Re: [RFT][PATCH] ide-disk.c: more write cache fixes

[Rene Herman <rene.herman@keyaccess.nl>]

Bartlomiej Zolnierkiewicz wrote:

> On Sunday 16 of May 2004 21:58, Alan Cox wrote:
> 
>>On Fri, May 14, 2004 at 01:58:58PM +0200, Rene Herman wrote:
>>
>>>Have again attached a 'rollup' patch against vanilla 2.6.6, including
>>>this, Andrew's SYSTEM_SHUTDOWN split and the quick "don't switch of
>>>spindle if rebooting" hack. Again, just in case anyone finds it useful.
>>
>>This reintroduces corruption on my thinkpad 600.
> 
> [ this corruption was fixed by kernel 2.6.6 ]
> 
> Please see if reverting changes to ide_device_shutdown() helps.

Bart, could something like:

if (system_state == SYSTEM_RESTART) {
	ide_cache_flush_p(drive)
	return;
}

(as opposed to just the return in that patch and -mm3) possibly help? I 
sort of expect that ide_cache_flush_p() is already called further on up?

Alan, if you know, that drive fails ide_id_has_flush_cache()?

Note, very aware I don't know what the fuck I'm doing here (and equally 
aware I don't _want_ to be here :-) Having the drive spin down on each 
reboot is totally unacceptable though. Not only does spinning up again 
take significant time and noise, it's also actively bad for the drive.

If there's no sane way to fix this, an explicit blacklist for drives 
that really need to be shutdown? Eew.

Rene.