pidfile) 4. lock(pidfile) 5. open(pidfile) 6. unlink(pidfile) 7. close(pidfile) 8. lock(pidfile) IOW, at step 8 the new QEMU has successfully acquired the lock, but the pidfile no longer exists on disk because it was deleted after the original QEMU exited. While we could just say no external app should ever delete the pidfile, I don't think that is satisfactory as people don't read docs, and admins don't like stale pidfiles being left around on disk. To make this robust, I think we might want to copy libvirt's approach to pidfile acquisition which runs in a loop and checks that the file on disk /after/ acquiring the lock matches the file that was locked. Then we could in fact safely let QEMU delete its own pidfiles on clean exit.. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20180831145314.14736-2-marcandre.lureau@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-08-31 16:53:12 +02:00
if (fd == -1) {
return false;
}
if (fstat(fd, &b) < 0) {
error_setg_errno(errp, errno, "Cannot stat file");
goto fail_close;
}
if (fcntl(fd, F_SETLK, &lock)) {
util: add qemu_write_pidfile() There are variants of qemu_create_pidfile() in qemu-pr-helper and qemu-ga. Let's have a common implementation in libqemuutil. The code is initially based from pr-helper write_pidfile(), with various improvements and suggestions from Daniel Berrangé: QEMU will leave the pidfile existing on disk when it exits which initially made me think it avoids the deletion race. The app managing QEMU, however, may well delete the pidfile after it has seen QEMU exit, and even if the app locks the pidfile before deleting it, there is still a race. eg consider the following sequence QEMU 1 libvirtd QEMU 2 1. lock(pidfile) 2. exit() 3. open(pidfile) 4. lock(pidfile) 5. open(pidfile) 6. unlink(pidfile) 7. close(pidfile) 8. lock(pidfile) IOW, at step 8 the new QEMU has successfully acquired the lock, but the pidfile no longer exists on disk because it was deleted after the original QEMU exited. While we could just say no external app should ever delete the pidfile, I don't think that is satisfactory as people don't read docs, and admins don't like stale pidfiles being left around on disk. To make this robust, I think we might want to copy libvirt's approach to pidfile acquisition which runs in a loop and checks that the file on disk /after/ acquiring the lock matches the file that was locked. Then we could in fact safely let QEMU delete its own pidfiles on clean exit.. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20180831145314.14736-2-marcandre.lureau@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-08-31 16:53:12 +02:00
error_setg_errno(errp, errno, "Cannot lock pid file");
goto fail_close;
}
/*
* Now make sure the path we locked is the same one that now
* exists on the filesystem.
*/
if (stat(path, &a) < 0) {
/*
* PID file disappeared, someone else must be racing with
* us, so try again.
*/
close(fd);
continue;
}
if (a.st_ino == b.st_ino) {
break;
}
/*
* PID file was recreated, someone else must be racing with
* us, so try again.
