Re: 2.6.0-test9-mm2 - AIO tests still gets slab corruption

[Andrew Morton]

Daniel McNeil <daniel@xxxxxxxx> wrote:
>
> Andrew,
> 
> test9-mm2 is still getting slab corruption with AIO:

Why?

> Maximal retry count.  Bytes done 0
> Slab corruption: start=dc70f91c, expend=dc70f9eb, problemat=dc70f91c
> Last user: [<c0192fa3>](__aio_put_req+0xbf/0x200)
> Data: 00 01 10 00 00 02 20 00 *********6C ******************************A5
> Next: 71 F0 2C .A3 2F 19 C0 71 F0 2C .********************
> slab error in check_poison_obj(): cache `kiocb': object was modified after freeing
> 
> With suparna's retry-based-aio-dio patch, there are no kernel messages
> and the tests do not see any uninitialized data.
> 
> Any reason not to add suparna's patch to -mm to fix these problems?

It relies on infrastructure which is not present in Linus's kernel.  We
should only be interested in fixing mainline 2.6.x.

Furthermore I'd like to see the direct-vs-buffered locking fixes fully
implemented against Linus's tree, not -mm.  They're almost there, but are
not quite complete.  Running off and making it dependent on the retry
infrastructure is not really helpful.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/