[PATCH] md: Remove risk of overflow via sprintf) by using snprintf()in md_check_recovery()

sprintf() is dangerous - given the wrong source string it will overflow
the destination. snprintf() is safer in that at least we'll never overflow
the destination. Even if overflow will never happen today, code changes
over time and snprintf() is just safer in the long run.

Signed-off-by: Jesper Juhl
---
md.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

just compile tested

diff --git a/drivers/md/md.c b/drivers/md/md.c
index 0cc30ec..6283658 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -7164,7 +7164,7 @@ void md_check_recovery(mddev_t *mddev)
if (mddev->pers->hot_remove_disk(
mddev, rdev->raid_disk)==0) {
char nm[20];
- sprintf(nm,"rd%d", rdev->raid_disk);
+ snprintf(nm, sizeof(nm), "rd%d", rdev->raid_disk);
sysfs_remove_link(&mddev->kobj, nm);
rdev->raid_disk = -1;
}


--
Jesper Juhl http://www.chaosbits.net/
Plain text mails only, please.
Don't top-post http://www.catb.org/~esr/jargon/html/T/top-post.html

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH] md: Remove risk of overflow via sprintf) by using snprintf()in md_check_recovery()

Jesper Juhl wrote:
> sprintf() is dangerous - given the wrong source string it will overflow
> the destination. snprintf() is safer in that at least we'll never overflow
> the destination. Even if overflow will never happen today, code changes
> over time and snprintf() is just safer in the long run.

> - sprintf(nm,"rd%d", rdev->raid_disk);
> + snprintf(nm, sizeof(nm), "rd%d", rdev->raid_disk);
> sysfs_remove_link(&mddev->kobj, nm);

What if "rd1234" get truncated to "rd123" and you remove the wrong link.
(No, I didn't actually bother to check how much room was allocated.)

Isn't it better to overflow than silently to unlink the wrong file?

What will happen when you try to unlink the "rd123" file again, when the
actual 123 is meant?

Whatever the real fix is, should this be checked for at create_link time
as well?


Daniel K.

Re: [PATCH] md: Remove risk of overflow via sprintf) by using snprintf()in md_check_recovery()

12.02.2011 12:34, Daniel K. wrote:
> Jesper Juhl wrote:
>> sprintf() is dangerous - given the wrong source string it will
>> overflow the destination. snprintf() is safer in that at least we'll
>> never overflow the destination. Even if overflow will never happen
>> today, code changes over time and snprintf() is just safer in the long
>> run.
>
>> - sprintf(nm,"rd%d", rdev->raid_disk);
>> + snprintf(nm, sizeof(nm), "rd%d",
>> rdev->raid_disk);
>> sysfs_remove_link(&mddev->kobj, nm);
>
> What if "rd1234" get truncated to "rd123" and you remove the wrong link.
> (No, I didn't actually bother to check how much room was allocated.)

That allocation is in the line above first sprintf which you deleted.
Sure, didn't bother, it's very difficult.

C'mon guys, this is pointless. 20 bytes allocated for the device
name, and this is for raid disk number. It is impossible to have
more than 10^17 (20 bytes total, 2 for "rd" and on for the zero
terminator) drives in a single array.

/mjt
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH] md: Remove risk of overflow via sprintf) by using snprintf()in md_check_recovery()

Michael Tokarev wrote:
> 12.02.2011 12:34, Daniel K. wrote:
>> Jesper Juhl wrote:
>>> sprintf() is dangerous - given the wrong source string it will
>>> overflow the destination. snprintf() is safer in that at least we'll
>>> never overflow the destination. Even if overflow will never happen
>>> today, code changes over time and snprintf() is just safer in the long
>>> run.
>>> - sprintf(nm,"rd%d", rdev->raid_disk);
>>> + snprintf(nm, sizeof(nm), "rd%d", rdev->raid_disk);
>>> sysfs_remove_link(&mddev->kobj, nm);
>> What if "rd1234" get truncated to "rd123" and you remove the wrong link.
>> (No, I didn't actually bother to check how much room was allocated.)
>
> That allocation is in the line above first sprintf which you deleted.
> Sure, didn't bother, it's very difficult.

Yeah, early morning, I cut to much, and I didn't bother to look it up
again, sorry for being lazy. Nevertheless, the actual size is of the
allocation is of no particular importance. As you've shown, the current
allocation of 20 bytes is more than enough.

> C'mon guys, this is pointless. 20 bytes allocated for the device
> name, and this is for raid disk number. It is impossible to have
> more than 10^17 (20 bytes total, 2 for "rd" and on for the zero
> terminator) drives in a single array.

Agreed, and this was sort of the point.

In all probability it would not overflow, and if it did, it would be
better for it to crash and burn, than to unlink the wrong files.


Daniel K.
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH] md: Remove risk of overflow via sprintf) by usingsnprintf() in md_check_recovery()

On Sat, 12 Feb 2011, Daniel K. wrote:

> Michael Tokarev wrote:
> > 12.02.2011 12:34, Daniel K. wrote:
> > > Jesper Juhl wrote:
> > > > sprintf() is dangerous - given the wrong source string it will
> > > > overflow the destination. snprintf() is safer in that at least we'll
> > > > never overflow the destination. Even if overflow will never happen
> > > > today, code changes over time and snprintf() is just safer in the long
> > > > run.
> > > > - sprintf(nm,"rd%d", rdev->raid_disk);
> > > > + snprintf(nm, sizeof(nm), "rd%d",
> > > > rdev->raid_disk);
> > > > sysfs_remove_link(&mddev->kobj, nm);
> > > What if "rd1234" get truncated to "rd123" and you remove the wrong link.
> > > (No, I didn't actually bother to check how much room was allocated.)
> >
> > That allocation is in the line above first sprintf which you deleted.
> > Sure, didn't bother, it's very difficult.
>
> Yeah, early morning, I cut to much, and I didn't bother to look it up again,
> sorry for being lazy. Nevertheless, the actual size is of the allocation is of
> no particular importance. As you've shown, the current allocation of 20 bytes
> is more than enough.
>
> > C'mon guys, this is pointless. 20 bytes allocated for the device
> > name, and this is for raid disk number. It is impossible to have
> > more than 10^17 (20 bytes total, 2 for "rd" and on for the zero
> > terminator) drives in a single array.
>
> Agreed, and this was sort of the point.
>
> In all probability it would not overflow, and if it did, it would be better
> for it to crash and burn, than to unlink the wrong files.
>

Point taken. Ignore the patch.

--
Jesper Juhl http://www.chaosbits.net/
Plain text mails only, please.
Don't top-post http://www.catb.org/~esr/jargon/html/T/top-post.html

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html