*/
close(fd);
}
if (ftruncate(fd, 0) < 0) {
error_setg_errno(errp, errno, "Failed to truncate pid file");
goto fail_unlink;
}
snprintf(pidstr, sizeof(pidstr), FMT_pid "\n", getpid());
if (qemu_write_full(fd, pidstr, strlen(pidstr)) != strlen(pidstr)) {
util: add qemu_write_pidfile() There are variants of qemu_create_pidfile() in qemu-pr-helper and qemu-ga. Let's have a common implementation in libqemuutil. The code is initially based from pr-helper write_pidfile(), with various improvements and suggestions from Daniel Berrangé: QEMU will leave the pidfile existing on disk when it exits which initially made me think it avoids the deletion race. The app managing QEMU, however, may well delete the pidfile after it has seen QEMU exit, and even if the app locks the pidfile before deleting it, there is still a race. eg consider the following sequence QEMU 1 libvirtd QEMU 2 1. lock(pidfile) 2. exit() 3. open(pidfile) 4. lock(pidfile) 5. open(pidfile) 6. unlink(pidfile) 7. close(pidfile) 8. lock(pidfile) IOW, at step 8 the new QEMU has successfully acquired the lock, but the pidfile no longer exists on disk because it was deleted after the original QEMU exited. While we could just say no external app should ever delete the pidfile, I don't think that is satisfactory as people don't read docs, and admins don't like stale pidfiles being left around on disk. To make this robust, I think we might want to copy libvirt's approach to pidfile acquisition which runs in a loop and checks that the file on disk /after/ acquiring the lock matches the file that was locked. Then we could in fact safely let QEMU delete its own pidfiles on clean exit.. Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Message-Id: <20180831145314.14736-2-marcandre.lureau@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-08-31 16:53:12 +02:00
error_setg(errp, "Failed to write pid file");
goto fail_unlink;
}
return true;
fail_unlink:
unlink(path);
fail_close:
close(fd);
return false;
}
/* alloc shared memory pages */
void *qemu_anon_ram_alloc(size_t size, uint64_t *alignment, bool shared,
bool noreserve)
{
const uint32_t qemu_map_flags = (shared ? QEMU_MAP_SHARED : 0) |
(noreserve ? QEMU_MAP_NORESERVE : 0);
size_t align = QEMU_VMALLOC_ALIGN;
void *ptr = qemu_ram_mmap(-1, size, align, qemu_map_flags, 0);
if (ptr == MAP_FAILED) {
return NULL;
}
if (alignment) {
*alignment = align;
}
trace_qemu_anon_ram_alloc(size, ptr);
return ptr;
}
void qemu_anon_ram_free(void *ptr, size_t size)
{
trace_qemu_anon_ram_free(ptr, size);
qemu_ram_munmap(-1, ptr, size);
}
void qemu_socket_set_block(int fd)
{
g_unix_set_fd_nonblocking(fd, false, NULL);
}
int qemu_socket_try_set_nonblock(int fd)
{
return g_unix_set_fd_nonblocking(fd, true, NULL) ? 0 : -errno;
net: check if the file descriptor is valid before using it qemu_set_nonblock() checks that the file descriptor can be used and, if not, crashes QEMU. An assert() is used for that. The use of assert() is used to detect programming error and the coredump will allow to debug the problem. But in the case of the tap device, this assert() can be triggered by a misconfiguration by the user. At startup, it's not a real problem, but it can also happen during the hot-plug of a new device, and here it's a problem because we can crash a perfectly healthy system. For instance: # ip link add link virbr0 name macvtap0 type macvtap mode bridge # ip link set macvtap0 up # TAP=/dev/tap$(ip -o link show macvtap0 | cut -d: -f1) # qemu-system-x86_64 -machine q35 -device pcie-root-port,id=pcie-root-port-0 -monitor stdio 9<> $TAP (qemu) netdev_add type=tap,id=hostnet0,vhost=on,fd=9 (qemu) device_add driver=virtio-net-pci,netdev=hostnet0,id=net0,bus=pcie-root-port-0 (qemu) device_del net0 (qemu) netdev_del hostnet0 (qemu) netdev_add type=tap,id=hostnet1,vhost=on,fd=9 qemu-system-x86_64: .../util/oslib-posix.c:247: qemu_set_nonblock: Assertion `f != -1' failed. Aborted (core dumped) To avoid that, add a function, qemu_try_set_nonblock(), that allows to report the problem without crashing. In the same way, we also update the function for vhostfd in net_init_tap_one() and for fd in net_init_socket() (both descriptors are provided by the user and can be wrong). Signed-off-by: Laurent Vivier <lvivier@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2020-07-07 20:45:14 +02:00
}
void qemu_socket_set_nonblock(int fd)
