From: peterx@redhat.com
To: Peter Maydell <peter.maydell@linaro.org>, qemu-devel@nongnu.org
Cc: peterx@redhat.com, Fabiano Rosas <farosas@suse.de>
Subject: [PULL 30/34] migration/multifd: Unify multifd and TLS connection paths
Date: Thu, 8 Feb 2024 11:05:24 +0800 [thread overview]
Message-ID: <20240208030528.368214-31-peterx@redhat.com> (raw)
In-Reply-To: <20240208030528.368214-1-peterx@redhat.com>
From: Fabiano Rosas <farosas@suse.de>
During multifd channel creation (multifd_send_new_channel_async) when
TLS is enabled, the multifd_channel_connect function is called twice,
once to create the TLS handshake thread and another time after the
asynchrounous TLS handshake has finished.
This creates a slightly confusing call stack where
multifd_channel_connect() is called more times than the number of
channels. It also splits error handling between the two callers of
multifd_channel_connect() causing some code duplication. Lastly, it
gets in the way of having a single point to determine whether all
channel creation tasks have been initiated.
Refactor the code to move the reentrancy one level up at the
multifd_new_send_channel_async() level, de-duplicating the error
handling and allowing for the next patch to introduce a
synchronization point common to all the multifd channel creation,
regardless of TLS.
Note that the previous code would never fail once p->c had been set.
This patch changes this assumption, which affects refcounting, so add
comments around object_unref to explain the situation.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240206215118.6171-6-farosas@suse.de
Signed-off-by: Peter Xu <peterx@redhat.com>
---
migration/multifd.c | 83 ++++++++++++++++++++++-----------------------
1 file changed, 40 insertions(+), 43 deletions(-)
diff --git a/migration/multifd.c b/migration/multifd.c
index cf865edba0..3db18dc79e 100644
--- a/migration/multifd.c
+++ b/migration/multifd.c
@@ -869,30 +869,7 @@ out:
return NULL;
}
-static bool multifd_channel_connect(MultiFDSendParams *p,
- QIOChannel *ioc,
- Error **errp);
-
-static void multifd_tls_outgoing_handshake(QIOTask *task,
- gpointer opaque)
-{
- MultiFDSendParams *p = opaque;
- QIOChannel *ioc = QIO_CHANNEL(qio_task_get_source(task));
- Error *err = NULL;
-
- if (!qio_task_propagate_error(task, &err)) {
- trace_multifd_tls_outgoing_handshake_complete(ioc);
- if (multifd_channel_connect(p, ioc, &err)) {
- return;
- }
- }
-
- trace_multifd_tls_outgoing_handshake_error(ioc, error_get_pretty(err));
-
- multifd_send_set_error(err);
- multifd_send_kick_main(p);
- error_free(err);
-}
+static void multifd_new_send_channel_async(QIOTask *task, gpointer opaque);
static void *multifd_tls_handshake_thread(void *opaque)
{
@@ -900,7 +877,7 @@ static void *multifd_tls_handshake_thread(void *opaque)
QIOChannelTLS *tioc = QIO_CHANNEL_TLS(p->c);
qio_channel_tls_handshake(tioc,
- multifd_tls_outgoing_handshake,
+ multifd_new_send_channel_async,
p,
NULL,
NULL);
@@ -920,6 +897,10 @@ static bool multifd_tls_channel_connect(MultiFDSendParams *p,
return false;
}
+ /*
+ * Ownership of the socket channel now transfers to the newly
+ * created TLS channel, which has already taken a reference.
+ */
object_unref(OBJECT(ioc));
trace_multifd_tls_outgoing_handshake_start(ioc, tioc, hostname);
qio_channel_set_name(QIO_CHANNEL(tioc), "multifd-tls-outgoing");
@@ -936,18 +917,7 @@ static bool multifd_channel_connect(MultiFDSendParams *p,
QIOChannel *ioc,
Error **errp)
{
- trace_multifd_set_outgoing_channel(
- ioc, object_get_typename(OBJECT(ioc)),
- migrate_get_current()->hostname);
-
- if (migrate_channel_requires_tls_upgrade(ioc)) {
- /*
- * tls_channel_connect will call back to this
- * function after the TLS handshake,
- * so we mustn't call multifd_send_thread until then
- */
- return multifd_tls_channel_connect(p, ioc, errp);
- }
+ qio_channel_set_delay(ioc, false);
migration_ioc_register_yank(ioc);
p->registered_yank = true;
@@ -959,24 +929,51 @@ static bool multifd_channel_connect(MultiFDSendParams *p,
return true;
}
+/*
+ * When TLS is enabled this function is called once to establish the
+ * TLS connection and a second time after the TLS handshake to create
+ * the multifd channel. Without TLS it goes straight into the channel
+ * creation.
