Files
qemu/migration/cpr.c
hongmianquan a181df93bb migration/cpr: use hashtable for cpr fds
Use a GHashTable to store cpr fds to reduce the time
consumption of `cpr_find_fd` in scenarios with a large
number of fds. The time complexity for `cpr_find_fd` is
reduced from O(N) to O(1).
Keep cpr fds lookups in a GHashTable during normal runtime
while preserving the existing QLIST migration ABI. Build a
temporary QLIST from the hash table in pre_save and rebuild
the hash table from the loaded QLIST in post_load.

To demonstrate the performance improvement, we tested the total time
consumed by `cpr_find_fd` (called N times for N fds) under our real-world
business scenarios with different numbers of file descriptors. The results
are measured in nanoseconds:

| Number of FDs | Total time with QLIST (ns) | Total time with GHashTable (ns) |
|---------------|----------------------------|---------------------------------|
| 540           | 936,753                    | 393,358                         |
| 2,870         | 24,102,342                 | 2,212,113                       |
| 7,530         | 152,715,916                | 5,474,310                       |

As shown in the data, the lookup time grows exponentially with the QLIST
as the number of fds increases. With the GHashTable, the time consumption
remains linear (O(1) per lookup), significantly reducing the downtime during
the CPR process.

Signed-off-by: hongmianquan <hongmianquan@bytedance.com>
Link: https://lore.kernel.org/r/20260519134315.27997-1-hongmianquan@bytedance.com
Signed-off-by: Peter Xu <peterx@redhat.com>
2026-05-20 15:49:10 -04:00

