lkml.org 
[lkml]   [2008]   [Jun]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [PATCH] IDE: Fix HDIO_DRIVE_RESET handling
Date
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> wrote:
> On Monday 23 June 2008, Elias Oltmanns wrote:
[...]
>> Indeed, the following patch series is based on nex-20080620. Just to be
>> absolutely clear though, this is actually a bug fix since a
>>
>> # hdparm -w /dev/hda
>>
>> currently freezes the system if there happens to be any I/O operation in
>> progress. Not sure whether this is serious enough for -rc or, indeed,
>> -stable trees, but I thought I'd mention it.
>
> According to 'man hdparm':
>
> -w Perform a device reset (DANGEROUS). Do NOT use this option. It
> exists for unlikely situations where a reboot might otherwise be
> required to get a confused drive back into a useable state.
>
> so I don't think that a rush is necessary (however we may still want to get
> patch #1 in for 2.6.26).

I see.

>
> BTW Your fix adds framework which can be re-used for fixing locking of
>IDE
> settings (+ it finally makes sense to dust-off my IDE settings rework which
> was always low-prio and was never posted :) and maybe it could be also used
> for drive->special handling.

Alan's remark made me think again and I'm still not quite sure whether I
can (and indeed should) fix ide_abort() and allow for out of band
execution of HDIO_DRIVE_RESET as well as idle immediate with head unload
for disk shock protection. What are your views on Alan's remark and my
reply?

[...]
>> 4. Adds some more error reporting facilities and documents them as well.
>>
>> Please be particularly alert when reviewing the last patch. I merely
>> did what seemed to be the right and obvious thing to do but I ironed out
>> some irregularities along the way which (for all improbability) may have
>> been there for some reason or other. It beats me, for instance, why
>> ->polling but not ->resetting should be reset to 0 when
>> sil_sata_reset_poll() returns non zero. So, I now both are 0 once any
>> of the poll functions returns ide_stopped.
>
> Looks OK but please move the above description to the patch description.

Will do.

>
> Patches #2 and #3 also look good.

Thanks for reviewing.

Regards,

Elias


\
 
 \ /
  Last update: 2008-06-24 09:25    [from the cache]
©2003-2014 Jasper Spaans. Advertise on this site