Hi Jens.
I wanted to try out blk-7 to see if it cured the abysmal I/O
performance
on 2.4.0-test10, but it won't boot on my system. The last message I
see
is the banner of the SCSI host adapter init (it found the card) but
it
never actually lists the devices on the host... weird place to crash
actually. I tried it twice, and waited about a minute each time for it
to
make progress.
In other words, this appears:
(scsi0) <Adaptec AHA-294X Ultra2 SCSI host adapter> found at PCI 0/11/0
(scsi0) Wide Channel, SCSI ID=7, 32/255 SCBs
(scsi0) Downloading sequencer code... 392 instructions downloaded
scsi0 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 5.2.1/5.2.0
<Adaptec AHA-294X Ultra2 SCSI host adapter>
But it never makes it to:
(scsi0:0:0:0) Synchronous at 80.0 Mbyte/sec, offset 15.
Vendor: SEAGATE Model: ST39102LW Rev: 0006
Type: Direct-Access ANSI SCSI revision: 02
(scsi0:0:2:0) Synchronous at 80.0 Mbyte/sec, offset 31.
Vendor: IBM Model: DNES-318350W Rev: SA30
Type: Direct-Access ANSI SCSI revision: 03
Does this ring a bell? I tried using SYSRQ and it shows the 'current'
as kapmd-idled and the showPc shows the PC at c01108ff ->
apm_bios_call_simple. Hmm.
I'll try any subsequent patch you can offer.
My system is a single processor Athlon 700, 256mb ram, 2.4.0-test10 plus
blk-7.
David Mansfield
P.S. I added the #define ELEVATOR_MERGE_HOLE 3 like someone else
mentioned, and the patch applied cleanly except filemap.c which was
offset -20 lines.
On Thu, Nov 02 2000, David Mansfield wrote:
> Hi Jens.
>
> I wanted to try out blk-7 to see if it cured the abysmal I/O
> performance
> on 2.4.0-test10, but it won't boot on my system. The last message I
> see
> is the banner of the SCSI host adapter init (it found the card) but
> it
Yes, known bug. The two scsi queueing functions need to plug the
device, it's fixed here. I don't have a clean blk-7 tree atm, but
I'll put up a blk-8 in an hour or so with that fix and others.
--
* Jens Axboe <[email protected]>
* SuSE Labs
On Thu, Nov 02 2000, David Mansfield wrote:
> Hi Jens.
>
> I wanted to try out blk-7 to see if it cured the abysmal I/O
> performance on 2.4.0-test10, but it won't boot on my system.
Could you try blk-8?
*.kernel.org/pub/linux/kernel/people/axboe/patches/2.4.0-test10/blk-8.bz2
It also has other fixes over blk-7 (saves the linear back scan when
a buffer can't be merged, simple aging of requests in queue, and
more "fair" accouting of merges)
--
* Jens Axboe <[email protected]>
* SuSE Labs