[Qemu-devel] [PATCH] raw-posix: Fix .bdrv_co_get_block_status() for unaligned image size

[Qemu-devel] [PATCH] raw-posix: Fix .bdrv_co_get_block_status() for unaligned image size

[Kevin Wolf]

Image files with an unaligned image size have a final hole that starts
at EOF, i.e. in the middle of a sector. Currently, *pnum == 0 is
returned when checking the status of this sector. In qemu-img, this
triggers an assertion failure.

In order to fix this, one type for the sector that contains EOF must be
found. Treating a hole as data is safe, so this patch rounds the
calculated number of data sectors up, so that a partial sector at EOF is
treated as a full data sector.

This fixes https://bugzilla.redhat.com/show_bug.cgi?id=1229394

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
---
 block/raw-posix.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/block/raw-posix.c b/block/raw-posix.c
index 2990e95..44ade8c 100644
--- a/block/raw-posix.c
+++ b/block/raw-posix.c
@@ -1848,8 +1848,9 @@ static int64_t coroutine_fn raw_co_get_block_status(BlockDriverState *bs,
         *pnum = nb_sectors;
         ret = BDRV_BLOCK_DATA;
     } else if (data == start) {
-        /* On a data extent, compute sectors to the end of the extent.  */
-        *pnum = MIN(nb_sectors, (hole - start) / BDRV_SECTOR_SIZE);
+        /* On a data extent, compute sectors to the end of the extent,
+         * possibly including a partial sector at EOF. */
+        *pnum = MIN(nb_sectors, DIV_ROUND_UP(hole - start, BDRV_SECTOR_SIZE));
         ret = BDRV_BLOCK_DATA;
     } else {
         /* On a hole, compute sectors to the beginning of the next extent.  */
-- 
1.8.3.1

Re: [Qemu-devel] [PATCH] raw-posix: Fix .bdrv_co_get_block_status() for unaligned image size

[Wen Congyang]

On 06/09/2015 04:55 PM, Kevin Wolf wrote:
> Image files with an unaligned image size have a final hole that starts
> at EOF, i.e. in the middle of a sector. Currently, *pnum == 0 is
> returned when checking the status of this sector. In qemu-img, this
> triggers an assertion failure.
> 
> In order to fix this, one type for the sector that contains EOF must be
> found. Treating a hole as data is safe, so this patch rounds the
> calculated number of data sectors up, so that a partial sector at EOF is
> treated as a full data sector.
> 
> This fixes https://bugzilla.redhat.com/show_bug.cgi?id=1229394
> 
> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
> ---
>  block/raw-posix.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/block/raw-posix.c b/block/raw-posix.c
> index 2990e95..44ade8c 100644
> --- a/block/raw-posix.c
> +++ b/block/raw-posix.c
> @@ -1848,8 +1848,9 @@ static int64_t coroutine_fn raw_co_get_block_status(BlockDriverState *bs,
>          *pnum = nb_sectors;
>          ret = BDRV_BLOCK_DATA;
>      } else if (data == start) {
> -        /* On a data extent, compute sectors to the end of the extent.  */
> -        *pnum = MIN(nb_sectors, (hole - start) / BDRV_SECTOR_SIZE);
> +        /* On a data extent, compute sectors to the end of the extent,
> +         * possibly including a partial sector at EOF. */

Not only for EOF. If the hole and start are in the same sector, (hole - start) / BDRV_SECTOR_SIZE
will be 0

> +        *pnum = MIN(nb_sectors, DIV_ROUND_UP(hole - start, BDRV_SECTOR_SIZE));
>          ret = BDRV_BLOCK_DATA;
>      } else {
>          /* On a hole, compute sectors to the beginning of the next extent.  */
> 

So, if start is hole, data and  start are in the same sector, (data - start) / BDRV_SECTOR_SIZE
will be 0, you also need to fix it here.

