2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9

2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9

[Manuel Lauss]

Hello,

Latest git, insert a CF card into cardbus slot gives:

pccard: PCMCIA card inserted into slot 0
cs: memory probe 0x0c0000-0x0fffff: excluding 0xc0000-0xcffff 0xe0000-0xfffff
cs: memory probe 0x60000000-0x60ffffff: clean.
cs: memory probe 0xa0000000-0xa0ffffff: clean.
cs: memory probe 0xdea00000-0xdeafffff: excluding 0xdea00000-0xdeafffff
cs: memory probe 0xff600000-0xff6fffff: excluding 0xff600000-0xff60ffff 0xff6a0000-0xff6dffff 0xff6f0000-0xff6fffff
pcmcia: registering new device pcmcia0.0
irq 9: nobody cared (try booting with the "irqpoll" option)
  [<c01041f2>] show_trace_log_lvl+0x1a/0x2f
  [<c01042ed>] show_trace+0x12/0x14
  [<c0104ea0>] dump_stack+0x15/0x17
  [<c0133b48>] __report_bad_irq+0x39/0x79
  [<c0133d80>] note_interrupt+0x1f8/0x234
  [<c0134527>] handle_level_irq+0x8e/0xd1
  [<c0105a6d>] do_IRQ+0x92/0xad
  [<c0103f57>] common_interrupt+0x23/0x28
  [<c0134513>] handle_level_irq+0x7a/0xd1
  [<c0105a5e>] do_IRQ+0x83/0xad
  =======================
handlers:
[<c02a8a38>] (acpi_irq+0x0/0x18)
[<c03268e0>] (yenta_interrupt+0x0/0xb7)
[<c032b1d7>] (usb_hcd_irq+0x0/0x51)
[<c032b1d7>] (usb_hcd_irq+0x0/0x51)
[<c032b1d7>] (usb_hcd_irq+0x0/0x51)
[<c037c08c>] (snd_intel8x0_interrupt+0x0/0x1e8)
[<c02eeaff>] (e1000_intr+0x0/0xeb)
Disabling IRQ #9
scsi2 : pata_pcmcia
ata3: PATA max PIO0 cmd 0x0001d100 ctl 0x0001d10e bmdma 0x00000000 irq 9
ata3.00: qc timeout (cmd 0x91)
ata3.00: failed to IDENTIFY (INIT_DEV_PARAMS failed, err_mask=0x4)

However, if I rename pata_pcmcia.ko so that udev cannot autoload it, insert the
card and manually load pata_pcmcia.ko it works:

pccard: PCMCIA card inserted into slot 0
cs: memory probe 0x0c0000-0x0fffff: excluding 0xc0000-0xcffff 0xe0000-0xfffff
cs: memory probe 0x60000000-0x60ffffff: clean.
cs: memory probe 0xa0000000-0xa0ffffff: clean.
cs: memory probe 0xdea00000-0xdeafffff: excluding 0xdea00000-0xdeafffff
cs: memory probe 0xff600000-0xff6fffff: excluding 0xff600000-0xff60ffff 0xff6a0000-0xff6dffff 0xff6f0000-0xff6fffff
pcmcia: registering new device pcmcia0.0

# insmod pata_pcmcia.ko

scsi2 : pata_pcmcia
ata3: PATA max PIO0 cmd 0x0001d100 ctl 0x0001d10e bmdma 0x00000000 irq 9
ata3.00: CFA: , verB2.10, max PIO2
ata3.00: 507904 sectors, multi 0: LBA
ata3.00: configured for PIO0
scsi 2:0:0:0: Direct-Access     ATA                       verB PQ: 0 ANSI: 5
sd 2:0:0:0: [sdb] 507904 512-byte hardware sectors (260 MB)
sd 2:0:0:0: [sdb] Write Protect is off
sd 2:0:0:0: [sdb] Mode Sense: 00 3a 00 00
sd 2:0:0:0: [sdb] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
sd 2:0:0:0: [sdb] 507904 512-byte hardware sectors (260 MB)
sd 2:0:0:0: [sdb] Write Protect is off
sd 2:0:0:0: [sdb] Mode Sense: 00 3a 00 00
sd 2:0:0:0: [sdb] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
  sdb: sdb1
sd 2:0:0:0: [sdb] Attached SCSI removable disk
sd 2:0:0:0: Attached scsi generic sg2 type 0


System is a Vaio with ICH4 and TI PCI7420 Cardbus bridge.
If you need more info, please don't hesitate to ask.

