Re: [PATCH v2] floppy: refactor open() flags handling (was Re: mm: uninterruptable tasks hanged on mmap_sem)

[Dmitry Vyukov]

On Fri, Feb 5, 2016 at 2:51 PM, Jiri Kosina <jikos@xxxxxxxxxx> wrote:
> On Fri, 5 Feb 2016, Dmitry Vyukov wrote:
>
>> > could you please feed the patch below (on top of the previous floppy fix)
>> > to your syzkaller machinery and test whether you are still able to
>> > reproduce the problem? It passess my local testing here.
>>
>> Now that open exits early with EWOULDBLOCK, I guess the reproduced is
>> not doing anything particularly interesting.
>
> Yeah. But as I explained in the changelog, I think it's a valid thing to
> do (opinions welcome).
>
> I don't think having a huge discussion about what nonblocking really means
> for floppy and then try to refactor the whole driver to support that would
> make sense.

I don't have any objections. And I agree that it does not make sense
to spend any considerable time on optimizing this driver.


> Alternatively we can take more conservative aproach, accept the
> nonblocking flag, but do the regular business of the driver.
>
> Actually, let's try that, to make sure that we don't introduce userspace
> breakage.
>
> Could you please retest with the patch below?

Reapplied.
Agree that it's better to not bail out on O_NONBLOCK.