Thanks
Wen Congyang

Re: [Qemu-devel] [PATCH] raw-posix: Fix .bdrv_co_get_block_status() for unaligned image size

[Kevin Wolf]

Am 09.06.2015 um 11:47 hat Wen Congyang geschrieben:
> On 06/09/2015 04:55 PM, Kevin Wolf wrote:
> > Image files with an unaligned image size have a final hole that starts
> > at EOF, i.e. in the middle of a sector. Currently, *pnum == 0 is
> > returned when checking the status of this sector. In qemu-img, this
> > triggers an assertion failure.
> > 
> > In order to fix this, one type for the sector that contains EOF must be
> > found. Treating a hole as data is safe, so this patch rounds the
> > calculated number of data sectors up, so that a partial sector at EOF is
> > treated as a full data sector.
> > 
> > This fixes https://bugzilla.redhat.com/show_bug.cgi?id=1229394
> > 
> > Signed-off-by: Kevin Wolf <kwolf@redhat.com>
> > ---
> >  block/raw-posix.c | 5 +++--
> >  1 file changed, 3 insertions(+), 2 deletions(-)
> > 
> > diff --git a/block/raw-posix.c b/block/raw-posix.c
> > index 2990e95..44ade8c 100644
> > --- a/block/raw-posix.c
> > +++ b/block/raw-posix.c
> > @@ -1848,8 +1848,9 @@ static int64_t coroutine_fn raw_co_get_block_status(BlockDriverState *bs,
> >          *pnum = nb_sectors;
> >          ret = BDRV_BLOCK_DATA;
> >      } else if (data == start) {
> > -        /* On a data extent, compute sectors to the end of the extent.  */
> > -        *pnum = MIN(nb_sectors, (hole - start) / BDRV_SECTOR_SIZE);
> > +        /* On a data extent, compute sectors to the end of the extent,
> > +         * possibly including a partial sector at EOF. */
> 
> Not only for EOF. If the hole and start are in the same sector, (hole - start) / BDRV_SECTOR_SIZE
> will be 0
> 
> > +        *pnum = MIN(nb_sectors, DIV_ROUND_UP(hole - start, BDRV_SECTOR_SIZE));
> >          ret = BDRV_BLOCK_DATA;
> >      } else {
> >          /* On a hole, compute sectors to the beginning of the next extent.  */
> > 
> 
> So, if start is hole, data and  start are in the same sector, (data - start) / BDRV_SECTOR_SIZE
> will be 0, you also need to fix it here.

At first, I thought the same. But how would you ever get a hole that
starts in the middle of a sector? You would have to have a filesystem
with a block size smaller than 512. I don't think that it exists.

Kevin

Re: [Qemu-devel] [PATCH] raw-posix: Fix .bdrv_co_get_block_status() for unaligned image size

[Wen Congyang]

On 06/09/2015 05:49 PM, Kevin Wolf wrote:
> Am 09.06.2015 um 11:47 hat Wen Congyang geschrieben:
>> On 06/09/2015 04:55 PM, Kevin Wolf wrote:
>>> Image files with an unaligned image size have a final hole that starts
>>> at EOF, i.e. in the middle of a sector. Currently, *pnum == 0 is
>>> returned when checking the status of this sector. In qemu-img, this
>>> triggers an assertion failure.
>>>
>>> In order to fix this, one type for the sector that contains EOF must be
>>> found. Treating a hole as data is safe, so this patch rounds the
>>> calculated number of data sectors up, so that a partial sector at EOF is
>>> treated as a full data sector.
>>>
>>> This fixes https://bugzilla.redhat.com/show_bug.cgi?id=1229394
>>>
>>> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
>>> ---
>>>  block/raw-posix.c | 5 +++--
>>>  1 file changed, 3 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/block/raw-posix.c b/block/raw-posix.c
>>> index 2990e95..44ade8c 100644
>>> --- a/block/raw-posix.c
>>> +++ b/block/raw-posix.c
>>> @@ -1848,8 +1848,9 @@ static int64_t coroutine_fn raw_co_get_block_status(BlockDriverState *bs,
>>>          *pnum = nb_sectors;
>>>          ret = BDRV_BLOCK_DATA;
>>>      } else if (data == start) {
>>> -        /* On a data extent, compute sectors to the end of the extent.  */
>>> -        *pnum = MIN(nb_sectors, (hole - start) / BDRV_SECTOR_SIZE);
>>> +        /* On a data extent, compute sectors to the end of the extent,
>>> +         * possibly including a partial sector at EOF. */
>>
>> Not only for EOF. If the hole and start are in the same sector, (hole - start) / BDRV_SECTOR_SIZE
>> will be 0
>>
>>> +        *pnum = MIN(nb_sectors, DIV_ROUND_UP(hole - start, BDRV_SECTOR_SIZE));
>>>          ret = BDRV_BLOCK_DATA;
>>>      } else {
>>>          /* On a hole, compute sectors to the beginning of the next extent.  */
>>>
>>
>> So, if start is hole, data and  start are in the same sector, (data - start) / BDRV_SECTOR_SIZE
>> will be 0, you also need to fix it here.
> 
> At first, I thought the same. But how would you ever get a hole that
> starts in the middle of a sector? You would have to have a filesystem
> with a block size smaller than 512. I don't think that it exists.