Thank you,
	Manuel Lauss

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Peter F. Patel-Schneider]

Hi:

There have been several threads related to continuing problems with CF 
flash memory card in PCMCIA adapters not being recognized and leading to 
traces like:
Mar 20 19:51:44 idefix kernel: pccard: PCMCIA card inserted into slot 0
Mar 20 19:51:44 idefix kernel: pcmcia: registering new device pcmcia0.0
Mar 20 19:51:45 idefix kernel: scsi17 : pata_pcmcia
Mar 20 19:51:45 idefix kernel: ata12: PATA max PIO0 cmd 0xa100 ctl 
0xa10e irq 3
Mar 20 19:52:15 idefix kernel: ata12.00: qc timeout (cmd 0x91)
Mar 20 19:52:15 idefix kernel: ata12.00: failed to IDENTIFY 
(INIT_DEV_PARAMS failed, err_mask=0x4)
Mar 20 19:52:15 idefix kernel: ata12: failed to recover some devices, 
retrying in 5 secs
Mar 20 19:52:50 idefix kernel: ata12.00: qc timeout (cmd 0x91)
Mar 20 19:52:50 idefix kernel: ata12.00: failed to IDENTIFY 
(INIT_DEV_PARAMS failed, err_mask=0x4)
Mar 20 19:52:50 idefix kernel: ata12: failed to recover some devices, 
retrying in 5 secs

I, too, am experiencing this problem (and have been for about a year).  
Currently I'm running
2.6.24.3-34.fc8 #1 SMP on a Thinkpad T61p, but the problem has surfaced 
in all recent (to me) kernels.  I've tried with two different PCMCIA CF 
adapters and quite a few CF cards - all exhibit the same broken behaviour.

I can't find a resolution.  If there isn't one I'm willing to donate one 
of my adapters and a CF card so that anyone with a PCMCIA slot can test 
out the situation.

By the way, the problem occurs for me on insertion into a running 
system.  If I boot with the PCMCIA adapter and CF card in, I either get 
a freeze during udev startup (until I remove the card and then the 
system works fine) or I get the card mounted as a disk, and everything 
works fine.  I'm willing to provide more information about these two 
situations if anyone is interested.  The same CF card works fine when 
put into a USB multi-card reader as well.

Peter F. Patel-Schneider
Bell Labs Research

PS:  Please cc me directly if possible on replies, thanks.

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Alan Cox]

O> Mar 20 19:51:45 idefix kernel: ata12: PATA max PIO0 cmd 0xa100 ctl 
> 0xa10e irq 3
> Mar 20 19:52:15 idefix kernel: ata12.00: qc timeout (cmd 0x91)
> Mar 20 19:52:15 idefix kernel: ata12.00: failed to IDENTIFY 
> (INIT_DEV_PARAMS failed, err_mask=0x4)

Thinkpad T61p .. I've seen a few of these.

Firstly try the card and adapter in a different PC and let me know if
that works. So far they do. If that is the case get the latest firmware
updates (BIOS etc) for your thinkpad and see if the problem goes away.
They (or our ACPI code I guess it may be) have some serious IRQ routing
problems on this box.

Alan

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Tejun Heo]

Alan Cox wrote:
> O> Mar 20 19:51:45 idefix kernel: ata12: PATA max PIO0 cmd 0xa100 ctl 
>> 0xa10e irq 3
>> Mar 20 19:52:15 idefix kernel: ata12.00: qc timeout (cmd 0x91)
>> Mar 20 19:52:15 idefix kernel: ata12.00: failed to IDENTIFY 
>> (INIT_DEV_PARAMS failed, err_mask=0x4)
> 
> Thinkpad T61p .. I've seen a few of these.
> 
> Firstly try the card and adapter in a different PC and let me know if
> that works. So far they do. If that is the case get the latest firmware
> updates (BIOS etc) for your thinkpad and see if the problem goes away.
> They (or our ACPI code I guess it may be) have some serious IRQ routing
> problems on this box.

Any chance the thing just chokes up on INIT_DEV_PARAMS?