net: check if the file descriptor is valid before using it qemu_set_nonblock() checks that the file descriptor can be used and, if not, crashes QEMU. An assert() is used for that. The use of assert() is used to detect programming error and the coredump will allow to debug the problem. But in the case of the tap device, this assert() can be triggered by a misconfiguration by the user. At startup, it's not a real problem, but it can also happen during the hot-plug of a new device, and here it's a problem because we can crash a perfectly healthy system. For instance: # ip link add link virbr0 name macvtap0 type macvtap mode bridge # ip link set macvtap0 up # TAP=/dev/tap$(ip -o link show macvtap0 | cut -d: -f1) # qemu-system-x86_64 -machine q35 -device pcie-root-port,id=pcie-root-port-0 -monitor stdio 9<> $TAP (qemu) netdev_add type=tap,id=hostnet0,vhost=on,fd=9 (qemu) device_add driver=virtio-net-pci,netdev=hostnet0,id=net0,bus=pcie-root-port-0 (qemu) device_del net0 (qemu) netdev_del hostnet0 (qemu) netdev_add type=tap,id=hostnet1,vhost=on,fd=9 qemu-system-x86_64: .../util/oslib-posix.c:247: qemu_set_nonblock: Assertion `f != -1' failed. Aborted (core dumped) To avoid that, add a function, qemu_try_set_nonblock(), that allows to report the problem without crashing. In the same way, we also update the function for vhostfd in net_init_tap_one() and for fd in net_init_socket() (both descriptors are provided by the user and can be wrong). Signed-off-by: Laurent Vivier <lvivier@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2020-07-07 20:45:14 +02:00
{
int f;
f = qemu_socket_try_set_nonblock(fd);
net: check if the file descriptor is valid before using it qemu_set_nonblock() checks that the file descriptor can be used and, if not, crashes QEMU. An assert() is used for that. The use of assert() is used to detect programming error and the coredump will allow to debug the problem. But in the case of the tap device, this assert() can be triggered by a misconfiguration by the user. At startup, it's not a real problem, but it can also happen during the hot-plug of a new device, and here it's a problem because we can crash a perfectly healthy system. For instance: # ip link add link virbr0 name macvtap0 type macvtap mode bridge # ip link set macvtap0 up # TAP=/dev/tap$(ip -o link show macvtap0 | cut -d: -f1) # qemu-system-x86_64 -machine q35 -device pcie-root-port,id=pcie-root-port-0 -monitor stdio 9<> $TAP (qemu) netdev_add type=tap,id=hostnet0,vhost=on,fd=9 (qemu) device_add driver=virtio-net-pci,netdev=hostnet0,id=net0,bus=pcie-root-port-0 (qemu) device_del net0 (qemu) netdev_del hostnet0 (qemu) netdev_add type=tap,id=hostnet1,vhost=on,fd=9 qemu-system-x86_64: .../util/oslib-posix.c:247: qemu_set_nonblock: Assertion `f != -1' failed. Aborted (core dumped) To avoid that, add a function, qemu_try_set_nonblock(), that allows to report the problem without crashing. In the same way, we also update the function for vhostfd in net_init_tap_one() and for fd in net_init_socket() (both descriptors are provided by the user and can be wrong). Signed-off-by: Laurent Vivier <lvivier@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2020-07-07 20:45:14 +02:00
assert(f == 0);
}
int socket_set_fast_reuse(int fd)
{
int val = 1, ret;
ret = setsockopt(fd, SOL_SOCKET, SO_REUSEADDR,
(const char *)&val, sizeof(val));
assert(ret == 0);
return ret;
}
void qemu_set_cloexec(int fd)
{
int f;
f = fcntl(fd, F_GETFD);
assert(f != -1);
f = fcntl(fd, F_SETFD, f | FD_CLOEXEC);
assert(f != -1);
}
char *
qemu_get_local_state_dir(void)
{
return get_relocated_path(CONFIG_QEMU_LOCALSTATEDIR);
}
void qemu_set_tty_echo(int fd, bool echo)
{
struct termios tty;
tcgetattr(fd, &tty);
if (echo) {
tty.c_lflag |= ECHO | ECHONL | ICANON | IEXTEN;
} else {
tty.c_lflag &= ~(ECHO | ECHONL | ICANON | IEXTEN);
}
tcsetattr(fd, TCSANOW, &tty);
}
#ifdef CONFIG_LINUX
static void sigbus_handler(int signal, siginfo_t *siginfo, void *ctx)
#else /* CONFIG_LINUX */
static void sigbus_handler(int signal)
#endif /* CONFIG_LINUX */
{
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
int i;
if (sigbus_memset_context) {
for (i = 0; i < sigbus_memset_context->num_threads; i++) {
MemsetThread *thread = &sigbus_memset_context->threads[i];
if (qemu_thread_is_self(&thread->pgthread)) {
siglongjmp(thread->env, 1);
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
}
}
}
#ifdef CONFIG_LINUX
/*
* We assume that the MCE SIGBUS handler could have been registered. We
* should never receive BUS_MCEERR_AO on any of our threads, but only on
* the main thread registered for PR_MCE_KILL_EARLY. Further, we should not
* receive BUS_MCEERR_AR triggered by action of other threads on one of
* our threads. So, no need to check for unrelated SIGBUS when seeing one
* for our threads.
