On Wed, Aug 14, 2013 at 10:40 AM, Geert Uytterhoeven
<geert@xxxxxxxxxxxxxx> wrote:
> On Tue, Aug 13, 2013 at 6:32 PM, Geert Uytterhoeven
> <geert@xxxxxxxxxxxxxx> wrote:
>> On Fri, 9 Aug 2013, Zach Brown wrote:
>>> On Fri, Aug 09, 2013 at 02:26:36PM +0200, Andreas Schwab wrote:
>>> > Josef Bacik <jbacik@xxxxxxxxxxxx> writes:
>>> >
>>> > > So stripe_len shouldn't be 0, if it is you have bigger problems :).
>>> >
>>> > The bigger problem is that stripe_nr is u64, this is completely bogus.
>>> > The first operand of do_div must be u32. This goes through the whole
>>> > file.
>>
>> This was introduced by commit 53b381b3abeb86f12787a6c40fee9b2f71edc23b
>> ("Btrfs: RAID5 and RAID6"), which changed the divisor from
>> map->stripe_len (struct map_lookup.stripe_len is int) to a 64-bit
>> temporary.
>>
>>> Definitely. Can we get some typeof() tricks in the macros to have the
>>> build fail if (when, evidently) someone gets it wrong?
>>
>> Not using typeof, as there are way too many callsites where int is used
>> instead of u32.
>>
>> However, checking that sizeof() equals to 4 seems to work.
>> Below is a patch for asm-generic, which is untested, but it works when
>> adding the same checks to arch/m68k/include/asm/div64.h
>>
>> This is not something we just want to drop in, as it has the potential of
>> breaking lots of things (yes, it breaks btrfs :-)
>
> Found so far:
> - Several calls to sector_div() in blkdev_issue_discard()
> - Two calls to do_div() in sd_completed_bytes()
>
> Some of these even operate on dividends that never exceed 32-bit, tss...
Two more:
drivers/md/dm-cache-target.c:too_many_discard_blocks
fs/jfs/jfs_dmap.c:dbDiscardAG
These bring in the 64-bit divisor from somewhere else, so they're less
trivial to fix.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
