crash upon rmmod aic7xxx (pcmcia)

crash upon rmmod aic7xxx (pcmcia)

[Kasper Peeters]

Hi,

Since the early 2.6.x kernels, removing the aic7xxx module (either by
doing a 'rmmod' by hand or by ejecting the pcmcia aic7xxx card) locks
the system hard after about 2 seconds, leaving no trace in syslog. 
Just checked with kernel 2.6.13 (and pcmcia-tools 3.2.8).

Upon insertion of the pcmcia card, this appears in syslog:

Sep 11 14:36:05 whiteroom2 kernel: PCI: Enabling device 0000:03:00.0 (0000 -> 0003)
Sep 11 14:36:05 whiteroom2 kernel: ACPI: PCI Interrupt 0000:03:00.0[A] -> Link [LNKA] -> GSI 11 (level, low
) -> IRQ 11
Sep 11 14:36:05 whiteroom2 kernel: aic7xxx: PCI Device 3:0:0 failed memory mapped test.  Using PIO.
Sep 11 14:36:05 whiteroom2 kernel: ahc_pci:3:0:0: Host Adapter Bios disabled.  Using default SCSI device pa
rameters
Sep 11 14:36:05 whiteroom2 kernel: scsi0 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.2.36
Sep 11 14:36:05 whiteroom2 kernel:         <Adaptec 1480A Ultra SCSI adapter>
Sep 11 14:36:05 whiteroom2 kernel:         aic7860: Ultra Single Channel A, SCSI Id=7, 3/253 SCBs
Sep 11 14:36:05 whiteroom2 kernel: 
Sep 11 14:36:12 whiteroom2 kernel: scsi0: PCI error Interrupt at seqaddr = 0x8
Sep 11 14:36:12 whiteroom2 kernel: scsi0: Signaled a Target Abort
Sep 11 14:36:21 whiteroom2 kernel:   Vendor: HP        Model: CD-Writer+ 9200   Rev: 1.0c
Sep 11 14:36:21 whiteroom2 kernel:   Type:   CD-ROM                             ANSI SCSI revision: 04
Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: asynchronous.
Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: Beginning Domain Validation
Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: Domain Validation skipping write tests
Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: FAST-10 SCSI 10.0 MB/s ST (100 ns, offset 15)
Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: Ending Domain Validation
Sep 11 14:36:21 whiteroom2 kernel: sr0: scsi3-mmc drive: 32x/32x writer cd/rw xa/form2 cdda tray
Sep 11 14:36:21 whiteroom2 kernel: Attached scsi generic sg0 at scsi0, channel 0, id 4, lun 0,  type 5
Sep 11 14:36:22 whiteroom2 scsi.agent[3014]: cdrom at /devices/pci0000:00/0000:00:1e.0/0000:02:0a.0/0000:03
:00.0/host0/target0:0:4/0:0:4:0


Can someone give me a hand pinning down this bug?

(Please CC replies to me directly).

Kasper

Re: crash upon rmmod aic7xxx (pcmcia)

[Daniel Drake]

Kasper Peeters wrote:
> Since the early 2.6.x kernels, removing the aic7xxx module (either by
> doing a 'rmmod' by hand or by ejecting the pcmcia aic7xxx card) locks
> the system hard after about 2 seconds, leaving no trace in syslog. 
> Just checked with kernel 2.6.13 (and pcmcia-tools 3.2.8).

Try enabling magic sysrq and press Alt+sysrq+9 then Alt+SysRq+P when the 
freeze occurs, and write down the call trace which hopefully appears on the 
console.

On a sidenote, this might be related to a Gentoo bug that I'm working on 
getting full details of:

