6e9beed308
block-nbd was refacted during release 6.2.0, but we didn't induced all the needed patches within the 6.2.0 baseline, which leads to vm crash during migration. the reasons are as below: when iothread is configured, the coroutines should get back to the exact iothread that was out of. But within the 6.2.0 baseline, patches were missing, nbd related coroutine didn't have its related aio_context. It in fact get to the mainline aio_context, the mistaken context leads to vm crash.
98 lines
3.3 KiB
Diff
98 lines
3.3 KiB
Diff
From ba810e3b3800f0d084a2dd324a9dea89c9320548 Mon Sep 17 00:00:00 2001
|
|
From: Zhang Bo <oscar.zhangbo@huawei.com>
|
|
Date: Mon, 29 Aug 2022 15:38:19 +0800
|
|
Subject: [PATCH 5/5] block/nbd: Move s->ioc on AioContext change
|
|
|
|
s->ioc must always be attached to the NBD node's AioContext. If that
|
|
context changes, s->ioc must be attached to the new context.
|
|
|
|
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2033626
|
|
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
|
|
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
|
|
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
|
|
Signed-off-by: Zhang Bo <oscar.zhangbo@huawei.com>
|
|
---
|
|
block/nbd.c | 45 +++++++++++++++++++++++++++++++++++++++++++++
|
|
1 file changed, 45 insertions(+)
|
|
|
|
diff --git a/block/nbd.c b/block/nbd.c
|
|
index dc6c3f3bbc..5853d85d60 100644
|
|
--- a/block/nbd.c
|
|
+++ b/block/nbd.c
|
|
@@ -2055,6 +2055,42 @@ static void nbd_cancel_in_flight(BlockDriverState *bs)
|
|
nbd_co_establish_connection_cancel(s->conn);
|
|
}
|
|
|
|
+static void nbd_attach_aio_context(BlockDriverState *bs,
|
|
+ AioContext *new_context)
|
|
+{
|
|
+ BDRVNBDState *s = bs->opaque;
|
|
+
|
|
+ /* The open_timer is used only during nbd_open() */
|
|
+ assert(!s->open_timer);
|
|
+
|
|
+ /*
|
|
+ * The reconnect_delay_timer is scheduled in I/O paths when the
|
|
+ * connection is lost, to cancel the reconnection attempt after a
|
|
+ * given time. Once this attempt is done (successfully or not),
|
|
+ * nbd_reconnect_attempt() ensures the timer is deleted before the
|
|
+ * respective I/O request is resumed.
|
|
+ * Since the AioContext can only be changed when a node is drained,
|
|
+ * the reconnect_delay_timer cannot be active here.
|
|
+ */
|
|
+ assert(!s->reconnect_delay_timer);
|
|
+
|
|
+ if (s->ioc) {
|
|
+ qio_channel_attach_aio_context(s->ioc, new_context);
|
|
+ }
|
|
+}
|
|
+
|
|
+static void nbd_detach_aio_context(BlockDriverState *bs)
|
|
+{
|
|
+ BDRVNBDState *s = bs->opaque;
|
|
+
|
|
+ assert(!s->open_timer);
|
|
+ assert(!s->reconnect_delay_timer);
|
|
+
|
|
+ if (s->ioc) {
|
|
+ qio_channel_detach_aio_context(s->ioc);
|
|
+ }
|
|
+}
|
|
+
|
|
static BlockDriver bdrv_nbd = {
|
|
.format_name = "nbd",
|
|
.protocol_name = "nbd",
|
|
@@ -2078,6 +2114,9 @@ static BlockDriver bdrv_nbd = {
|
|
.bdrv_dirname = nbd_dirname,
|
|
.strong_runtime_opts = nbd_strong_runtime_opts,
|
|
.bdrv_cancel_in_flight = nbd_cancel_in_flight,
|
|
+
|
|
+ .bdrv_attach_aio_context = nbd_attach_aio_context,
|
|
+ .bdrv_detach_aio_context = nbd_detach_aio_context,
|
|
};
|
|
|
|
static BlockDriver bdrv_nbd_tcp = {
|
|
@@ -2103,6 +2142,9 @@ static BlockDriver bdrv_nbd_tcp = {
|
|
.bdrv_dirname = nbd_dirname,
|
|
.strong_runtime_opts = nbd_strong_runtime_opts,
|
|
.bdrv_cancel_in_flight = nbd_cancel_in_flight,
|
|
+
|
|
+ .bdrv_attach_aio_context = nbd_attach_aio_context,
|
|
+ .bdrv_detach_aio_context = nbd_detach_aio_context,
|
|
};
|
|
|
|
static BlockDriver bdrv_nbd_unix = {
|
|
@@ -2128,6 +2170,9 @@ static BlockDriver bdrv_nbd_unix = {
|
|
.bdrv_dirname = nbd_dirname,
|
|
.strong_runtime_opts = nbd_strong_runtime_opts,
|
|
.bdrv_cancel_in_flight = nbd_cancel_in_flight,
|
|
+
|
|
+ .bdrv_attach_aio_context = nbd_attach_aio_context,
|
|
+ .bdrv_detach_aio_context = nbd_detach_aio_context,
|
|
};
|
|
|
|
static void bdrv_nbd_init(void)
|
|
--
|
|
2.27.0
|
|
|