+ */
static void multifd_new_send_channel_async(QIOTask *task, gpointer opaque)
{
MultiFDSendParams *p = opaque;
QIOChannel *ioc = QIO_CHANNEL(qio_task_get_source(task));
Error *local_err = NULL;
+ bool ret;
trace_multifd_new_send_channel_async(p->id);
- if (!qio_task_propagate_error(task, &local_err)) {
- qio_channel_set_delay(ioc, false);
- if (multifd_channel_connect(p, ioc, &local_err)) {
- return;
- }
+
+ if (qio_task_propagate_error(task, &local_err)) {
+ ret = false;
+ goto out;
+ }
+
+ trace_multifd_set_outgoing_channel(ioc, object_get_typename(OBJECT(ioc)),
+ migrate_get_current()->hostname);
+
+ if (migrate_channel_requires_tls_upgrade(ioc)) {
+ ret = multifd_tls_channel_connect(p, ioc, &local_err);
+ } else {
+ ret = multifd_channel_connect(p, ioc, &local_err);
}
+ if (ret) {
+ return;
+ }
+
+out:
trace_multifd_new_send_channel_async_error(p->id, local_err);
multifd_send_set_error(local_err);
multifd_send_kick_main(p);
- object_unref(OBJECT(ioc));
+ if (!p->c) {
+ /*
+ * If no channel has been created, drop the initial
+ * reference. Otherwise cleanup happens at
+ * multifd_send_channel_destroy()
+ */
+ object_unref(OBJECT(ioc));
+ }
error_free(local_err);
}
--
2.43.0
next prev parent reply other threads:[~2024-02-08 3:08 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-02-08 3:04 [PULL 00/34] Migration staging patches peterx
2024-02-08 3:04 ` [PULL 01/34] migration: prevent migration when VM has poisoned memory peterx
2024-02-08 3:04 ` [PULL 02/34] migration/multifd: Drop stale comment for multifd zero copy peterx
2024-02-08 3:04 ` [PULL 03/34] migration/multifd: multifd_send_kick_main() peterx
2024-02-08 3:04 ` [PULL 04/34] migration/multifd: Drop MultiFDSendParams.quit, cleanup error paths peterx
2024-02-08 3:04 ` [PULL 05/34] migration/multifd: Postpone reset of MultiFDPages_t peterx
2024-02-08 3:05 ` [PULL 06/34] migration/multifd: Drop MultiFDSendParams.normal[] array peterx
2024-02-08 3:05 ` [PULL 07/34] migration/multifd: Separate SYNC request with normal jobs peterx
2024-02-08 3:05 ` [PULL 08/34] migration/multifd: Simplify locking in sender thread peterx
2024-02-08 3:05 ` [PULL 09/34] migration/multifd: Drop pages->num check " peterx
2024-02-08 3:05 ` [PULL 10/34] migration/multifd: Rename p->num_packets and clean it up peterx
2024-02-08 3:05 ` [PULL 11/34] migration/multifd: Move total_normal_pages accounting peterx
2024-02-08 3:05 ` [PULL 12/34] migration/multifd: Move trace_multifd_send|recv() peterx
2024-02-08 3:05 ` [PULL 13/34] migration/multifd: multifd_send_prepare_header() peterx
2024-02-08 3:05 ` [PULL 14/34] migration/multifd: Move header prepare/fill into send_prepare() peterx
2024-02-08 3:05 ` [PULL 15/34] migration/multifd: Forbid spurious wakeups peterx
2024-02-08 3:05 ` [PULL 16/34] migration/multifd: Split multifd_send_terminate_threads() peterx
2024-02-08 3:05 ` [PULL 17/34] migration/multifd: Change retval of multifd_queue_page() peterx
2024-02-08 3:05 ` [PULL 18/34] migration/multifd: Change retval of multifd_send_pages() peterx
2024-02-08 3:05 ` [PULL 19/34] migration/multifd: Rewrite multifd_queue_page() peterx
2024-02-08 3:05 ` [PULL 20/34] migration/multifd: Cleanup multifd_save_cleanup() peterx
2024-02-08 3:05 ` [PULL 21/34] migration/multifd: Cleanup multifd_load_cleanup() peterx
2024-02-08 3:05 ` [PULL 22/34] migration/multifd: Stick with send/recv on function names peterx
2024-02-08 3:05 ` [PULL 23/34] migration/multifd: Fix MultiFDSendParams.packet_num race peterx
2024-02-08 3:05 ` [PULL 24/34] migration/multifd: Optimize sender side to be lockless peterx
2024-02-08 3:05 ` [PULL 25/34] migration: Fix logic of channels and transport compatibility check peterx
2024-02-08 3:05 ` [PULL 26/34] migration/multifd: Join the TLS thread peterx
2024-02-10 9:18 ` Michael Tokarev
2024-02-10 9:30 ` Michael Tokarev
2024-02-14 13:27 ` Fabiano Rosas
2024-02-14 13:58 ` Michael Tokarev
2024-02-15 13:24 ` Fabiano Rosas
2024-02-08 3:05 ` [PULL 27/34] migration/multifd: Remove p->running peterx
2024-02-08 3:05 ` [PULL 28/34] migration/multifd: Move multifd_send_setup error handling in to the function peterx
2024-02-08 3:05 ` [PULL 29/34] migration/multifd: Move multifd_send_setup into migration thread peterx
2024-02-08 3:05 ` peterx [this message]
2024-02-08 3:05 ` [PULL 31/34] migration/multifd: Add a synchronization point for channel creation peterx
2024-02-08 3:05 ` [PULL 32/34] tests/migration-test: Stick with gicv3 in aarch64 test peterx
2024-02-08 3:05 ` [PULL 33/34] ci: Remove tag dependency for build-previous-qemu peterx
2024-02-08 3:05 ` [PULL 34/34] ci: Update comment for migration-compat-aarch64 peterx
2024-02-09 16:14 ` [PULL 00/34] Migration staging patches Peter Maydell
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20240208030528.368214-31-peterx@redhat.com \
--to=peterx@redhat.com \
--cc=farosas@suse.de \
--cc=peter.maydell@linaro.org \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).