[ltp] Re: enabling protect file for hdaps on t61

[Elias Oltmanns]

Vincent C Jones <v.jones@networkingunlimited.com> wrote:
> On Mon, 2008-02-11 at 01:00 +0100, Rolf Kutz wrote:
>> On 10/02/08 00:18 +0100, Elias Oltmanns wrote:
>> >
>> >Is this with libata (and ata_piix in particular)? Try the following
>> >experiment:
>> 
>> This is an SATA hdd, maybe that's the difference.
>> ata_piix is just for the DVD.:
>> # lsmod|grep ata
>> ata_piix               23236  0
>> libata                174960  2 ahci,ata_piix
>> scsi_mod              179192  4 sd_mod,sg,sr_mod,libata

Yes, I mainly wanted to make sure that we are talking about libata
rather than old style ide drivers. As it turns out, the differences
between ahci and ata_piix don't come into it.

>> 
>> >Switch to single user mode and remount the partitions read only. Perform
>> >the following commands:
>> >
>> ># cat some-big-file > /dev/null &
>> ># echo 1 > /sys/block/sda/queue/protect
>> > 
>> >Repeat the second one several times while the read operation of the
>> >first one is in progress. Make sure that you actually hear the drive
>> >park heads and resume normal operation. My system freezes after the
>> >third park command at the latest.
>> 
>> Done that and it didn't freeze the system even
>> after ten times. That doesn't mean that there
>> isn't a bug. It just doesn't trigger on this
>> system.
>> 
>> regards, Rolf
>
> SATA drive here as well (7675 X61, AHCI mode, SuSE 10.3, 2.6.24 with
> SuSE, acpi, protect, & tp_smapi patches applied) and I have no problem
> invoking a freeze. What is your magic incantation to avoid the problem?
> I really would like to get HDAPS back...

The difference isn't so much SATA vs PATA but SMP system vs uniprocessor
/ non-SMP system. Probably Rolf is running an SMP kernel on a dual core
machine whereas your system is lacking SMP support.

Yesterday, I've found the culprit and I have contacted libata developers
in order to find a proper solution for this issue.

Regards,

Elias