[PATCH v3 02/12] block/io.c: drop assertion on double waiting for request serialisation

Vladimir Sementsov-Ogievskiy posted 12 patches 5 years, 5 months ago
Maintainers: Fam Zheng <fam@euphon.net>, Kevin Wolf <kwolf@redhat.com>, Stefan Hajnoczi <stefanha@redhat.com>, Max Reitz <mreitz@redhat.com>, Markus Armbruster <armbru@redhat.com>, Eric Blake <eblake@redhat.com>
There is a newer version of this series
[PATCH v3 02/12] block/io.c: drop assertion on double waiting for request serialisation
Posted by Vladimir Sementsov-Ogievskiy 5 years, 5 months ago
The comments states, that on misaligned request we should have already
been waiting. But for bdrv_padding_rmw_read, we called
bdrv_mark_request_serialising with align = request_alignment, and now
we serialise with align = cluster_size. So we may have to wait again
with larger alignment.

Note, that the only user of BDRV_REQ_SERIALISING is backup which issues
cluster-aligned requests, so seems the assertion should not fire for
now. But it's wrong anyway.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
---
 block/io.c | 11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)

diff --git a/block/io.c b/block/io.c
index ad3a51ed53..b18680a842 100644
--- a/block/io.c
+++ b/block/io.c
@@ -1881,7 +1881,6 @@ bdrv_co_write_req_prepare(BdrvChild *child, int64_t offset, uint64_t bytes,
                           BdrvTrackedRequest *req, int flags)
 {
     BlockDriverState *bs = child->bs;
-    bool waited;
     int64_t end_sector = DIV_ROUND_UP(offset + bytes, BDRV_SECTOR_SIZE);
 
     if (bs->read_only) {
@@ -1893,15 +1892,7 @@ bdrv_co_write_req_prepare(BdrvChild *child, int64_t offset, uint64_t bytes,
     assert(!(flags & ~BDRV_REQ_MASK));
 
     if (flags & BDRV_REQ_SERIALISING) {
-        waited = bdrv_mark_request_serialising(req, bdrv_get_cluster_size(bs));
-        /*
-         * For a misaligned request we should have already waited earlier,
-         * because we come after bdrv_padding_rmw_read which must be called
-         * with the request already marked as serialising.
-         */
-        assert(!waited ||
-               (req->offset == req->overlap_offset &&
-                req->bytes == req->overlap_bytes));
+        bdrv_mark_request_serialising(req, bdrv_get_cluster_size(bs));
     } else {
         bdrv_wait_serialising_requests(req);
     }
-- 
2.18.0


Re: [PATCH v3 02/12] block/io.c: drop assertion on double waiting for request serialisation
Posted by Stefan Hajnoczi 5 years, 5 months ago
On Mon, Aug 17, 2020 at 12:15:43PM +0300, Vladimir Sementsov-Ogievskiy wrote:
> The comments states, that on misaligned request we should have already
> been waiting. But for bdrv_padding_rmw_read, we called
> bdrv_mark_request_serialising with align = request_alignment, and now
> we serialise with align = cluster_size. So we may have to wait again
> with larger alignment.
> 
> Note, that the only user of BDRV_REQ_SERIALISING is backup which issues
> cluster-aligned requests, so seems the assertion should not fire for
> now. But it's wrong anyway.
> 
> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> ---
>  block/io.c | 11 +----------
>  1 file changed, 1 insertion(+), 10 deletions(-)

This code was added by Paolo, CCing him.

> diff --git a/block/io.c b/block/io.c
> index ad3a51ed53..b18680a842 100644
> --- a/block/io.c
> +++ b/block/io.c
> @@ -1881,7 +1881,6 @@ bdrv_co_write_req_prepare(BdrvChild *child, int64_t offset, uint64_t bytes,
>                            BdrvTrackedRequest *req, int flags)
>  {
>      BlockDriverState *bs = child->bs;
> -    bool waited;
>      int64_t end_sector = DIV_ROUND_UP(offset + bytes, BDRV_SECTOR_SIZE);
>  
>      if (bs->read_only) {
> @@ -1893,15 +1892,7 @@ bdrv_co_write_req_prepare(BdrvChild *child, int64_t offset, uint64_t bytes,
>      assert(!(flags & ~BDRV_REQ_MASK));
>  
>      if (flags & BDRV_REQ_SERIALISING) {
> -        waited = bdrv_mark_request_serialising(req, bdrv_get_cluster_size(bs));
> -        /*
> -         * For a misaligned request we should have already waited earlier,
> -         * because we come after bdrv_padding_rmw_read which must be called
> -         * with the request already marked as serialising.
> -         */
> -        assert(!waited ||
> -               (req->offset == req->overlap_offset &&
> -                req->bytes == req->overlap_bytes));
> +        bdrv_mark_request_serialising(req, bdrv_get_cluster_size(bs));
>      } else {
>          bdrv_wait_serialising_requests(req);
>      }
> -- 
> 2.18.0
> 
Re: [PATCH v3 02/12] block/io.c: drop assertion on double waiting for request serialisation
Posted by Paolo Bonzini 5 years, 5 months ago
On 19/08/20 17:28, Stefan Hajnoczi wrote:
> On Mon, Aug 17, 2020 at 12:15:43PM +0300, Vladimir Sementsov-Ogievskiy wrote:
>> The comments states, that on misaligned request we should have already
>> been waiting. But for bdrv_padding_rmw_read, we called
>> bdrv_mark_request_serialising with align = request_alignment, and now
>> we serialise with align = cluster_size. So we may have to wait again
>> with larger alignment.
>>
>> Note, that the only user of BDRV_REQ_SERIALISING is backup which issues
>> cluster-aligned requests, so seems the assertion should not fire for
>> now. But it's wrong anyway.
>>
>> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
>> ---
>>  block/io.c | 11 +----------
>>  1 file changed, 1 insertion(+), 10 deletions(-)
> 
> This code was added by Paolo, CCing him.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>

bdrv_mark_request_serialising already calls
bdrv_wait_serialising_requests_locked so it is okay.

Thanks,

Paolo

>> diff --git a/block/io.c b/block/io.c
>> index ad3a51ed53..b18680a842 100644
>> --- a/block/io.c
>> +++ b/block/io.c
>> @@ -1881,7 +1881,6 @@ bdrv_co_write_req_prepare(BdrvChild *child, int64_t offset, uint64_t bytes,
>>                            BdrvTrackedRequest *req, int flags)
>>  {
>>      BlockDriverState *bs = child->bs;
>> -    bool waited;
>>      int64_t end_sector = DIV_ROUND_UP(offset + bytes, BDRV_SECTOR_SIZE);
>>  
>>      if (bs->read_only) {
>> @@ -1893,15 +1892,7 @@ bdrv_co_write_req_prepare(BdrvChild *child, int64_t offset, uint64_t bytes,
>>      assert(!(flags & ~BDRV_REQ_MASK));
>>  
>>      if (flags & BDRV_REQ_SERIALISING) {
>> -        waited = bdrv_mark_request_serialising(req, bdrv_get_cluster_size(bs));
>> -        /*
>> -         * For a misaligned request we should have already waited earlier,
>> -         * because we come after bdrv_padding_rmw_read which must be called
>> -         * with the request already marked as serialising.
>> -         */
>> -        assert(!waited ||
>> -               (req->offset == req->overlap_offset &&
>> -                req->bytes == req->overlap_bytes));
>> +        bdrv_mark_request_serialising(req, bdrv_get_cluster_size(bs));
>>      } else {
>>          bdrv_wait_serialising_requests(req);
>>      }
>> -- 
>> 2.18.0
>>