rpms
/
qemu-kvm
7
0
Fork 0
Code Issues Pull Requests Releases Activity
qemu-kvm/SOURCES/kvm-migration-check-magic-v...

297 lines
11 KiB
Diff
Raw Blame History

From f21a343af4b4d0c6e5181ae0abd0f6280dc8296c Mon Sep 17 00:00:00 2001
From: "manish.mishra" <manish.mishra@nutanix.com>
Date: Tue, 20 Dec 2022 18:44:18 +0000
Subject: [PATCH 2/3] migration: check magic value for deciding the mapping of
channels
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
RH-Author: Peter Xu <peterx@redhat.com>
RH-MergeRequest: 258: migration: Fix multifd crash due to channel disorder
RH-Bugzilla: 2137740
RH-Acked-by: quintela1 <quintela@redhat.com>
RH-Acked-by: Leonardo Brás <leobras@redhat.com>
RH-Acked-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
RH-Commit: [2/2] f97bebef3d3e372cfd660e5ddb6cffba791840d2
Conflicts:
migration/migration.c
migration/multifd.c
migration/postcopy-ram.c
migration/postcopy-ram.h
There're a bunch of conflicts due to missing upstream patches on
e.g. on qemufile reworks, postcopy preempt. We don't plan to have
preempt in rhel8 at all, probably the same as the rest.
Current logic assumes that channel connections on the destination side are
always established in the same order as the source and the first one will
always be the main channel followed by the multifid or post-copy
preemption channel. This may not be always true, as even if a channel has a
connection established on the source side it can be in the pending state on
the destination side and a newer connection can be established first.
Basically causing out of order mapping of channels on the destination side.
Currently, all channels except post-copy preempt send a magic number, this
patch uses that magic number to decide the type of channel. This logic is
applicable only for precopy(multifd) live migration, as mentioned, the
post-copy preempt channel does not send any magic number. Also, tls live
migrations already does tls handshake before creating other channels, so
this issue is not possible with tls, hence this logic is avoided for tls
live migrations. This patch uses read peek to check the magic number of
channels so that current data/control stream management remains
un-effected.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Suggested-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: manish.mishra <manish.mishra@nutanix.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
(cherry picked from commit 6720c2b32725e6ac404f22851a0ecd0a71d0cbe2)
Signed-off-by: Peter Xu <peterx@redhat.com>
---
migration/channel.c | 45 ++++++++++++++++++++++++++++++++++++++
migration/channel.h | 5 +++++
migration/migration.c | 51 +++++++++++++++++++++++++++++++------------
migration/multifd.c | 19 ++++++++--------
migration/multifd.h | 2 +-
5 files changed, 98 insertions(+), 24 deletions(-)
diff --git a/migration/channel.c b/migration/channel.c
index 086b5c0d8b..ee308fef23 100644
--- a/migration/channel.c
+++ b/migration/channel.c
@@ -98,3 +98,48 @@ void migration_channel_connect(MigrationState *s,
g_free(s->hostname);
error_free(error);
}
+
+
+/**
+ * @migration_channel_read_peek - Peek at migration channel, without
+ * actually removing it from channel buffer.
+ *
+ * @ioc: the channel object
+ * @buf: the memory region to read data into
+ * @buflen: the number of bytes to read in @buf
+ * @errp: pointer to a NULL-initialized error object
+ *
+ * Returns 0 if successful, returns -1 and sets @errp if fails.
+ */
+int migration_channel_read_peek(QIOChannel *ioc,
+ const char *buf,
+ const size_t buflen,
+ Error **errp)
+{
+ ssize_t len = 0;
+ struct iovec iov = { .iov_base = (char *)buf, .iov_len = buflen };
+
+ while (true) {
+ len = qio_channel_readv_full(ioc, &iov, 1, NULL, NULL,
+ QIO_CHANNEL_READ_FLAG_MSG_PEEK, errp);
+
+ if (len <= 0 && len != QIO_CHANNEL_ERR_BLOCK) {
+ error_setg(errp,
+ "Failed to peek at channel");
+ return -1;
+ }
+
+ if (len == buflen) {
+ break;
+ }
+
+ /* 1ms sleep. */
+ if (qemu_in_coroutine()) {
+ qemu_co_sleep_ns(QEMU_CLOCK_REALTIME, 1000000);
+ } else {
+ g_usleep(1000);
+ }
+ }
+
+ return 0;
+}
diff --git a/migration/channel.h b/migration/channel.h
index 67a461c28a..5bdb8208a7 100644
--- a/migration/channel.h
+++ b/migration/channel.h
@@ -24,4 +24,9 @@ void migration_channel_connect(MigrationState *s,
QIOChannel *ioc,
const char *hostname,
Error *error_in);
+
+int migration_channel_read_peek(QIOChannel *ioc,
+ const char *buf,
+ const size_t buflen,
+ Error **errp);
#endif
diff --git a/migration/migration.c b/migration/migration.c
index d8b24a2c91..0885549de0 100644
--- a/migration/migration.c
+++ b/migration/migration.c
@@ -32,6 +32,7 @@
#include "savevm.h"
#include "qemu-file-channel.h"
#include "qemu-file.h"
+#include "channel.h"
#include "migration/vmstate.h"
#include "block/block.h"
#include "qapi/error.h"
@@ -637,10 +638,6 @@ static bool migration_incoming_setup(QEMUFile *f, Error **errp)
{
MigrationIncomingState *mis = migration_incoming_get_current();
- if (multifd_load_setup(errp) != 0) {
- return false;
- }
-
if (!mis->from_src_file) {
mis->from_src_file = f;
}
@@ -701,10 +698,42 @@ void migration_fd_process_incoming(QEMUFile *f, Error **errp)
void migration_ioc_process_incoming(QIOChannel *ioc, Error **errp)
{
MigrationIncomingState *mis = migration_incoming_get_current();
+ bool default_channel = true;
+ uint32_t channel_magic = 0;
Error *local_err = NULL;
- bool start_migration;
+ int ret = 0;
- if (!mis->from_src_file) {
+ if (migrate_use_multifd() && !migrate_postcopy_ram() &&
+ qio_channel_has_feature(ioc, QIO_CHANNEL_FEATURE_READ_MSG_PEEK)) {
+ /*
+ * With multiple channels, it is possible that we receive channels
+ * out of order on destination side, causing incorrect mapping of
+ * source channels on destination side. Check channel MAGIC to
+ * decide type of channel. Please note this is best effort, postcopy
+ * preempt channel does not send any magic number so avoid it for
+ * postcopy live migration. Also tls live migration already does
+ * tls handshake while initializing main channel so with tls this
+ * issue is not possible.
