Am 12.02.2014 um 20:46 hat Jeff Cody geschrieben:
> When starting a block job, commit_active_start() relies on whether *errp
> is set by mirror_start_job. This allows it to determine if the mirror
> job start failed, so that it can clean up any changes to open flags from
> the bdrv_reopen(). If errp is NULL, then it will not be able to
> determine if mirror_start_job failed or not.
>
> Reported-by: Markus Armbruster <arm...@redhat.com>
> Signed-off-by: Jeff Cody <jc...@redhat.com>
> ---
> block/mirror.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/block/mirror.c b/block/mirror.c
> index 2a43334..41bb83c 100644
> --- a/block/mirror.c
> +++ b/block/mirror.c
> @@ -634,6 +634,8 @@ void commit_active_start(BlockDriverState *bs,
> BlockDriverState *base,
> int64_t length, base_length;
> int orig_base_flags;
>
> + assert(errp != NULL);
> +
> orig_base_flags = bdrv_get_flags(base);
>
> if (bdrv_reopen(base, bs->open_flags, errp)) {
This is surprising behaviour. Without looking at the function
implementation, I expect that errp == NULL works and means the same as
everywhere else: The caller doesn't care about errors.
I wouldn't mind if violators were detected at compile time, but this is
merely a run-time error (and strictly speaking only an error at all
without NDEBUG), so I would prefer to use the normal local_err pattern.
Kevin