You are right, I don't find such filesystem.

Thanks
Wen Congyang


> 
> Kevin
> .
> 

Re: [Qemu-devel] [PATCH] raw-posix: Fix .bdrv_co_get_block_status() for unaligned image size

[Eric Blake]

[-- Attachment #1: Type: text/plain, Size: 918 bytes --]

On 06/09/2015 02:55 AM, Kevin Wolf wrote:
> Image files with an unaligned image size have a final hole that starts
> at EOF, i.e. in the middle of a sector. Currently, *pnum == 0 is
> returned when checking the status of this sector. In qemu-img, this
> triggers an assertion failure.
> 
> In order to fix this, one type for the sector that contains EOF must be
> found. Treating a hole as data is safe, so this patch rounds the
> calculated number of data sectors up, so that a partial sector at EOF is
> treated as a full data sector.
> 
> This fixes https://bugzilla.redhat.com/show_bug.cgi?id=1229394
> 
> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
> ---
>  block/raw-posix.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)

Reviewed-by: Eric Blake <eblake@redhat.com>

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 604 bytes --]

Re: [Qemu-devel] [PATCH] raw-posix: Fix .bdrv_co_get_block_status() for unaligned image size

[Cole Robinson]

On 06/09/2015 04:55 AM, Kevin Wolf wrote:
> Image files with an unaligned image size have a final hole that starts
> at EOF, i.e. in the middle of a sector. Currently, *pnum == 0 is
> returned when checking the status of this sector. In qemu-img, this
> triggers an assertion failure.
> 
> In order to fix this, one type for the sector that contains EOF must be
> found. Treating a hole as data is safe, so this patch rounds the
> calculated number of data sectors up, so that a partial sector at EOF is
> treated as a full data sector.
> 
> This fixes https://bugzilla.redhat.com/show_bug.cgi?id=1229394
> 
> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
> ---
>  block/raw-posix.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/block/raw-posix.c b/block/raw-posix.c
> index 2990e95..44ade8c 100644
> --- a/block/raw-posix.c
> +++ b/block/raw-posix.c
> @@ -1848,8 +1848,9 @@ static int64_t coroutine_fn raw_co_get_block_status(BlockDriverState *bs,
>          *pnum = nb_sectors;
>          ret = BDRV_BLOCK_DATA;
>      } else if (data == start) {
> -        /* On a data extent, compute sectors to the end of the extent.  */
> -        *pnum = MIN(nb_sectors, (hole - start) / BDRV_SECTOR_SIZE);
> +        /* On a data extent, compute sectors to the end of the extent,
> +         * possibly including a partial sector at EOF. */
> +        *pnum = MIN(nb_sectors, DIV_ROUND_UP(hole - start, BDRV_SECTOR_SIZE));
>          ret = BDRV_BLOCK_DATA;
>      } else {
>          /* On a hole, compute sectors to the beginning of the next extent.  */
> 

Fixes the testcase from the bug for me:

Tested-by: Cole Robinson <crobinso@redhat.com>

- Cole

Re: [Qemu-devel] [Qemu-block] [PATCH] raw-posix: Fix .bdrv_co_get_block_status() for unaligned image size

[Stefan Hajnoczi]

[-- Attachment #1: Type: text/plain, Size: 873 bytes --]

On Tue, Jun 09, 2015 at 10:55:08AM +0200, Kevin Wolf wrote:
> Image files with an unaligned image size have a final hole that starts
> at EOF, i.e. in the middle of a sector. Currently, *pnum == 0 is
> returned when checking the status of this sector. In qemu-img, this
> triggers an assertion failure.
> 
> In order to fix this, one type for the sector that contains EOF must be
> found. Treating a hole as data is safe, so this patch rounds the
> calculated number of data sectors up, so that a partial sector at EOF is
> treated as a full data sector.
> 
> This fixes https://bugzilla.redhat.com/show_bug.cgi?id=1229394
> 
> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
> ---
>  block/raw-posix.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)

Thanks, applied to my block tree:
https://github.com/stefanha/qemu/commits/block

Stefan

[-- Attachment #2: Type: application/pgp-signature, Size: 473 bytes --]