-- 
tejun

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Alan Cox]

On Fri, 21 Mar 2008 10:38:31 +0900
Tejun Heo <htejun@gmail.com> wrote:

> Alan Cox wrote:
> > O> Mar 20 19:51:45 idefix kernel: ata12: PATA max PIO0 cmd 0xa100 ctl 
> >> 0xa10e irq 3
> >> Mar 20 19:52:15 idefix kernel: ata12.00: qc timeout (cmd 0x91)
> >> Mar 20 19:52:15 idefix kernel: ata12.00: failed to IDENTIFY 
> >> (INIT_DEV_PARAMS failed, err_mask=0x4)
> > 
> > Thinkpad T61p .. I've seen a few of these.
> > 
> > Firstly try the card and adapter in a different PC and let me know if
> > that works. So far they do. If that is the case get the latest firmware
> > updates (BIOS etc) for your thinkpad and see if the problem goes away.
> > They (or our ACPI code I guess it may be) have some serious IRQ routing
> > problems on this box.
> 
> Any chance the thing just chokes up on INIT_DEV_PARAMS?

Possibly but I've seen several previosu reports showing no IRQ being
delivered off these machines. You get IRQ 3 reported as assigned and an
IRQ 9 turns up.

Alan

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Peter F. Patel-Schneider]

Alan Cox wrote:
> On Fri, 21 Mar 2008 10:38:31 +0900
> Tejun Heo <htejun@gmail.com> wrote:
>
>   
>> Alan Cox wrote:
>>     
>>> O> Mar 20 19:51:45 idefix kernel: ata12: PATA max PIO0 cmd 0xa100 ctl 
>>>       
>>>> 0xa10e irq 3
>>>> Mar 20 19:52:15 idefix kernel: ata12.00: qc timeout (cmd 0x91)
>>>> Mar 20 19:52:15 idefix kernel: ata12.00: failed to IDENTIFY 
>>>> (INIT_DEV_PARAMS failed, err_mask=0x4)
>>>>         
>>> Thinkpad T61p .. I've seen a few of these.
>>>
>>> Firstly try the card and adapter in a different PC and let me know if
>>> that works. So far they do. If that is the case get the latest firmware
>>> updates (BIOS etc) for your thinkpad and see if the problem goes away.
>>> They (or our ACPI code I guess it may be) have some serious IRQ routing
>>> problems on this box.
>>>       
>> Any chance the thing just chokes up on INIT_DEV_PARAMS?
>>     
>
> Possibly but I've seen several previosu reports showing no IRQ being
> delivered off these machines. You get IRQ 3 reported as assigned and an
> IRQ 9 turns up.
>
> Alan
>
>   
How can I check to see if this is the case?  How would this explain that 
the adapter and card sometimes work when they are in when the system boots?

peter

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Alan Cox]

 
> How can I check to see if this is the case?  How would this explain that 

As I said before, take a CF card/adapter combination you know fails
on a T60p/T61p and test it on a Linux box that isn't one of those types
(and preferably not a Lenovo at all)

> the adapter and card sometimes work when they are in when the system boots?

Ask the BIOS authors. Trying to fathom what they have done is frequently
interesting. Everyone else I've had do this has found the card works
perfectly with Linux except in the Lenovo box.

Alan

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Peter F. Patel-Schneider]

From: Alan Cox <alan@lxorguk.ukuu.org.uk>
Subject: Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure
Date: Fri, 21 Mar 2008 11:13:13 +0000

>  
> > How can I check to see if this is the case?  How would this explain that 
> 
> As I said before, take a CF card/adapter combination you know fails
> on a T60p/T61p and test it on a Linux box that isn't one of those types
> (and preferably not a Lenovo at all)

Actually I was wondering if there was something that I could do on the
box itself.

> > the adapter and card sometimes work when they are in when the system boots?
> 
> Ask the BIOS authors. Trying to fathom what they have done is frequently
> interesting. Everyone else I've had do this has found the card works
> perfectly with Linux except in the Lenovo box.

I'll investigate further.

> Alan

peter

Reporting ThinkPad BIOS bugs

[Henrique de Moraes Holschuh]

On Fri, 21 Mar 2008, Alan Cox wrote:
> > How can I check to see if this is the case?  How would this explain that 
> 
> As I said before, take a CF card/adapter combination you know fails
> on a T60p/T61p and test it on a Linux box that isn't one of those types
> (and preferably not a Lenovo at all)
> 
> > the adapter and card sometimes work when they are in when the system boots?
> 
> Ask the BIOS authors. Trying to fathom what they have done is frequently

I can actually try to do just that.  So can Thomas Renninger, added to cc.