*
* We will forward to the MCE handler, which will either handle the SIGBUS
* or reinstall the default SIGBUS handler and reraise the SIGBUS. The
* default SIGBUS handler will crash the process, so we don't care.
*/
if (sigbus_oldact.sa_flags & SA_SIGINFO) {
sigbus_oldact.sa_sigaction(signal, siginfo, ctx);
return;
}
#endif /* CONFIG_LINUX */
warn_report("os_mem_prealloc: unrelated SIGBUS detected and ignored");
}
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
static void *do_touch_pages(void *arg)
{
MemsetThread *memset_args = (MemsetThread *)arg;
sigset_t set, oldset;
int ret = 0;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
/*
* On Linux, the page faults from the loop below can cause mmap_sem
* contention with allocation of the thread stacks. Do not start
* clearing until all threads have been created.
*/
qemu_mutex_lock(&page_mutex);
while (!memset_args->context->all_threads_created) {
qemu_cond_wait(&page_cond, &page_mutex);
}
qemu_mutex_unlock(&page_mutex);
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
/* unblock SIGBUS */
sigemptyset(&set);
sigaddset(&set, SIGBUS);
pthread_sigmask(SIG_UNBLOCK, &set, &oldset);
if (sigsetjmp(memset_args->env, 1)) {
ret = -EFAULT;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
} else {
char *addr = memset_args->addr;
size_t numpages = memset_args->numpages;
size_t hpagesize = memset_args->hpagesize;
size_t i;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
for (i = 0; i < numpages; i++) {
/*
* Read & write back the same value, so we don't
* corrupt existing user/app data that might be
* stored.
*
* 'volatile' to stop compiler optimizing this away
* to a no-op
*/
*(volatile char *)addr = *addr;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
addr += hpagesize;
}
}
pthread_sigmask(SIG_SETMASK, &oldset, NULL);
return (void *)(uintptr_t)ret;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
}
static void *do_madv_populate_write_pages(void *arg)
{
MemsetThread *memset_args = (MemsetThread *)arg;
const size_t size = memset_args->numpages * memset_args->hpagesize;
char * const addr = memset_args->addr;
int ret = 0;
/* See do_touch_pages(). */
qemu_mutex_lock(&page_mutex);
while (!memset_args->context->all_threads_created) {
qemu_cond_wait(&page_cond, &page_mutex);
}
qemu_mutex_unlock(&page_mutex);
if (size && qemu_madvise(addr, size, QEMU_MADV_POPULATE_WRITE)) {
ret = -errno;
}
return (void *)(uintptr_t)ret;
}
static inline int get_memset_num_threads(size_t hpagesize, size_t numpages,
int smp_cpus)
{
long host_procs = sysconf(_SC_NPROCESSORS_ONLN);
int ret = 1;
if (host_procs > 0) {
ret = MIN(MIN(host_procs, MAX_MEM_PREALLOC_THREAD_COUNT), smp_cpus);
}
/* Especially with gigantic pages, don't create more threads than pages. */
ret = MIN(ret, numpages);
/* Don't start threads to prealloc comparatively little memory. */
ret = MIN(ret, MAX(1, hpagesize * numpages / (64 * MiB)));
/* In case sysconf() fails, we fall back to single threaded */
return ret;
}
static int touch_all_pages(char *area, size_t hpagesize, size_t numpages,
int smp_cpus, bool use_madv_populate_write)
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
{
static gsize initialized = 0;
MemsetContext context = {
.num_threads = get_memset_num_threads(hpagesize, numpages, smp_cpus),
};
size_t numpages_per_thread, leftover;
void *(*touch_fn)(void *);
int ret = 0, i = 0;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
