block: Don't allow illegal discard requests
diff mbox

Message ID 1446628721-11030-1-git-send-email-jack@suse.com
State New
Headers show

Commit Message

Jan Kara Nov. 4, 2015, 9:18 a.m. UTC
Currently blkdev_issue_discard() doesn't check that submitted discard
request matches granularity required by the underlying device.
Submitting such requests to the device can have unexpected consequences
(e.g. brd driver just discards more / less data depending on how exactly
the request is aligned). So test that submitted discard request has the
granularity required by the driver.

Signed-off-by: Jan Kara <jack@suse.com>
---
 block/blk-lib.c | 7 +++++++
 1 file changed, 7 insertions(+)

Comments

kernel test robot Nov. 4, 2015, 10:46 a.m. UTC | #1
Hi Jan,

[auto build test ERROR on: ext3/for_next]
[cannot apply to: v4.3 next-20151103]

url:    https://github.com/0day-ci/linux/commits/Jan-Kara/block-Don-t-allow-illegal-discard-requests/20151104-172203
base:   https://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs.git for_next
config: arm-prima2_defconfig (attached as .config)
reproduce:
        wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # save the attached .config to linux build tree
        make.cross ARCH=arm 

All errors (new ones prefixed by >>):

   block/built-in.o: In function `blkdev_issue_discard':
>> block/blk-lib.c:67: undefined reference to `__aeabi_uldivmod'
>> block/blk-lib.c:67: undefined reference to `__aeabi_uldivmod'

vim +67 block/blk-lib.c

    61		if (!blk_queue_discard(q))
    62			return -EOPNOTSUPP;
    63	
    64		if (q->limits.discard_granularity) {
    65			unsigned int gran_sect = q->limits.discard_granularity >> 9;
    66	
  > 67			if (sector % gran_sect || nr_sects % gran_sect)
    68				return -EINVAL;
    69		}
    70	

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation
kernel test robot Nov. 4, 2015, 10:48 a.m. UTC | #2
Hi Jan,

[auto build test ERROR on: ext3/for_next]
[cannot apply to: v4.3 next-20151103]

url:    https://github.com/0day-ci/linux/commits/Jan-Kara/block-Don-t-allow-illegal-discard-requests/20151104-172203
base:   https://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs.git for_next
config: avr32-atngw100_defconfig (attached as .config)
reproduce:
        wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # save the attached .config to linux build tree
        make.cross ARCH=avr32 

All errors (new ones prefixed by >>):

   block/built-in.o: In function `ll_back_merge_fn':
>> (.text+0x90dc): undefined reference to `__avr32_umod64'

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation
kernel test robot Nov. 4, 2015, 10:53 a.m. UTC | #3
Hi Jan,

[auto build test ERROR on: ext3/for_next]
[cannot apply to: v4.3 next-20151103]

url:    https://github.com/0day-ci/linux/commits/Jan-Kara/block-Don-t-allow-illegal-discard-requests/20151104-172203
base:   https://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs.git for_next
config: arm-sa1100 (attached as .config)
reproduce:
        wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # save the attached .config to linux build tree
        make.cross ARCH=arm 

All errors (new ones prefixed by >>):

   block/built-in.o: In function `blkdev_issue_discard':
>> :(.text+0xcdf4): undefined reference to `__umoddi3'
   :(.text+0xce0c): undefined reference to `__umoddi3'

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation
Christoph Hellwig Nov. 4, 2015, 11:30 a.m. UTC | #4
On Wed, Nov 04, 2015 at 10:18:41AM +0100, Jan Kara wrote:
> Currently blkdev_issue_discard() doesn't check that submitted discard
> request matches granularity required by the underlying device.
> Submitting such requests to the device can have unexpected consequences
> (e.g. brd driver just discards more / less data depending on how exactly
> the request is aligned). So test that submitted discard request has the
> granularity required by the driver.

How is this going to work with stacked drivers?  I think brd needs to
be fixed to take care of the granularity itself, just like we did for
request based drivers.
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Jan Kara Nov. 4, 2015, 2:17 p.m. UTC | #5
On Wed 04-11-15 03:30:19, Christoph Hellwig wrote:
> On Wed, Nov 04, 2015 at 10:18:41AM +0100, Jan Kara wrote:
> > Currently blkdev_issue_discard() doesn't check that submitted discard
> > request matches granularity required by the underlying device.
> > Submitting such requests to the device can have unexpected consequences
> > (e.g. brd driver just discards more / less data depending on how exactly
> > the request is aligned). So test that submitted discard request has the
> > granularity required by the driver.
> 
> How is this going to work with stacked drivers?  I think brd needs to
> be fixed to take care of the granularity itself, just like we did for
> request based drivers.

So you mean that a stacked driver would split a discard request so that it
is no longer of the required granularity? OK, makes sense. I'll add the
check directly into brd itself.

								Honza

Patch
diff mbox

diff --git a/block/blk-lib.c b/block/blk-lib.c
index bd40292e5009..7170d02e9e9a 100644
--- a/block/blk-lib.c
+++ b/block/blk-lib.c
@@ -61,6 +61,13 @@  int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
 	if (!blk_queue_discard(q))
 		return -EOPNOTSUPP;
 
+	if (q->limits.discard_granularity) {
+		unsigned int gran_sect = q->limits.discard_granularity >> 9;
+
+		if (sector % gran_sect || nr_sects % gran_sect)
+			return -EINVAL;
+	}
+
 	if (flags & BLKDEV_DISCARD_SECURE) {
 		if (!blk_queue_secdiscard(q))
 			return -EOPNOTSUPP;