Please someone send either of us a *detailed*, ready for "upstream
submission" bug report of the thinkpad BIOS IRQ routing issues, and we might
be able to get it to someone who can fix it for good.

Be clear, concise, and remember that whomever will read it has no obligation
to know much about Linux kernel internals, so explain things from a
*hardware* and ACPI model (as in what is in the ACPI spec, not what ACPICA
code does) point of view.

Do attach the trimmed kernel logs showing the bugs and ACPI/IOAPIC init.
Even if the person reading the report doesn't know Linux well, they can be
helpful.  And if he does know Linux well...

The report should be done with the latest available BIOS when possible, and
include the dmidecode BIOS information (version, date), as well as the
dmidecode System Information/Product Name (i.e. the numeric model number).

Reports about non-Lenovo ThinkPad BIOSes are unlikely to be acted upon.  I
have made a bunch of those, but they couldn't do much about it.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Re: Reporting ThinkPad BIOS bugs

[Peter F. Patel-Schneider]

Great.

I'm willing to do what is needed to get this moving along, if someone
tells me what is needed.

It does appear that the problem is specific to particular machines, as
my Lenovo T60p running BIOS version 2.21 of 02/05/2008 (up to date as of
yesterday) has the problem when running various kernels (including all
Fedora 8, currently 2.6.24.3-34.fc8, and Fedora 9 alpha kernels,
currently some 2.6.25 kernel).  However, everything works fine on a
Compac EVO N620c running from the Fedora 8 Live CD.

I have standard kernel logs with the behaviors, including a full bootup
log (when the device works).  I just got the dmidecode information.  

If anyone wants, I can send you this information (about 900 lines), or I
can get other information.  I also can (with a bit of difficulty
probably get information from a T60 and an X60).

peter


From: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Subject: Reporting ThinkPad BIOS bugs
Date: Fri, 21 Mar 2008 09:51:35 -0300

> On Fri, 21 Mar 2008, Alan Cox wrote:
> > > How can I check to see if this is the case?  How would this explain that 
> > 
> > As I said before, take a CF card/adapter combination you know fails
> > on a T60p/T61p and test it on a Linux box that isn't one of those types
> > (and preferably not a Lenovo at all)
> > 
> > > the adapter and card sometimes work when they are in when the system boots?
> > 
> > Ask the BIOS authors. Trying to fathom what they have done is frequently
> 
> I can actually try to do just that.  So can Thomas Renninger, added to cc.
> 
> Please someone send either of us a *detailed*, ready for "upstream
> submission" bug report of the thinkpad BIOS IRQ routing issues, and we might
> be able to get it to someone who can fix it for good.
> 
> Be clear, concise, and remember that whomever will read it has no obligation
> to know much about Linux kernel internals, so explain things from a
> *hardware* and ACPI model (as in what is in the ACPI spec, not what ACPICA
> code does) point of view.
> 
> Do attach the trimmed kernel logs showing the bugs and ACPI/IOAPIC init.
> Even if the person reading the report doesn't know Linux well, they can be
> helpful.  And if he does know Linux well...
> 
> The report should be done with the latest available BIOS when possible, and
> include the dmidecode BIOS information (version, date), as well as the
> dmidecode System Information/Product Name (i.e. the numeric model number).
> 
> Reports about non-Lenovo ThinkPad BIOSes are unlikely to be acted upon.  I
> have made a bunch of those, but they couldn't do much about it.

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Peter F. Patel-Schneider]