char *addr = area;
if (g_once_init_enter(&initialized)) {
qemu_mutex_init(&page_mutex);
qemu_cond_init(&page_cond);
g_once_init_leave(&initialized, 1);
}
if (use_madv_populate_write) {
/* Avoid creating a single thread for MADV_POPULATE_WRITE */
if (context.num_threads == 1) {
if (qemu_madvise(area, hpagesize * numpages,
QEMU_MADV_POPULATE_WRITE)) {
return -errno;
}
return 0;
}
touch_fn = do_madv_populate_write_pages;
} else {
touch_fn = do_touch_pages;
}
context.threads = g_new0(MemsetThread, context.num_threads);
numpages_per_thread = numpages / context.num_threads;
leftover = numpages % context.num_threads;
for (i = 0; i < context.num_threads; i++) {
context.threads[i].addr = addr;
context.threads[i].numpages = numpages_per_thread + (i < leftover);
context.threads[i].hpagesize = hpagesize;
context.threads[i].context = &context;
qemu_thread_create(&context.threads[i].pgthread, "touch_pages",
touch_fn, &context.threads[i],
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
QEMU_THREAD_JOINABLE);
addr += context.threads[i].numpages * hpagesize;
}
if (!use_madv_populate_write) {
sigbus_memset_context = &context;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
}
qemu_mutex_lock(&page_mutex);
context.all_threads_created = true;
qemu_cond_broadcast(&page_cond);
qemu_mutex_unlock(&page_mutex);
for (i = 0; i < context.num_threads; i++) {
int tmp = (uintptr_t)qemu_thread_join(&context.threads[i].pgthread);
if (tmp) {
ret = tmp;
}
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
}
if (!use_madv_populate_write) {
sigbus_memset_context = NULL;
}
g_free(context.threads);
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
return ret;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
}
static bool madv_populate_write_possible(char *area, size_t pagesize)
{
return !qemu_madvise(area, pagesize, QEMU_MADV_POPULATE_WRITE) ||
errno != EINVAL;
}
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
void os_mem_prealloc(int fd, char *area, size_t memory, int smp_cpus,
Error **errp)
{
static gsize initialized;
int ret;
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
size_t hpagesize = qemu_fd_getpagesize(fd);
size_t numpages = DIV_ROUND_UP(memory, hpagesize);
bool use_madv_populate_write;
struct sigaction act;
/*
* Sense on every invocation, as MADV_POPULATE_WRITE cannot be used for
* some special mappings, such as mapping /dev/mem.
*/
use_madv_populate_write = madv_populate_write_possible(area, hpagesize);
if (!use_madv_populate_write) {
if (g_once_init_enter(&initialized)) {
qemu_mutex_init(&sigbus_mutex);
g_once_init_leave(&initialized, 1);
}
qemu_mutex_lock(&sigbus_mutex);
memset(&act, 0, sizeof(act));
#ifdef CONFIG_LINUX
act.sa_sigaction = &sigbus_handler;
act.sa_flags = SA_SIGINFO;
#else /* CONFIG_LINUX */
act.sa_handler = &sigbus_handler;
act.sa_flags = 0;
#endif /* CONFIG_LINUX */
ret = sigaction(SIGBUS, &act, &sigbus_oldact);
if (ret) {
qemu_mutex_unlock(&sigbus_mutex);
error_setg_errno(errp, errno,
"os_mem_prealloc: failed to install signal handler");
return;
}
}