411 lines
9.7 KiB
C

/*
* Copyright (c) 2021-2024 Oracle and/or its affiliates.
*
* This work is licensed under the terms of the GNU GPL, version 2 or later.
* See the COPYING file in the top-level directory.
*/
#include "qemu/osdep.h"
#include "qemu/error-report.h"
#include "qapi/error.h"
#include "qemu/error-report.h"
#include "hw/vfio/vfio-cpr.h"
#include "migration/cpr.h"
#include "migration/misc.h"
#include "migration/options.h"
#include "migration/qemu-file.h"
#include "migration/savevm.h"
#include "migration/vmstate.h"
#include "monitor/monitor.h"
#include "system/runstate.h"
#include "trace.h"
/*************************************************************************/
/* cpr state container for all information to be saved. */
CprState cpr_state;
static GHashTable *cpr_fds_hash;
/****************************************************************************/
typedef struct CprFd {
char *name;
unsigned int namelen;
int id;
int fd;
QLIST_ENTRY(CprFd) next;
} CprFd;
static const VMStateDescription vmstate_cpr_fd = {
.name = "cpr fd",
.version_id = 1,
.minimum_version_id = 1,
.fields = (VMStateField[]) {
VMSTATE_UINT32(namelen, CprFd),
VMSTATE_VBUFFER_ALLOC_UINT32(name, CprFd, 0, NULL, namelen),
VMSTATE_INT32(id, CprFd),
VMSTATE_FD(fd, CprFd),
VMSTATE_END_OF_LIST()
}
};
static guint cpr_fd_hash(gconstpointer v)
{
const CprFd *elem = v;
return g_str_hash(elem->name) ^ elem->id;
}
static gboolean cpr_fd_equal(gconstpointer a, gconstpointer b)
{
const CprFd *elem_a = a;
const CprFd *elem_b = b;
return !strcmp(elem_a->name, elem_b->name) && elem_a->id == elem_b->id;
}
static void cpr_fd_destroy(gpointer data)
{
CprFd *elem = data;
g_free(elem->name);
g_free(elem);
}
static GHashTable *get_cpr_fds_hash(void)
{
if (!cpr_fds_hash) {
cpr_fds_hash = g_hash_table_new_full(cpr_fd_hash, cpr_fd_equal,
cpr_fd_destroy, NULL);
}
return cpr_fds_hash;
}
static void cpr_fd_hash_insert(CprFd *elem)
{
/* Use the same CprFd as key and value. */
g_hash_table_insert(get_cpr_fds_hash(), elem, elem);
}
static int cpr_fd_pre_save(void *opaque)
{
CprState *state = (CprState *)opaque;
GHashTableIter iter;
CprFd *elem;
QLIST_INIT(&state->fds);
g_hash_table_iter_init(&iter, get_cpr_fds_hash());
while (g_hash_table_iter_next(&iter, (gpointer *)&elem, NULL)) {
QLIST_INSERT_HEAD(&state->fds, elem, next);
}
return 0;
}
static int cpr_fd_post_load(void *opaque, int version_id)
{
CprState *state = (CprState *)opaque;
CprFd *elem;
while ((elem = QLIST_FIRST(&state->fds))) {
QLIST_REMOVE(elem, next);
/*
* Preserve legacy QLIST lookup semantics if duplicate keys exist in
* the incoming stream: the first matching entry wins.
*/
if (g_hash_table_contains(get_cpr_fds_hash(), elem)) {
cpr_fd_destroy(elem);
continue;
}
cpr_fd_hash_insert(elem);
}
return 0;
}
void cpr_save_fd(const char *name, int id, int fd)
{
CprFd *elem = g_new0(CprFd, 1);
trace_cpr_save_fd(name, id, fd);
elem->name = g_strdup(name);
elem->namelen = strlen(name) + 1;
elem->id = id;
elem->fd = fd;
cpr_fd_hash_insert(elem);
}
static CprFd *find_fd(const char *name, int id)
{
CprFd key = {
.name = (char *)name,
.id = id,
};
return g_hash_table_lookup(get_cpr_fds_hash(), &key);
}
void cpr_delete_fd(const char *name, int id)
{
CprFd key = {
.name = (char *)name,
.id = id,
};
g_hash_table_remove(get_cpr_fds_hash(), &key);
trace_cpr_delete_fd(name, id);
}
int cpr_find_fd(const char *name, int id)
{
CprFd *elem = find_fd(name, id);
int fd = elem ? elem->fd : -1;
trace_cpr_find_fd(name, id, fd);
return fd;
}
void cpr_resave_fd(const char *name, int id, int fd)
{
CprFd *elem = find_fd(name, id);
int old_fd = elem ? elem->fd : -1;
if (old_fd < 0) {
cpr_save_fd(name, id, fd);
} else if (old_fd != fd) {
error_report("internal error: cpr fd '%s' id %d value %d "
"already saved with a different value %d",
name, id, fd, old_fd);
g_assert_not_reached();
}
}
int cpr_open_fd(const char *path, int flags, const char *name, int id,
Error **errp)
{
int fd = cpr_find_fd(name, id);
if (fd < 0) {
fd = qemu_open(path, flags, errp);
if (fd >= 0) {
cpr_save_fd(name, id, fd);
}
}
return fd;
}
bool cpr_walk_fd(cpr_walk_fd_cb cb)
{
GHashTableIter iter;
CprFd *elem;
g_hash_table_iter_init(&iter, get_cpr_fds_hash());
while (g_hash_table_iter_next(&iter, (gpointer *)&elem, NULL)) {