Alan Cox wrote:
> On Fri, 21 Mar 2008 10:38:31 +0900
> Tejun Heo <htejun@gmail.com> wrote:
>
>   
>> Alan Cox wrote:
>>     
>>> O> Mar 20 19:51:45 idefix kernel: ata12: PATA max PIO0 cmd 0xa100 ctl 
>>>       
>>>> 0xa10e irq 3
>>>> Mar 20 19:52:15 idefix kernel: ata12.00: qc timeout (cmd 0x91)
>>>> Mar 20 19:52:15 idefix kernel: ata12.00: failed to IDENTIFY 
>>>> (INIT_DEV_PARAMS failed, err_mask=0x4)
>>>>         
>>> Thinkpad T61p .. I've seen a few of these.
>>>
>>> Firstly try the card and adapter in a different PC and let me know if
>>> that works. So far they do. If that is the case get the latest firmware
>>> updates (BIOS etc) for your thinkpad and see if the problem goes away.
>>> They (or our ACPI code I guess it may be) have some serious IRQ routing
>>> problems on this box.
>>>       
>> Any chance the thing just chokes up on INIT_DEV_PARAMS?
>>     
>
> Possibly but I've seen several previosu reports showing no IRQ being
> delivered off these machines. You get IRQ 3 reported as assigned and an
> IRQ 9 turns up.
>
> Alan
>
>   

Also, these devices used to work about a year ago (same machine, earlier 
kernel) and then stopped working. 

It is entirely possible that some change triggered a BIOS(?) bug, or 
even that a BIOS upgrade caused the breakage, of course.

I'll try to try out the hardware in another Linux box - it may take a 
while, as I have to find out who at work has liberated their machine 
from corporate IT.  :-)

peter

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Alan Cox]

> It is entirely possible that some change triggered a BIOS(?) bug, or 
> even that a BIOS upgrade caused the breakage, of course.

Yes, and if your testing shows what I think it will then the bug can
instead become an ACPI bug and maybe Len and the ACPI folks can work
around it.
> 
> I'll try to try out the hardware in another Linux box - it may take a 
> while, as I have to find out who at work has liberated their machine 
> from corporate IT.  :-)

Thanks
Alan

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Peter F. Patel-Schneider]

Alan Cox wrote:

 >> Mar 20 19:51:45 idefix kernel: ata12: PATA max PIO0 cmd 0xa100 ctl 
0xa10e irq 3
 >> Mar 20 19:52:15 idefix kernel: ata12.00: qc timeout (cmd 0x91)
 >> Mar 20 19:52:15 idefix kernel: ata12.00: failed to IDENTIFY 
(INIT_DEV_PARAMS failed, err_mask=0x4)
 >
 > Thinkpad T61p .. I've seen a few of these.

(Actually a T60p. I guess I'm getting envious.)

 > Firstly try the card and adapter in a different PC and let me know if
 > that works. So far they do. If that is the case get the latest firmware
 > updates (BIOS etc) for your thinkpad and see if the problem goes away.
 > They (or our ACPI code I guess it may be) have some serious IRQ routing
 > problems on this box.
 >
 > Alan

The card and adapter (right now a Kodak picture card adapter and an old
64MB Kodak CF card, but, as I said, the behaviour is the same with
different adapters and different CF cards) work correctly in my Wife's
Evo running Windows.   The adapter and card work, sometimes, when they
are in the machine at boot (like on my last boot - I have a 
/var/log/messages
with this behaviour, but it isn't very illuminating).   I have had 
several T60p's,
they all act the same.  (The on-site service center swaps machines for us
when we have problems.)

I just updated to the current BIOS (2.21) and ECP (1.07).  Not working.
(Just for the record the kernel is 2.6.24.3-34.fc8 #1 SMP.)

Here are the /var/log/messages after inserting the card and adapter.

Mar 20 22:13:19 idefix kernel: pccard: PCMCIA card inserted into slot 0
Mar 20 22:13:19 idefix kernel: cs: memory probe 0xe4300000-0xe7ffffff: 
excluding 0xe4300000-0xe46cffff 0xe4e70000-0xe523ffff 
0xe5db0000-0xe617ffff 0xe6cf0000-0xe70bffff
Mar 20 22:13:19 idefix kernel: pcmcia: registering new device pcmcia0.0
Mar 20 22:13:19 idefix kernel: scsi6 : pata_pcmcia
Mar 20 22:13:19 idefix kernel: ata7: PATA max PIO0 cmd 0xa100 ctl 0xa10e 
irq 3
Mar 20 22:13:49 idefix kernel: ata7.00: qc timeout (cmd 0x91)
Mar 20 22:13:49 idefix kernel: ata7.00: failed to IDENTIFY 
(INIT_DEV_PARAMS failed, err_mask=0x4)
Mar 20 22:13:49 idefix kernel: ata7: failed to recover some devices, 
retrying in 5 secs
Mar 20 22:14:24 idefix kernel: ata7.00: qc timeout (cmd 0x91)
Mar 20 22:14:24 idefix kernel: ata7.00: failed to IDENTIFY 
(INIT_DEV_PARAMS failed, err_mask=0x4)
Mar 20 22:14:24 idefix kernel: ata7: failed to recover some devices, 
retrying in 5 secs
Mar 20 22:14:59 idefix kernel: ata7.00: qc timeout (cmd 0x91)
Mar 20 22:14:59 idefix kernel: ata7.00: failed to IDENTIFY 
(INIT_DEV_PARAMS failed, err_mask=0x4)
Mar 20 22:14:59 idefix kernel: ata7: failed to recover some devices, 
retrying in 5 secs

