Re: [PATCH 2/2] Call BlkifRing(Acquire/Release)Lock before/after BlkifRingPoll

[Paul Durrant <xadimgnik@xxxxxxxxx>]

On 21/06/2021 13:54, Owen Smith wrote:
> Releasing the blkif ring lock drains the atomic SRB list into the queue. When
> attempting to move things along, this also includes the atomic SRB list.
>
> Signed-off-by: Owen Smith <owen.smith@xxxxxxxxxx>
Acked-by: Paul Durrant <paul@xxxxxxx>

> ---
>   src/xenvbd/ring.c | 4 +++-
>   1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/src/xenvbd/ring.c b/src/xenvbd/ring.c
> index 66e8cdd..1ea1250 100644
> --- a/src/xenvbd/ring.c
> +++ b/src/xenvbd/ring.c
> @@ -2158,7 +2158,6 @@ BlkifRingDisable(
>   
>       Trace("====> %u\n", BlkifRing->Index);
>   
> -    __BlkifRingAcquireLock(BlkifRing);
>       ASSERT(BlkifRing->Enabled);
>   
>       status = XENBUS_STORE(Read,
> @@ -2188,8 +2187,10 @@ BlkifRingDisable(
>                   break;
>   
>               // Try to move things along
> +            __BlkifRingAcquireLock(BlkifRing);
>               __BlkifRingSend(BlkifRing);
>               (VOID) BlkifRingPoll(BlkifRing);
> +            __BlkifRingReleaseLock(BlkifRing);
>   
>               // We are waiting for a watch event at DISPATCH_LEVEL so
>               // it is our responsibility to poll the store ring.
> @@ -2200,6 +2201,7 @@ BlkifRingDisable(
>           }
>       }
>   
> +    __BlkifRingAcquireLock(BlkifRing);
>       BlkifRing->Enabled = FALSE;
>   
>       while (!IsListEmpty(&BlkifRing->SubmittedList)) {