mem-prealloc: reduce large guest start-up and migration time. Using "-mem-prealloc" option for a large guest leads to higher guest start-up and migration time. This is because with "-mem-prealloc" option qemu tries to map every guest page (create address translations), and make sure the pages are available during runtime. virsh/libvirt by default, seems to use "-mem-prealloc" option in case the guest is configured to use huge pages. The patch tries to map all guest pages simultaneously by spawning multiple threads. Currently limiting the change to QEMU library functions on POSIX compliant host only, as we are not sure if the problem exists on win32. Below are some stats with "-mem-prealloc" option for guest configured to use huge pages. ------------------------------------------------------------------------ Idle Guest | Start-up time | Migration time ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - single threaded (existing code) ------------------------------------------------------------------------ 64 Core - 4TB | 54m11.796s | 75m43.843s 64 Core - 1TB | 8m56.576s | 14m29.049s 64 Core - 256GB | 2m11.245s | 3m26.598s ------------------------------------------------------------------------ Guest stats with 2M HugePage usage - map guest pages using 8 threads ------------------------------------------------------------------------ 64 Core - 4TB | 5m1.027s | 34m10.565s 64 Core - 1TB | 1m10.366s | 8m28.188s 64 Core - 256GB | 0m19.040s | 2m10.148s ----------------------------------------------------------------------- Guest stats with 2M HugePage usage - map guest pages using 16 threads ----------------------------------------------------------------------- 64 Core - 4TB | 1m58.970s | 31m43.400s 64 Core - 1TB | 0m39.885s | 7m55.289s 64 Core - 256GB | 0m11.960s | 2m0.135s ----------------------------------------------------------------------- Changed in v2: - modify number of memset threads spawned to min(smp_cpus, 16). - removed 64GB memory restriction for spawning memset threads. Changed in v3: - limit number of threads spawned based on min(sysconf(_SC_NPROCESSORS_ONLN), 16, smp_cpus) - implement memset thread specific siglongjmp in SIGBUS signal_handler. Changed in v4 - remove sigsetjmp/siglongjmp and SIGBUS unblock/block for main thread as main thread no longer touches any pages. - simplify code my returning memset_thread_failed status from touch_all_pages. Signed-off-by: Jitendra Kolhe <jitendra.kolhe@hpe.com> Message-Id: <1487907103-32350-1-git-send-email-jitendra.kolhe@hpe.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2017-02-24 04:31:43 +01:00
/* touch pages simultaneously */
ret = touch_all_pages(area, hpagesize, numpages, smp_cpus,
use_madv_populate_write);
if (ret) {
error_setg_errno(errp, -ret,
"os_mem_prealloc: preallocating memory failed");
}
if (!use_madv_populate_write) {
ret = sigaction(SIGBUS, &sigbus_oldact, NULL);
if (ret) {
/* Terminate QEMU since it can't recover from error */
perror("os_mem_prealloc: failed to reinstall signal handler");
exit(1);
}
qemu_mutex_unlock(&sigbus_mutex);
}
}
char *qemu_get_pid_name(pid_t pid)
{
char *name = NULL;
#if defined(__FreeBSD__)
/* BSDs don't have /proc, but they provide a nice substitute */
struct kinfo_proc *proc = kinfo_getproc(pid);
if (proc) {
name = g_strdup(proc->ki_comm);
free(proc);
}
#else
/* Assume a system with reasonable procfs */
char *pid_path;
size_t len;
pid_path = g_strdup_printf("/proc/%d/cmdline", pid);
g_file_get_contents(pid_path, &name, &len, NULL);
g_free(pid_path);
#endif
return name;
}
pid_t qemu_fork(Error **errp)
{
sigset_t oldmask, newmask;
struct sigaction sig_action;
int saved_errno;
pid_t pid;
/*
* Need to block signals now, so that child process can safely
* kill off caller's signal handlers without a race.