peter

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Alan Cox]

> 64MB Kodak CF card, but, as I said, the behaviour is the same with
> different adapters and different CF cards) work correctly in my Wife's
> Evo running Windows.   The adapter and card work, sometimes, when they

Try them in a different Linux system and I bet they work there too.

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Peter F. Patel-Schneider]

Alan Cox wrote:
> O> Mar 20 19:51:45 idefix kernel: ata12: PATA max PIO0 cmd 0xa100 ctl 
>   
>> 0xa10e irq 3
>> Mar 20 19:52:15 idefix kernel: ata12.00: qc timeout (cmd 0x91)
>> Mar 20 19:52:15 idefix kernel: ata12.00: failed to IDENTIFY 
>> (INIT_DEV_PARAMS failed, err_mask=0x4)
>>     
>
> Thinkpad T61p .. I've seen a few of these.
>
> Firstly try the card and adapter in a different PC and let me know if
> that works. So far they do. If that is the case get the latest firmware
> updates (BIOS etc) for your thinkpad and see if the problem goes away.
> They (or our ACPI code I guess it may be) have some serious IRQ routing
> problems on this box.
>
> Alan
>
>   
So I tried again just now (on the same T60p running the current Fedora 
8),  and got different behaviour!

Mar 26 10:06:33 idefix kernel: pccard: PCMCIA card inserted into slot 0
Mar 26 10:06:33 idefix kernel: cs: memory probe 0xe4300000-0xe7ffffff: 
excluding 0xe4300000-0xe46cffff 0xe4e70000-0xe523ffff 
0xe5db0000-0xe617ffff 0xe6cf0000-0xe70bffff
Mar 26 10:06:33 idefix kernel: pcmcia: registering new device pcmcia0.0
Mar 26 10:06:34 idefix kernel: scsi11 : pata_pcmcia
Mar 26 10:06:34 idefix kernel: ata7: PATA max PIO0 cmd 0xa100 ctl 0xa10e 
irq 3
Mar 26 10:06:34 idefix kernel: ata7.00: CFA: SanDisk SDCFH-512, 
FhCA0145, max PIO4
Mar 26 10:06:34 idefix kernel: ata7.00: 1000944 sectors, multi 0: LBA
Mar 26 10:06:34 idefix kernel: ata7.00: configured for PIO0
Mar 26 10:06:34 idefix kernel: isa bounce pool size: 16 pages
Mar 26 10:06:34 idefix kernel: scsi 11:0:0:0: Direct-Access     ATA      
SanDisk SDCFH-51 FhCA PQ: 0 ANSI: 5
Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] 1000944 512-byte 
hardware sectors (512 MB)
Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] Write Protect is off
Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] Write cache: disabled, 
read cache: enabled, doesn't support DPO or FUA
Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] 1000944 512-byte 
hardware sectors (512 MB)
Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] Write Protect is off
Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] Write cache: disabled, 
read cache: enabled, doesn't support DPO or FUA
Mar 26 10:07:04 idefix kernel:  sdc:<3>ata7.00: exception Emask 0x0 SAct 
0x0 SErr 0x0 action 0x2 frozen
Mar 26 10:07:04 idefix kernel: ata7.00: cmd 
20/00:08:00:00:00/00:00:00:00:00/e0 tag 0 pio 4096 in
Mar 26 10:07:04 idefix kernel:          res 
40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
Mar 26 10:07:04 idefix kernel: ata7.00: status: { DRDY }
Mar 26 10:07:04 idefix kernel: ata7: soft resetting link
Mar 26 10:07:04 idefix kernel: ata7.00: configured for PIO0
Mar 26 10:07:04 idefix kernel: ata7: EH complete
Mar 26 10:07:34 idefix kernel: ata7.00: exception Emask 0x0 SAct 0x0 
SErr 0x0 action 0x2 frozen
Mar 26 10:07:34 idefix kernel: ata7.00: cmd 
20/00:08:00:00:00/00:00:00:00:00/e0 tag 0 pio 4096 in
Mar 26 10:07:34 idefix kernel:          res 
40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
Mar 26 10:07:34 idefix kernel: ata7.00: status: { DRDY }
Mar 26 10:07:34 idefix kernel: ata7: soft resetting link
Mar 26 10:07:34 idefix kernel: ata7.00: configured for PIO0
Mar 26 10:07:34 idefix kernel: ata7: EH complete
Mar 26 10:08:04 idefix kernel: ata7.00: exception Emask 0x0 SAct 0x0 
SErr 0x0 action 0x2 frozen
Mar 26 10:08:04 idefix kernel: ata7.00: cmd 
20/00:08:00:00:00/00:00:00:00:00/e0 tag 0 pio 4096 in
Mar 26 10:08:04 idefix kernel:          res 
40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
Mar 26 10:08:04 idefix kernel: ata7.00: status: { DRDY }
Mar 26 10:08:04 idefix kernel: ata7: soft resetting link
Mar 26 10:08:04 idefix kernel: ata7.00: configured for PIO0
Mar 26 10:08:04 idefix kernel: ata7: EH complete