http://bugs.gentoo.org/show_bug.cgi?id=102636

> Upon insertion of the pcmcia card, this appears in syslog:
> 
> Sep 11 14:36:05 whiteroom2 kernel: PCI: Enabling device 0000:03:00.0 (0000 -> 0003)
> Sep 11 14:36:05 whiteroom2 kernel: ACPI: PCI Interrupt 0000:03:00.0[A] -> Link [LNKA] -> GSI 11 (level, low
> ) -> IRQ 11
> Sep 11 14:36:05 whiteroom2 kernel: aic7xxx: PCI Device 3:0:0 failed memory mapped test.  Using PIO.
> Sep 11 14:36:05 whiteroom2 kernel: ahc_pci:3:0:0: Host Adapter Bios disabled.  Using default SCSI device pa
> rameters
> Sep 11 14:36:05 whiteroom2 kernel: scsi0 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.2.36
> Sep 11 14:36:05 whiteroom2 kernel:         <Adaptec 1480A Ultra SCSI adapter>
> Sep 11 14:36:05 whiteroom2 kernel:         aic7860: Ultra Single Channel A, SCSI Id=7, 3/253 SCBs
> Sep 11 14:36:05 whiteroom2 kernel: 
> Sep 11 14:36:12 whiteroom2 kernel: scsi0: PCI error Interrupt at seqaddr = 0x8
> Sep 11 14:36:12 whiteroom2 kernel: scsi0: Signaled a Target Abort
> Sep 11 14:36:21 whiteroom2 kernel:   Vendor: HP        Model: CD-Writer+ 9200   Rev: 1.0c
> Sep 11 14:36:21 whiteroom2 kernel:   Type:   CD-ROM                             ANSI SCSI revision: 04
> Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: asynchronous.
> Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: Beginning Domain Validation
> Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: Domain Validation skipping write tests
> Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: FAST-10 SCSI 10.0 MB/s ST (100 ns, offset 15)
> Sep 11 14:36:21 whiteroom2 kernel:  target0:0:4: Ending Domain Validation
> Sep 11 14:36:21 whiteroom2 kernel: sr0: scsi3-mmc drive: 32x/32x writer cd/rw xa/form2 cdda tray
> Sep 11 14:36:21 whiteroom2 kernel: Attached scsi generic sg0 at scsi0, channel 0, id 4, lun 0,  type 5
> Sep 11 14:36:22 whiteroom2 scsi.agent[3014]: cdrom at /devices/pci0000:00/0000:00:1e.0/0000:02:0a.0/0000:03
> :00.0/host0/target0:0:4/0:0:4:0

Re: crash upon rmmod aic7xxx (pcmcia)

[Daniel Drake]

(please use reply-to-all)

Kasper Peeters wrote:
>>Try enabling magic sysrq and press Alt+sysrq+9 then Alt+SysRq+P 
> 
> 
> I just spent an hour trying to make this work: it's turned on in the
> kernel, I have done
> 
>   echo '1' > /proc/sys/kernel/sysrq
> 
> , I have done a 'showkey -s' to find out that my sysrq generates
> '0x53', I have done 'setkeycodes 53 84', but none of the
> alt-sysrq-command things does anything (even before triggering the
> aic7xxx bug, that is). I must have overlooked something stupid.
> 
> Anyhow, for what it's worth, the 'rmmod' does spit out some more stuff
> to the console just before the crash (inasfar it has not yet scrolled
> off the screen):
> 
>   Sequencer Free SCB List 0 1 2
>   Sequencer SCB Info
>     0 SCB_CONTROL[0x0]:SCB_SCSIID[0x67]:SCB_LOW[0x0]:SCB_TAG[0xff]
>     1 SCB_CONTROL[0x0]:SCB_SCSIID[0xff]:(TWIN_CHNLB|OID|TWIN_TIB)SCB_LUN[0xff]:(SCB_XFERLEN_ODD|LID)SCB_TAG[0xff]
>     2 SCB_CONTROL[0x0]:SCB_SCSIID[0xff]:(TWIN_CHNLB|OID|TWIN_TIB)SCB_LUN[0xff]:(SCB_XFERLEN_ODD|LID)SCB_TAG[0xff]
>   Pending List:
>     2 SCB_CONTROL[0x40]:(DISCENB)SCB_SCSIID[0x47]:SCB_LUN[0x0]
>   Kernel Free SCB List: 1 0
>   Untagged Q(4):2
>   
>   <<< Dump Card State End >>>
>   qindex = 0, SCB index=0
>   Kernel Panic - not syncing: Loop 1
> 

Ok, if its a kernel panic then I'm fairly sure sysrq would be useless for 
debugging even if it worked on your setup. Also, this is the exact same 
message from the Gentoo bug that I mentioned previously: 
http://bugs.gentoo.org/102636

Perhaps you could file a bug for this at http://bugzilla.kernel.org and I'll 
get the Gentoo bug reporter to post his experience on your bug too.

Thanks,
Daniel

Re: crash upon rmmod aic7xxx (pcmcia)

[Kasper Peeters]

> Perhaps you could file a bug for this at http://bugzilla.kernel.org
> and I'll get the Gentoo bug reporter to post his experience on your
> bug too.

Ok, #5224.

Thanks for the help,
Kasper