Qlogic ISP 1020

Qlogic ISP 1020

[Jeff Layton]

I recently acquired a new tape drive with a HVD SCSI interface. I also
got a QLogic ISP 1020 based SCSI card to drive it. Unfortunately, I seem
to be having some problems with the qlogicisp driver in the 2.6 kernel.
Occasionally I get a SCSI error, and it takes the tape drive offline,
since the qlogicisp driver doesn't have proper error handling for 2.6.

Is there a maintainer for the 2.6 qlogicisp driver? I saw that Christoph
Hellwig had posted part of a patch to add support for the 1020 and 1040
cards to the qla1280 driver. Has this been completed yet, and if so is
there a place I could download the complete patch? 
If not, I'd be happy to help test patches for this, as my tape setup is
more or less useless to me until I have a driver that has proper error
handling.

Alternately, if anyone knows of a working driver for the QLA1020 for a
2.6 kernel, I'd appreciate any pointers.

Thanks,
-- 
Jeff Layton <jtlayton@poochiereds.net>

Re: Qlogic ISP 1020

[Christoph Hellwig]

On Mon, Mar 22, 2004 at 08:19:54PM -0500, Jeff Layton wrote:
> I recently acquired a new tape drive with a HVD SCSI interface. I also
> got a QLogic ISP 1020 based SCSI card to drive it. Unfortunately, I seem
> to be having some problems with the qlogicisp driver in the 2.6 kernel.
> Occasionally I get a SCSI error, and it takes the tape drive offline,
> since the qlogicisp driver doesn't have proper error handling for 2.6.
> 
> Is there a maintainer for the 2.6 qlogicisp driver? I saw that Christoph
> Hellwig had posted part of a patch to add support for the 1020 and 1040
> cards to the qla1280 driver. Has this been completed yet, and if so is
> there a place I could download the complete patch? 

https://verein.lst.de/~hch/qla1280-isp1040.diff.bz2 has my current patch.

So far I only kno it works with 1040B's on SGI Origins.   If you want to
give it a try apply the patch and make sure you have the qla1280 but not the
qlogicisp driver in the kernel.

I'll try to split the patch into reasoable chunk soon and feed it to Jes.

Re: Qlogic ISP 1020

[Jeff Layton]

On Tue, 2004-03-23 at 01:29, Christoph Hellwig wrote:
> https://verein.lst.de/~hch/qla1280-isp1040.diff.bz2 has my current patch.
> 
> So far I only kno it works with 1040B's on SGI Origins.   If you want to
> give it a try apply the patch and make sure you have the qla1280 but not the
> qlogicisp driver in the kernel.
> 
> I'll try to split the patch into reasoable chunk soon and feed it to Jes.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

I tried this out today, and the results were mixed. Here's the lspci
output for this card:

00:0f.0 SCSI storage controller: QLogic Corp. ISP1020 Fast-wide SCSI
(rev 05)

When I built a patched driver and plugged it in, here's what I saw in
dmesg:

qla1280: no version magic, tainting kernel.
qla1280: QLA1040 found on PCI bus 0, dev 15
scsi(0:0): Resetting SCSI BUS
scsi0 : QLogic QLA1040 PCI to SCSI Host Adapter
       Firmware version:  7.65.00, Driver version 3.25
  Vendor: Quantum   Model: DLT4000           Rev: CD3C
  Type:   Sequential-Access                  ANSI SCSI revision: 02
scsi(0:0:2:0): Sync: period 10, offset 12
st: Version 20040213, fixed bufsize 32768, s/g segs 256
Attached scsi tape st0 at scsi0, channel 0, id 2, lun 0
st0: try direct i/o: yes, max page reachable by HBA 1048575

Looked pretty good, other than it seems to think that my card is a 1040
and not a 1020. I then tried to do some operations with my tape drive:

jtlayton@salusa:/usr/src/linux-2.6.4% sudo mt -f /dev/nst0 status
SCSI 2 tape drive:
File number=-1, block number=-1, partition=0.
Tape block size 0 bytes. Density code 0x0 (default).
Soft error count since last status=0
General status bits on (1010000):
 ONLINE IM_REP_EN

I then tried to use tar to write a file to the tape and it failed almost
immediately:

jtlayton@salusa:~% sudo tar cvf /dev/nst0 ./wolf-linux-1.33.x86.run
./wolf-linux-1.33.x86.run
tar: /dev/nst0: Wrote only 0 of 10240 bytes
tar: Error is not recoverable: exiting now

These errors seem to cause these messages to pop up in dmesg:

st0: Error 70000 (sugg. bt 0x0, driver bt 0x0, host bt 0x7).
st0: Error 70000 (sugg. bt 0x0, driver bt 0x0, host bt 0x7).
st0: Error 70000 (sugg. bt 0x0, driver bt 0x0, host bt 0x7).