g_assert(elem->fd >= 0);
if (!cb(elem->fd)) {
return false;
}
}
return true;
}
/*************************************************************************/
static const VMStateDescription vmstate_cpr_state = {
.name = CPR_STATE,
.version_id = 1,
.minimum_version_id = 1,
.pre_save = cpr_fd_pre_save,
.post_load = cpr_fd_post_load,
.fields = (VMStateField[]) {
VMSTATE_QLIST_V(fds, CprState, 1, vmstate_cpr_fd, CprFd, next),
VMSTATE_END_OF_LIST()
},
.subsections = (const VMStateDescription * const []) {
&vmstate_cpr_vfio_devices,
NULL
}
};
/*************************************************************************/
static QEMUFile *cpr_state_file;
QIOChannel *cpr_state_ioc(void)
{
return qemu_file_get_ioc(cpr_state_file);
}
static MigMode incoming_mode = MIG_MODE_NONE;
MigMode cpr_get_incoming_mode(void)
{
return incoming_mode;
}
void cpr_set_incoming_mode(MigMode mode)
{
incoming_mode = mode;
}
bool cpr_is_incoming(void)
{
return incoming_mode != MIG_MODE_NONE;
}
bool cpr_state_save(MigrationChannel *channel, Error **errp)
{
QEMUFile *f;
MigMode mode = migrate_mode();
trace_cpr_state_save(MigMode_str(mode));
if (mode == MIG_MODE_CPR_TRANSFER) {
g_assert(channel);
f = cpr_transfer_output(channel, errp);
} else if (mode == MIG_MODE_CPR_EXEC) {
f = cpr_exec_output(errp);
} else {
return true;
}
if (!f) {
return false;
}
qemu_put_be32(f, QEMU_CPR_FILE_MAGIC);
qemu_put_be32(f, QEMU_CPR_FILE_VERSION);
if (!vmstate_save_vmsd(f, &vmstate_cpr_state, &cpr_state, 0, errp)) {
qemu_fclose(f);
return false;
}
if (migrate_mode() == MIG_MODE_CPR_EXEC) {
if (!cpr_exec_persist_state(f, errp)) {
qemu_fclose(f);
return false;
}
}
/*
* Close the socket only partially so we can later detect when the other
* end closes by getting a HUP event.
*/
qemu_fflush(f);
qio_channel_shutdown(qemu_file_get_ioc(f), QIO_CHANNEL_SHUTDOWN_WRITE,
NULL);
cpr_state_file = f;
return true;
}
bool cpr_state_load(MigrationChannel *channel, Error **errp)
{
uint32_t v;
QEMUFile *f;
MigMode mode = 0;
if (cpr_exec_has_state()) {
mode = MIG_MODE_CPR_EXEC;
f = cpr_exec_input(errp);
if (channel) {
warn_report("ignoring cpr channel for migration mode cpr-exec");
}
} else if (channel) {
mode = MIG_MODE_CPR_TRANSFER;
cpr_set_incoming_mode(mode);
f = cpr_transfer_input(channel, errp);
} else {
return true;
}
if (!f) {
return false;
}
trace_cpr_state_load(MigMode_str(mode));
cpr_set_incoming_mode(mode);
v = qemu_get_be32(f);
if (v != QEMU_CPR_FILE_MAGIC) {
error_setg(errp, "Not a migration stream (bad magic %x)", v);
qemu_fclose(f);
return false;
}
v = qemu_get_be32(f);
if (v != QEMU_CPR_FILE_VERSION) {
error_setg(errp, "Unsupported migration stream version %d", v);
qemu_fclose(f);
return false;
}
if (!vmstate_load_vmsd(f, &vmstate_cpr_state, &cpr_state, 1, errp)) {
qemu_fclose(f);
return false;
}
if (migrate_mode() == MIG_MODE_CPR_EXEC) {
/* Set cloexec to prevent fd leaks from fork until the next cpr-exec */
cpr_exec_unpreserve_fds();
}
/*
* Let the caller decide when to close the socket (and generate a HUP event
* for the sending side).
*/
cpr_state_file = f;
return true;
}
void cpr_state_close(void)
{
if (cpr_state_file) {
qemu_fclose(cpr_state_file);
cpr_state_file = NULL;
}
}
bool cpr_incoming_needed(void *opaque)
{
MigMode mode = migrate_mode();
return mode == MIG_MODE_CPR_TRANSFER || mode == MIG_MODE_CPR_EXEC;
}
/*
* cpr_get_fd_param: find a descriptor and return its value.
*
* @name: CPR name for the descriptor
* @fdname: An integer-valued string, or a name passed to a getfd command
* @index: CPR index of the descriptor
* @errp: returned error message
*
* If CPR is not being performed, then use @fdname to find the fd.
* If CPR is being performed, then ignore @fdname, and look for @name
* and @index in CPR state.
*
* On success returns the fd value, else returns -1.
*/
int cpr_get_fd_param(const char *name, const char *fdname, int index,
Error **errp)
{
ERRP_GUARD();
int fd;
if (cpr_is_incoming()) {
fd = cpr_find_fd(name, index);
if (fd < 0) {
error_setg(errp, "cannot find saved value for fd %s", fdname);
}
} else {
fd = monitor_fd_param(monitor_cur(), fdname, errp);
if (fd >= 0) {
cpr_save_fd(name, index, fd);
} else {
error_prepend(errp, "Could not parse object fd %s:", fdname);
}
}
return fd;
}