]> Git Repo - qemu.git/commit - migration/migration.c
migration: Unify block node activation error handling
authorKevin Wolf <[email protected]>
Thu, 4 May 2017 16:52:36 +0000 (18:52 +0200)
committerKevin Wolf <[email protected]>
Thu, 11 May 2017 10:08:24 +0000 (12:08 +0200)
commitace21a58751824f9a3d399e332317233e880de3a
treed4486c7e1932c50d5390939f446510a89750829a
parentaa93c834f9c5b971ad3b54944a5dae97ca310225
migration: Unify block node activation error handling

Migration code activates all block driver nodes on the destination when
the migration completes. It does so by calling
bdrv_invalidate_cache_all() and blk_resume_after_migration(). There is
one code path for precopy and one for postcopy migration, resulting in
four function calls, which used to have three different failure modes.

This patch unifies the behaviour so that failure to activate all block
nodes is non-fatal, but the error message is logged and the VM isn't
automatically started. 'cont' will retry activating the block nodes.

Signed-off-by: Kevin Wolf <[email protected]>
Reviewed-by: Eric Blake <[email protected]>
migration/migration.c
migration/savevm.c
qmp.c
This page took 0.024582 seconds and 4 git commands to generate.