Re: 2.6.5-rc3: cat /proc/ide/hpt366 kills disk on second channel

["John Stoffel" <stoffel@lucent.com>]

>>>>> "Bartlomiej" == Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl> writes:

>> I'll also pull this patch forward to 2.6.5 and make sure to submit it
>> to Linus/Andrew, unless you'll do that part?

Bartlomiej> I'll take care of it. :)

Thanks.  I'm going to add your patch into my local 2.6.5 and test it
out, but so far it's been nice and stable.

>> I do wish the cable detection stuff worked though... too bad about the

Bartlomiej> Cable detection works just fine, see init_hwif_hpt366().

Will do.  Mostly, I'm just trying to see if my HPT302 is running a 66,
100, or 133.  Depending on which kernel I boot, I got different
numbers from /proc/ide/hpt366, so I'll see what I can find. 

>> outb() stuff.  Maybe I can poke at it and figure out what kind of
>> locking is required here to make this work right.  Would it need to be
>> queued up as a regular HWIF command?  Can you tell I don't know what
>> I'm talking about?  *grin*

Bartlomiej> regular HWIF command?  It can be fixed using REQ_SPECIAL
Bartlomiej> special request but I don't think it's worth the work -
Bartlomiej> remove #ifdef/#endif DEBUG around printk() in
Bartlomiej> init_hwif_hpt366() if you need info about cable.

God knows what I'm talking about here, but you've given me some good
pointers here, so I'll play around and see what I find.  Thanks again
for your quick work.

Oh yeah, I did a quick audit of the other drivers in drivers/ide/pci
and none of the rest did any outb() or pci_write*() calls in the
various *_get_info() routines, so they rest should be ok now.

Thanks,
John