Re: [Qemu-devel] [PATCH] migration: Fix block failure cases

[Peter Xu]

On Wed, Feb 28, 2018 at 04:49:37PM +0000, Dr. David Alan Gilbert (git) wrote:
> From: "Dr. David Alan Gilbert" <address@hidden>
> 
> This fixes a couple of cases where the block migration capability
> doesn't get cleared when a migration failed.
> 
> 1) When block migration is compiled out:
>   (qemu) migrate -d -b  "exec:cat > /dev/null"
>   QEMU compiled without old-style (blk/-b, inc/-i) block migration
>   Use drive_mirror+NBD instead.
>   (qemu) migrate_set_capability xbzrle off
>   QEMU compiled without old-style (blk/-b, inc/-i) block migration
>   Use drive_mirror+NBD instead.
> 
>   This corresponds to https://bugzilla.redhat.com/show_bug.cgi?id=1550022
> 
> 2) When a migration with a bad protocol is tried:
>   (qemu) migrate -d -b "foo:bah"
>   Parameter 'uri' expects a valid migration protocol
>   (qemu) info migrate_capabilities
>   xbzrle: off
>   rdma-pin-all: off
>   auto-converge: off
>   zero-blocks: off
>   compress: off
>   events: off
>   postcopy-ram: off
>   x-colo: off
>   release-ram: off
>   block: on   <<<<<<-----
>   return-path: off
>   pause-before-switchover: off
>   x-multifd: off
> 
> Fixes: 2833c59b947
> Signed-off-by: Dr. David Alan Gilbert <address@hidden>
> ---
>  migration/migration.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/migration/migration.c b/migration/migration.c
> index b913b98803..da0e4a1f56 100644
> --- a/migration/migration.c
> +++ b/migration/migration.c
> @@ -1391,11 +1391,12 @@ void qmp_migrate(const char *uri, bool has_blk, bool 
> blk,
>              return;
>          }
>          migrate_set_block_enabled(true, &local_err);
> +        s->must_remove_block_options = true;
>          if (local_err) {
>              error_propagate(errp, local_err);
> +            block_cleanup_parameters(s);
>              return;
>          }
> -        s->must_remove_block_options = true;
>      }
>  
>      if (has_inc && inc) {
> @@ -1417,11 +1418,10 @@ void qmp_migrate(const char *uri, bool has_blk, bool 
> blk,
>      } else if (strstart(uri, "fd:", &p)) {
>          fd_start_outgoing_migration(s, p, &local_err);
>      } else {
> -        error_setg(errp, QERR_INVALID_PARAMETER_VALUE, "uri",
> +        error_setg(&local_err, QERR_INVALID_PARAMETER_VALUE, "uri",
>                     "a valid migration protocol");
>          migrate_set_state(&s->state, MIGRATION_STATUS_SETUP,
>                            MIGRATION_STATUS_FAILED);
> -        return;
>      }
>  
>      if (local_err) {
> -- 
> 2.14.3
> 

Ouch...

Instead it seems to be my fault in 4a84214ebe ("migration: provide
migrate_caps_check()", 2017-07-18).  For now I cannot understand why I
did that before since it's obviously strange if without this
squashed...

diff --git a/migration/migration.c b/migration/migration.c
index 0aa596f867..88ed9375aa 100644
--- a/migration/migration.c
+++ b/migration/migration.c
@@ -747,13 +747,15 @@ void 
qmp_migrate_set_capabilities(MigrationCapabilityStatusList *params,
 {
     MigrationState *s = migrate_get_current();
     MigrationCapabilityStatusList *cap;
+    bool cap_list[MIGRATION_CAPABILITY__MAX];
 
     if (migration_is_setup_or_active(s->state)) {
         error_setg(errp, QERR_MIGRATION_ACTIVE);
         return;
     }
 
-    if (!migrate_caps_check(s->enabled_capabilities, params, errp)) {
+    memcpy(cap_list, s->enabled_capabilities, sizeof(cap_list));
+    if (!migrate_caps_check(cap_list, params, errp)) {
         return;
     }

Otherwise I'll get:

(qemu) migrate_set_capability postcopy-ram on                                  
(qemu) migrate_set_capability compress on                                      
Postcopy is not currently compatible with compression                          
(qemu) info migrate_capabilities       
xbzrle: off                            
rdma-pin-all: off                      
auto-converge: off                     
zero-blocks: off                       
compress: on         <------------------------- :(
events: off                            
postcopy-ram: on                       
x-colo: off                            
release-ram: off                       
block: off                             
return-path: off                       
pause-before-switchover: off           
x-multifd: off
 
And it looks very likely that this should solve the block bug too.

(So I think either I got a brain fart last July, or now...)

Thanks,

-- 
Peter Xu