Please let me know if there's any info I can provide that might help, or
if you need me to test another patch.

Heck, I'll even offer up a case of beer (or an equivalent reward) if we
can get it working!

Thanks,
-- 
Jeff Layton <jtlayton@poochiereds.net>

Re: Qlogic ISP 1020

[Christoph Hellwig]

On Tue, Mar 23, 2004 at 04:00:53PM -0500, Jeff Layton wrote:
> On Tue, 2004-03-23 at 01:29, Christoph Hellwig wrote:
> > https://verein.lst.de/~hch/qla1280-isp1040.diff.bz2 has my current patch.
> > 
> > So far I only kno it works with 1040B's on SGI Origins.   If you want to
> > give it a try apply the patch and make sure you have the qla1280 but not the
> > qlogicisp driver in the kernel.
> > 
> > I'll try to split the patch into reasoable chunk soon and feed it to Jes.
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> I tried this out today, and the results were mixed. Here's the lspci
> output for this card:

In current mainline the qla1280 driver supports the 1020 now.  Care to test
again?  I've heard success reports for both i386 and alpha so far.

Re: Qlogic ISP 1020

[Ingvar Hagelund]

* Christoph Hellwig
> > > https://verein.lst.de/~hch/qla1280-isp1040.diff.bz2 has my current patch.
(...)
> > > So far I only kno it works with 1040B's on SGI Origins.  If you
> > > want to give it a try apply the patch and make sure you have the
> > > qla1280 but not the qlogicisp driver in the kernel.
> > >
> > > I'll try to split the patch into reasoable chunk soon and feed
> > > it to Jes.

* Jeff Layton
> > I tried this out today, and the results were mixed.

* Christoph Hellwig
> In current mainline the qla1280 driver supports the 1020 now. Care
> to test again?  I've heard success reports for both i386 and alpha
> so far.

Just to mention it, I tried this driver on our Alpha with 2.4.27 (with
the mentioned patch), and it runs rock stable, at least after the
first 24 hours. This finally solved my problems with md raid1 on the
1020, see posting <ujcwty6lp0n.fsf@nfsd.linpro.no>,
 "oops in qlogicisp.c when running 2.4.27 with md RAID1 on Alpha". 
The driver was patched upon 3.24.3 (stolen from kernel 2.6.8.1) and
compiles easily on kernel 2.4.27 with some minor updates in
include/asm-alpha/io.h. It is also a lot more stable than the infamous
feral driver.

Thanks guys, this really made my day.

Norbert, could this be something for the Debian alpha kernels?

Ingvar

-- 
Many that live deserve death. And some that die deserve life. Can you
give it to them? Then do not be too eager to deal out death in
judgement. For even the very wise cannot see all ends.
								Gandalf

RE: Qlogic ISP 1020

[Bailey, Scott]

To recap, I have a 3-processor Alphaserver 4100 5/466 with several KZPBA-BA
(ISP1020A-based) SCSI controllers that runs Debian Linux on top of LVM on
top of md RAID-1 devices. I've been patching the feral driver into my kernel
but on the lookout for something easier to maintain.

Christoph developed a patch for the qla1280 driver (under 2.6) that
introduced support for ISP1020/ISP1040 devices; Ingvar brought this to my
attention and provided patches to backport this support to 2.4. These are
available at http://users.linpro.no/ingvar/alpha/qla1280/ for those coming
late to this party. :-)

I tried building a kernel from the Debian 2.4.27 kernel source with these
patches included; on 3 separate attempts, my system would lock up part-way
through the boot process.

Following up on a last suggestion from Ingvar, I rebuilt my 2.4.27 kernel
with CONFIG_SMP turned off.

It wasn't a huge test, but I was able to boot my system cleanly without any
problems. That's much farther than I got running the SMP-enabled kernel.

Therefore I conclude that either I am the first person in the universe to
use the 2.4 qla1280 driver on a multiprocessor Alpha system, or the ISP10x0
support patch somehow is not SMP-safe (at least as backported).

I am happy and eager to do more playtesting if anybody has more suggestions,
requests for additional information, or revised patches appear. (As I
mentioned before, I am clueless about using magic SYSRQ keys on an
Alphaserver with a serial console.)

Tantalized,

	Scott