AIC7xxx in 2.4.19-pre8?

AIC7xxx in 2.4.19-pre8?

[Dana Lacoste]

(Kernel version is 2.4.19-pre8.  Let me know if you want more info :)

(Motherboard is Nexcom Peak 632 : 440BX with Adaptec 7890U2)

I'm getting (over and over and over and over) the message from line
1854 of drivers/scsi/aic7xxx/aic7xxx_pci.c :

scsi0: PCI error Interrupt at seqaddr = 0x8
scsi0: PCI error Interrupt at seqaddr = 0x9

(0x9 is more common, no pattern detected between 0x8 and 0x9)

It's interlaced with :
scsi0: Received a Target Abort

which means the status1 flag RTA was set, but I'm now officially out
of my league :)

This box was going fine till this morning (we upgraded from 2.4.18
last night) and a second box we did the same thing to also had the
same problem, so it looks like a kernel issue somehow, not just
faulty hardware.  Retrograding to 2.4.18 caused the problem to go
away.

Has anyone else seen this?
(or if you have suggestions, please let me know!)
 
Dana Lacoste
Ottawa, Canada

Re: AIC7xxx in 2.4.19-pre8?

[Justin T. Gibbs]

>(Kernel version is 2.4.19-pre8.  Let me know if you want more info :)
>
>(Motherboard is Nexcom Peak 632 : 440BX with Adaptec 7890U2)
>
>I'm getting (over and over and over and over) the message from line
>1854 of drivers/scsi/aic7xxx/aic7xxx_pci.c :
>
>scsi0: PCI error Interrupt at seqaddr = 0x8
>scsi0: PCI error Interrupt at seqaddr = 0x9
>
>(0x9 is more common, no pattern detected between 0x8 and 0x9)
>
>It's interlaced with :
>scsi0: Received a Target Abort
>
>which means the status1 flag RTA was set, but I'm now officially out
>of my league :)

Hmm.  The aic7xxx chip could have botched a master transaction somehow,
but that's still hard to say.  Can you have the driver dump out its
chip state (call ahc_dump_card_state(ahc) from within the pci interrupt
handler) to get some additional information?

>This box was going fine till this morning (we upgraded from 2.4.18
>last night) and a second box we did the same thing to also had the
>same problem, so it looks like a kernel issue somehow, not just
>faulty hardware.  Retrograding to 2.4.18 caused the problem to go
>away.

How much memory is in the box?  What version of the aic7xxx driver
were you using under 2.4.18?

--
Justin