*/
sigfillset(&newmask);
if (pthread_sigmask(SIG_SETMASK, &newmask, &oldmask) != 0) {
error_setg_errno(errp, errno,
"cannot block signals");
return -1;
}
pid = fork();
saved_errno = errno;
if (pid < 0) {
/* attempt to restore signal mask, but ignore failure, to
* avoid obscuring the fork failure */
(void)pthread_sigmask(SIG_SETMASK, &oldmask, NULL);
error_setg_errno(errp, saved_errno,
"cannot fork child process");
errno = saved_errno;
return -1;
} else if (pid) {
/* parent process */
/* Restore our original signal mask now that the child is
* safely running. Only documented failures are EFAULT (not
* possible, since we are using just-grabbed mask) or EINVAL
* (not possible, since we are using correct arguments). */
(void)pthread_sigmask(SIG_SETMASK, &oldmask, NULL);
} else {
/* child process */
size_t i;
/* Clear out all signal handlers from parent so nothing
* unexpected can happen in our child once we unblock
* signals */
sig_action.sa_handler = SIG_DFL;
sig_action.sa_flags = 0;
sigemptyset(&sig_action.sa_mask);
for (i = 1; i < NSIG; i++) {
/* Only possible errors are EFAULT or EINVAL The former
* won't happen, the latter we expect, so no need to check
* return value */
(void)sigaction(i, &sig_action, NULL);
}
/* Unmask all signals in child, since we've no idea what the
* caller's done with their signal mask and don't want to
* propagate that to children */
sigemptyset(&newmask);
if (pthread_sigmask(SIG_SETMASK, &newmask, NULL) != 0) {
Error *local_err = NULL;
error_setg_errno(&local_err, errno,
"cannot unblock signals");
error_report_err(local_err);
_exit(1);
}
}
return pid;
}
void *qemu_alloc_stack(size_t *sz)
{
void *ptr, *guardpage;
int flags;
#ifdef CONFIG_DEBUG_STACK_USAGE
void *ptr2;
#endif
size_t pagesz = qemu_real_host_page_size();
#ifdef _SC_THREAD_STACK_MIN
/* avoid stacks smaller than _SC_THREAD_STACK_MIN */
long min_stack_sz = sysconf(_SC_THREAD_STACK_MIN);
*sz = MAX(MAX(min_stack_sz, 0), *sz);
#endif
/* adjust stack size to a multiple of the page size */
*sz = ROUND_UP(*sz, pagesz);
/* allocate one extra page for the guard page */
*sz += pagesz;
flags = MAP_PRIVATE | MAP_ANONYMOUS;
#if defined(MAP_STACK) && defined(__OpenBSD__)
/* Only enable MAP_STACK on OpenBSD. Other OS's such as
* Linux/FreeBSD/NetBSD have a flag with the same name
* but have differing functionality. OpenBSD will SEGV
* if it spots execution with a stack pointer pointing
* at memory that was not allocated with MAP_STACK.
*/
flags |= MAP_STACK;
#endif
ptr = mmap(NULL, *sz, PROT_READ | PROT_WRITE, flags, -1, 0);
if (ptr == MAP_FAILED) {
perror("failed to allocate memory for stack");
abort();
}
#if defined(HOST_IA64)
/* separate register stack */
guardpage = ptr + (((*sz - pagesz) / 2) & ~pagesz);
#elif defined(HOST_HPPA)
/* stack grows up */
guardpage = ptr + *sz - pagesz;
#else
/* stack grows down */
guardpage = ptr;
#endif
if (mprotect(guardpage, pagesz, PROT_NONE) != 0) {
perror("failed to set up stack guard page");
abort();
}
#ifdef CONFIG_DEBUG_STACK_USAGE
for (ptr2 = ptr + pagesz; ptr2 < ptr + *sz; ptr2 += sizeof(uint32_t)) {
*(uint32_t *)ptr2 = 0xdeadbeaf;
}
#endif
return ptr;
}
#ifdef CONFIG_DEBUG_STACK_USAGE
static __thread unsigned int max_stack_usage;
#endif
void qemu_free_stack(void *stack, size_t sz)
{
#ifdef CONFIG_DEBUG_STACK_USAGE
unsigned int usage;
void *ptr;
for (ptr = stack + qemu_real_host_page_size(); ptr < stack + sz;
ptr += sizeof(uint32_t)) {
if (*(uint32_t *)ptr != 0xdeadbeaf) {
break;
}
}
usage = sz - (uintptr_t) (ptr - stack);
if (usage > max_stack_usage) {
error_report("thread %d max stack usage increased from %u to %u",
qemu_get_thread_id(), max_stack_usage, usage);
max_stack_usage = usage;
}
#endif
munmap(stack, sz);
}
