[Cbe-oss-dev] [regression/bisected] corrupt CD data after media change and delay

[Alessandro Suardi]

On Sun, Jul 13, 2008 at 3:33 PM, James Bottomley
<James.Bottomley at hansenpartnership.com> wrote:
> On Sat, 2008-07-12 at 14:08 +0200, Alessandro Suardi wrote:
>> On Fri, Jul 11, 2008 at 11:25 PM, Alessandro Suardi
>> sr0: rw=0, want=3291928, limit=2097151
>> [asuardi at sandman sdlmame0126]$ h | grep siz
>>   958  cat /sys/block/sr0/size
>>  1031  h | grep siz
>> [asuardi at sandman sdlmame0126]$ !958
>> cat /sys/block/sr0/size
>> 6993536
>>
>> <eject, remount DVD>
>>
>> [asuardi at sandman sdlmame0126]$ !mount
>> mount -r /cdrom
>> [asuardi at sandman sdlmame0126]$ cat /sys/block/sr0/size
>> 6993536
>>
>> <now everything works fine>
>>
>> Possibly a new bug then - since size is always the same ?
>
> It looks like a re-manifestation of Geert's bug.
>
> You said everything worked in -rc7, and there actually have been no sr
> changes at all between then and now, so I'm a bit confused as to how it
> could reappear.  Could you try a bisection search for it (first
> re-verify -rc7 is OK, just in case that's a false negative).

I have been away a few days - I'll try this weekend rebuilding
 older kernels (I save religiously my .configs for every kernel
 I built in the last, um, five years at least) and give the reboot/
 DVD mount/access sequence a go...

The good news is that in 2.6.26-git4, -git9, -git10 I have been
 unable to reproduce the issue a single time. Still, I'm willing
 to make a bit more effort in trying to reproduce.

--alessandro

 "Give me love / Or give me hate
 Give me anything that's not just ok"

 (Sophia, 'Weightless')