2001-02-28 13:10:23

by Igor Mozetic

[permalink] [raw]
Subject: 2.4.2 + aic7xxx still broken

I'm still unable to boot any 2.4.x kernel on this box:
Intel C440GX+, on-board Adaptec AIC-7896/7, 2GB RAM.

2.4.2 + stock aic7xxx:
----------------------
...
SCSI host 0 channel 0 reset (pid 0) timed out - trying harder
SCSI bus is being reset for host 0 channel 0
... ad infinitum

2.4.2 + http://people.FreeBSD.org/~gibbs/linux/ aic7xxx-6.1.4-2.4.2-pre4:
-------------------------------------------------------------------------
...
aic7xxx - abort returns 8194
scsi1:0:0:0 Attempting to queue an ABORT message
scsi1:0:0:0 Command already completed
scsi1:0:0:0 Attempting to queue a TARGET RESET message
scsi1:0:0:0 is not an active device
... ad infinitum

2.2.[17,18] stock driver works fine.
I can provide more info if needed. Any help appreciated.

-Igor Mozetic


2001-02-28 14:56:10

by Doug Ledford

[permalink] [raw]
Subject: Re: 2.4.2 + aic7xxx still broken

Igor Mozetic wrote:
>
> I'm still unable to boot any 2.4.x kernel on this box:
> Intel C440GX+, on-board Adaptec AIC-7896/7, 2GB RAM.
>
> 2.4.2 + stock aic7xxx:
> ----------------------
> ...
> SCSI host 0 channel 0 reset (pid 0) timed out - trying harder
> SCSI bus is being reset for host 0 channel 0
> ... ad infinitum
>
> 2.4.2 + http://people.FreeBSD.org/~gibbs/linux/ aic7xxx-6.1.4-2.4.2-pre4:
> -------------------------------------------------------------------------
> ...
> aic7xxx - abort returns 8194
> scsi1:0:0:0 Attempting to queue an ABORT message
> scsi1:0:0:0 Command already completed
> scsi1:0:0:0 Attempting to queue a TARGET RESET message
> scsi1:0:0:0 is not an active device
> ... ad infinitum
>
> 2.2.[17,18] stock driver works fine.
> I can provide more info if needed. Any help appreciated.

Sounds like an IRQ routing problem. Have you tried a UP kernel with IO-APIC
on UP support disabled or an SMP kernel both with and without the noapic
options to see if they make a difference?

--

Doug Ledford <[email protected]> http://people.redhat.com/dledford
Please check my web site for aic7xxx updates/answers before
e-mailing me about problems

2001-03-10 17:35:40

by Marc Lehmann

[permalink] [raw]
Subject: Re: 2.4.2 + aic7xxx still broken

On Wed, Feb 28, 2001 at 02:07:30PM +0100, Igor Mozetic <[email protected]> wrote:
> 2.4.2 + stock aic7xxx:
> ----------------------
> ...
> SCSI host 0 channel 0 reset (pid 0) timed out - trying harder

interestingly, I have exactly the same problems when booting my smp kernel
with either maxcpus=1, nosmp or the second cpu removed but NOT when the
kernel boots with two cpus (it works *perfectly*)

Unless macpus=! switches off apic (it doens't) this doesn't look like a
IRAQ problem, as the bios has no idea of the maxcpus=! option.

One thing that puzzles me is why the new driver looks for db_185.h in
/usr/include/db, which seems to be a rather nonstandard position for that
header (none my my slackware or home-grown boxes have that directory, all
of them have the db_185.h file in /usr/include, which is the standard
location I'd think since glibc-2.1 installed it there).

--
-----==- |
----==-- _ |
---==---(_)__ __ ____ __ Marc Lehmann +--
--==---/ / _ \/ // /\ \/ / [email protected] |e|
-=====/_/_//_/\_,_/ /_/\_\ XX11-RIPE --+
The choice of a GNU generation |
|

2001-03-15 12:31:39

by Igor Mozetic

[permalink] [raw]
Subject: Re: 2.4.2 + aic7xxx still broken

Doug Ledford writes:

> Either one should work. Try it with the UP_IOAPIC support enabled and see if
> that helps. If it doesn't, then I would try Justin's driver and see if it
> works.

Thanks, the machine now boots.
I'm using 2.4.3-pre3 + linux-aic7xxx-6.1.7 (UP on SMP board)
with the following APIC support enabled:

CONFIG_X86_GOOD_APIC=y
CONFIG_X86_UP_IOAPIC=y
CONFIG_X86_IO_APIC=y
CONFIG_X86_LOCAL_APIC=y

-Igor Mozetic