+ */
+ ret = migration_channel_read_peek(ioc, (void *)&channel_magic,
+ sizeof(channel_magic), &local_err);
+
+ if (ret != 0) {
+ error_propagate(errp, local_err);
+ return;
+ }
+
+ default_channel = (channel_magic == cpu_to_be32(QEMU_VM_FILE_MAGIC));
+ } else {
+ default_channel = !mis->from_src_file;
+ }
+
+ if (multifd_load_setup(errp) != 0) {
+ error_setg(errp, "Failed to setup multifd channels");
+ return;
+ }
+
+ if (default_channel) {
/* The first connection (multifd may have multiple) */
QEMUFile *f = qemu_fopen_channel_input(ioc);
@@ -716,23 +745,17 @@ void migration_ioc_process_incoming(QIOChannel *ioc, Error **errp)
if (!migration_incoming_setup(f, errp)) {
return;
}
-
- /*
- * Common migration only needs one channel, so we can start
- * right now. Multifd needs more than one channel, we wait.
- */
- start_migration = !migrate_use_multifd();
} else {
/* Multiple connections */
assert(migrate_use_multifd());
- start_migration = multifd_recv_new_channel(ioc, &local_err);
+ multifd_recv_new_channel(ioc, &local_err);
if (local_err) {
error_propagate(errp, local_err);
return;
}
}
- if (start_migration) {
+ if (migration_has_all_channels()) {
migration_incoming_process();
}
}
diff --git a/migration/multifd.c b/migration/multifd.c
index 7c16523e6b..75ac052d2f 100644
--- a/migration/multifd.c
+++ b/migration/multifd.c
@@ -1183,9 +1183,14 @@ int multifd_load_setup(Error **errp)
uint32_t page_count = MULTIFD_PACKET_SIZE / qemu_target_page_size();
uint8_t i;
- if (!migrate_use_multifd()) {
+ /*
+ * Return successfully if multiFD recv state is already initialised
+ * or multiFD is not enabled.
+ */
+ if (multifd_recv_state || !migrate_use_multifd()) {
return 0;
}
+
if (!migrate_multifd_is_allowed()) {
error_setg(errp, "multifd is not supported by current protocol");
return -1;
@@ -1244,11 +1249,9 @@ bool multifd_recv_all_channels_created(void)
/*
* Try to receive all multifd channels to get ready for the migration.
- * - Return true and do not set @errp when correctly receiving all channels;
- * - Return false and do not set @errp when correctly receiving the current one;
- * - Return false and set @errp when failing to receive the current channel.
+ * Sets @errp when failing to receive the current channel.
*/
-bool multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
+void multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
{
MultiFDRecvParams *p;
Error *local_err = NULL;
@@ -1261,7 +1264,7 @@ bool multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
"failed to receive packet"
" via multifd channel %d: ",
qatomic_read(&multifd_recv_state->count));
- return false;
+ return;
}
trace_multifd_recv_new_channel(id);
@@ -1271,7 +1274,7 @@ bool multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
id);
multifd_recv_terminate_threads(local_err);
error_propagate(errp, local_err);
- return false;
+ return;
}
p->c = ioc;
object_ref(OBJECT(ioc));
@@ -1282,6 +1285,4 @@ bool multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
qemu_thread_create(&p->thread, p->name, multifd_recv_thread, p,
QEMU_THREAD_JOINABLE);
qatomic_inc(&multifd_recv_state->count);
- return qatomic_read(&multifd_recv_state->count) ==
- migrate_multifd_channels();
}
diff --git a/migration/multifd.h b/migration/multifd.h
index 11d5e273e6..9c0a2a0701 100644
--- a/migration/multifd.h
+++ b/migration/multifd.h
@@ -20,7 +20,7 @@ void multifd_save_cleanup(void);
int multifd_load_setup(Error **errp);
int multifd_load_cleanup(Error **errp);
bool multifd_recv_all_channels_created(void);
-bool multifd_recv_new_channel(QIOChannel *ioc, Error **errp);
+void multifd_recv_new_channel(QIOChannel *ioc, Error **errp);
void multifd_recv_sync_main(void);
int multifd_send_sync_main(QEMUFile *f);
int multifd_queue_page(QEMUFile *f, RAMBlock *block, ram_addr_t offset);
--
2.37.3
Reference in New Issue View Git Blame Copy Permalink