cfi: Initial support for cfi-icall in QEMU LLVM/Clang, supports runtime checks for forward-edge Control-Flow Integrity (CFI). CFI on indirect function calls (cfi-icall) ensures that, in indirect function calls, the function called is of the right signature for the pointer type defined at compile time. For this check to work, the code must always respect the function signature when using function pointer, the function must be defined at compile time, and be compiled with link-time optimization. This rules out, for example, shared libraries that are dynamically loaded (given that functions are not known at compile time), and code that is dynamically generated at run-time. This patch: 1) Introduces the CONFIG_CFI flag to support cfi in QEMU 2) Introduces a decorator to allow the definition of "sensitive" functions, where a non-instrumented function may be called at runtime through a pointer. The decorator will take care of disabling cfi-icall checks on such functions, when cfi is enabled. 3) Marks functions currently in QEMU that exhibit such behavior, in particular: - The function in TCG that calls pre-compiled TBs - The function in TCI that interprets instructions - Functions in the plugin infrastructures that jump to callbacks - Functions in util that directly call a signal handler Signed-off-by: Daniele Buono <dbuono@linux.vnet.ibm.com> Acked-by: Alex Bennée <alex.bennee@linaro.org Message-Id: <20201204230615.2392-3-dbuono@linux.vnet.ibm.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2020-12-05 00:06:12 +01:00
/*
* Disable CFI checks.
* We are going to call a signal hander directly. Such handler may or may not
* have been defined in our binary, so there's no guarantee that the pointer
* used to set the handler is a cfi-valid pointer. Since the handlers are
* stored in kernel memory, changing the handler to an attacker-defined
* function requires being able to call a sigaction() syscall,
* which is not as easy as overwriting a pointer in memory.
*/
QEMU_DISABLE_CFI
void sigaction_invoke(struct sigaction *action,
struct qemu_signalfd_siginfo *info)
{
siginfo_t si = {};
si.si_signo = info->ssi_signo;
si.si_errno = info->ssi_errno;
si.si_code = info->ssi_code;
/* Convert the minimal set of fields defined by POSIX.
* Positive si_code values are reserved for kernel-generated
* signals, where the valid siginfo fields are determined by
* the signal number. But according to POSIX, it is unspecified
* whether SI_USER and SI_QUEUE have values less than or equal to
* zero.
*/
if (info->ssi_code == SI_USER || info->ssi_code == SI_QUEUE ||
info->ssi_code <= 0) {
/* SIGTERM, etc. */
si.si_pid = info->ssi_pid;
si.si_uid = info->ssi_uid;
} else if (info->ssi_signo == SIGILL || info->ssi_signo == SIGFPE ||
info->ssi_signo == SIGSEGV || info->ssi_signo == SIGBUS) {
si.si_addr = (void *)(uintptr_t)info->ssi_addr;
} else if (info->ssi_signo == SIGCHLD) {
si.si_pid = info->ssi_pid;
si.si_status = info->ssi_status;
si.si_uid = info->ssi_uid;
}
action->sa_sigaction(info->ssi_signo, &si, NULL);
}
size_t qemu_get_host_physmem(void)
{
#ifdef _SC_PHYS_PAGES
long pages = sysconf(_SC_PHYS_PAGES);
if (pages > 0) {
if (pages > SIZE_MAX / qemu_real_host_page_size()) {
return SIZE_MAX;
} else {
return pages * qemu_real_host_page_size();
}
}
#endif
return 0;
}
int qemu_msync(void *addr, size_t length, int fd)
{
size_t align_mask = ~(qemu_real_host_page_size() - 1);
/**
* There are no strict reqs as per the length of mapping
* to be synced. Still the length needs to follow the address
* alignment changes. Additionally - round the size to the multiple
* of PAGE_SIZE
*/
length += ((uintptr_t)addr & (qemu_real_host_page_size() - 1));
length = (length + ~align_mask) & align_mask;
addr = (void *)((uintptr_t)addr & align_mask);
return msync(addr, length, MS_SYNC);
}