2011-02-06 01:33:50

by Nitin Gupta

[permalink] [raw]
Subject: [PATCH] zram: fix data corruption issue

In zram_read() and zram_write() we were not incrementing the
index number and thus were reading/writing values from/to
incorrect sectors on zram disk, resulting in data corruption.

Signed-off-by: Nitin Gupta <[email protected]>
---
drivers/staging/zram/zram_drv.c | 4 ++++
1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/drivers/staging/zram/zram_drv.c b/drivers/staging/zram/zram_drv.c
index 5415712..4bd8cbd 100644
--- a/drivers/staging/zram/zram_drv.c
+++ b/drivers/staging/zram/zram_drv.c
@@ -227,6 +227,7 @@ static int zram_read(struct zram *zram, struct bio *bio)

if (zram_test_flag(zram, index, ZRAM_ZERO)) {
handle_zero_page(page);
+ index++;
continue;
}

@@ -235,12 +236,14 @@ static int zram_read(struct zram *zram, struct bio *bio)
pr_debug("Read before write: sector=%lu, size=%u",
(ulong)(bio->bi_sector), bio->bi_size);
/* Do nothing */
+ index++;
continue;
}

/* Page is stored uncompressed since it's incompressible */
if (unlikely(zram_test_flag(zram, index, ZRAM_UNCOMPRESSED))) {
handle_uncompressed_page(zram, page, index);
+ index++;
continue;
}

@@ -320,6 +323,7 @@ static int zram_write(struct zram *zram, struct bio *bio)
mutex_unlock(&zram->lock);
zram_stat_inc(&zram->stats.pages_zero);
zram_set_flag(zram, index, ZRAM_ZERO);
+ index++;
continue;
}

--
1.7.3.5


2011-02-06 05:00:11

by Greg KH

[permalink] [raw]
Subject: Re: [PATCH] zram: fix data corruption issue

On Sat, Feb 05, 2011 at 08:34:20PM -0500, Nitin Gupta wrote:
> In zram_read() and zram_write() we were not incrementing the
> index number and thus were reading/writing values from/to
> incorrect sectors on zram disk, resulting in data corruption.

Is this something that needs to go into .38 or can it wait until .39?

thanks,

greg k-h

2011-02-06 05:02:36

by Nitin Gupta

[permalink] [raw]
Subject: Re: [PATCH] zram: fix data corruption issue

On 02/06/2011 12:00 AM, Greg KH wrote:
> On Sat, Feb 05, 2011 at 08:34:20PM -0500, Nitin Gupta wrote:
>> In zram_read() and zram_write() we were not incrementing the
>> index number and thus were reading/writing values from/to
>> incorrect sectors on zram disk, resulting in data corruption.
>
> Is this something that needs to go into .38 or can it wait until .39?
>


This really needs to get into .38 or users will see random data
corruption on zram devices.

Thanks,
Nitin

2011-02-06 05:10:08

by Greg KH

[permalink] [raw]
Subject: Re: [PATCH] zram: fix data corruption issue

On Sun, Feb 06, 2011 at 12:03:12AM -0500, Nitin Gupta wrote:
> On 02/06/2011 12:00 AM, Greg KH wrote:
> >On Sat, Feb 05, 2011 at 08:34:20PM -0500, Nitin Gupta wrote:
> >>In zram_read() and zram_write() we were not incrementing the
> >>index number and thus were reading/writing values from/to
> >>incorrect sectors on zram disk, resulting in data corruption.
> >
> >Is this something that needs to go into .38 or can it wait until .39?
> >
>
>
> This really needs to get into .38 or users will see random data
> corruption on zram devices.

Ok, I'll queue it up for there. What about .37 and/or .36?

thanks,

greg k-h

2011-02-06 05:32:59

by Nitin Gupta

[permalink] [raw]
Subject: Re: [PATCH] zram: fix data corruption issue

On 02/06/2011 12:11 AM, Greg KH wrote:
> On Sun, Feb 06, 2011 at 12:03:12AM -0500, Nitin Gupta wrote:
>> On 02/06/2011 12:00 AM, Greg KH wrote:
>>> On Sat, Feb 05, 2011 at 08:34:20PM -0500, Nitin Gupta wrote:
>>>> In zram_read() and zram_write() we were not incrementing the
>>>> index number and thus were reading/writing values from/to
>>>> incorrect sectors on zram disk, resulting in data corruption.
>>>
>>> Is this something that needs to go into .38 or can it wait until .39?
>>>
>>
>>
>> This really needs to get into .38 or users will see random data
>> corruption on zram devices.
>
> Ok, I'll queue it up for there. What about .37 and/or .36?
>


The same problem exists on .37 and .36. I just checked that the same
patch applies cleanly to both these version so may be included in their
respective maintenance releases.

Thanks,
Nitin