megaraid irq disable after days
From: COMBES Julien - CETE Lyon/DI/ET/PAMELA
Date: Tue May 24 2005 - 09:15:25 EST
I have a problem with a kernel 2.6.10 (sources from debian) which
disable IRQ of my megraid (driver megaraid_mbox) on several servers
after days of work and several millions of interuptions.
When the IRQ is disable, the servers do that :
May 22 03:02:20 relternet-01 kernel: irq 17: nobody cared!
May 22 03:02:20 relternet-01 kernel: [__report_bad_irq+42/160]
May 22 03:02:20 relternet-01 kernel: [handle_IRQ_event+48/112]
May 22 03:02:20 relternet-01 kernel: [note_interrupt+112/176]
May 22 03:02:20 relternet-01 kernel: [__do_IRQ+304/320]
May 22 03:02:20 relternet-01 kernel: [do_IRQ+25/48] do_IRQ+0x19/0x30
May 22 03:02:20 relternet-01 kernel: [common_interrupt+26/32]
May 22 03:02:20 relternet-01 kernel: [mwait_idle+51/80]
May 22 03:02:20 relternet-01 kernel: [cpu_idle+59/80] cpu_idle+0x3b/0x50
May 22 03:02:20 relternet-01 kernel: handlers:
May 22 03:02:20 relternet-01 kernel: [pg0+944120576/1069794304]
May 22 03:02:20 relternet-01 kernel: Disabling IRQ #17
I haven't noticed something else curious on the servers.
hardware of these servers :
- bi Intel Xeon 2.4Ghz
- 4.5 GB of RAM
- MegaRaid SCSI 320-2 
The kernel use SMP, HT, high memory support 64GB, megaraid_mbox driver
(v22.214.171.124, in module with initrd until yesterday) and don't use preempt.
As I didn't find anything that match IRQ disable and megaraid under
internet, I have tried several ways :
- at the beginning, the IRQ of eth0 and megaraid was shared. I have
corrected this  but the problem is staying.
- I have try newer version of the kernel 2.6.9 to 2.6.10. No benefic
- I have try the boot option "acpi=ht". No benefic result.
- the firmeware of the megaraid has been upgraded (to the 1L37
version). No benefic result.
Since yesterday, I am trying, on all server which have the problem,
kernel 126.96.36.199 (source take kernel.org) with megaraid_mbox built-in
(v188.8.131.52) and with differents boot options on servers :
- "noirqdebug" and "acpi=ht"
- "noirqdebug" and "acpi=off"
I have this problem since I installed them few weeks ago with my new
FAI (Debian Fully autmatic Installation). I have a lot of difficulty for
solving the probleme because servers can run without problem one or two
weeks. These servers are not yet in production but still in tests of
charge; they should be in production in one or two weeks... if I find a
way to correct this problem !
Do you have any ideas of which way I can search ?
 0000:03:08.0 RAID bus controller: LSI Logic / Symbios Logic MegaRAID
Subsystem: LSI Logic / Symbios Logic MegaRAID 518 SCSI 320-2
Flags: bus master, fast Back2Back, 66MHz, slow devsel, latency
32, IRQ 17
Memory at d0500000 (32-bit, prefetchable) [size=64K]
Capabilities:  Power Management version 2
irq 0: 88018734 timer irq 16: 21659113 eth0
irq 1: 9 i8042 irq 17: 12532775 megaraid
irq 2: 0 cascade  irq 18: 30 aic79xx
irq 12: 3 irq 19: 30 aic79xx
irq 14: 1 ide0 irq 23: 0 ehci_hcd
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/