Any ideas on where this "report" should go?

Peter F. Patel-Schneider

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Tejun Heo]

Peter F. Patel-Schneider wrote:
> Alan Cox wrote:
>> O> Mar 20 19:51:45 idefix kernel: ata12: PATA max PIO0 cmd 0xa100 ctl  
>>> 0xa10e irq 3
>>> Mar 20 19:52:15 idefix kernel: ata12.00: qc timeout (cmd 0x91)
>>> Mar 20 19:52:15 idefix kernel: ata12.00: failed to IDENTIFY
>>> (INIT_DEV_PARAMS failed, err_mask=0x4)
>>>     
>>
>> Thinkpad T61p .. I've seen a few of these.
>>
>> Firstly try the card and adapter in a different PC and let me know if
>> that works. So far they do. If that is the case get the latest firmware
>> updates (BIOS etc) for your thinkpad and see if the problem goes away.
>> They (or our ACPI code I guess it may be) have some serious IRQ routing
>> problems on this box.
>>
>> Alan
>>
>>   
> So I tried again just now (on the same T60p running the current Fedora
> 8),  and got different behaviour!
> 
> Mar 26 10:06:33 idefix kernel: pccard: PCMCIA card inserted into slot 0
> Mar 26 10:06:33 idefix kernel: cs: memory probe 0xe4300000-0xe7ffffff:
> excluding 0xe4300000-0xe46cffff 0xe4e70000-0xe523ffff
> 0xe5db0000-0xe617ffff 0xe6cf0000-0xe70bffff
> Mar 26 10:06:33 idefix kernel: pcmcia: registering new device pcmcia0.0
> Mar 26 10:06:34 idefix kernel: scsi11 : pata_pcmcia
> Mar 26 10:06:34 idefix kernel: ata7: PATA max PIO0 cmd 0xa100 ctl 0xa10e
> irq 3
> Mar 26 10:06:34 idefix kernel: ata7.00: CFA: SanDisk SDCFH-512,
> FhCA0145, max PIO4
> Mar 26 10:06:34 idefix kernel: ata7.00: 1000944 sectors, multi 0: LBA
> Mar 26 10:06:34 idefix kernel: ata7.00: configured for PIO0
> Mar 26 10:06:34 idefix kernel: isa bounce pool size: 16 pages
> Mar 26 10:06:34 idefix kernel: scsi 11:0:0:0: Direct-Access     ATA     
> SanDisk SDCFH-51 FhCA PQ: 0 ANSI: 5
> Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] 1000944 512-byte
> hardware sectors (512 MB)
> Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] Write Protect is off
> Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] Write cache: disabled,
> read cache: enabled, doesn't support DPO or FUA
> Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] 1000944 512-byte
> hardware sectors (512 MB)
> Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] Write Protect is off
> Mar 26 10:06:34 idefix kernel: sd 11:0:0:0: [sdc] Write cache: disabled,
> read cache: enabled, doesn't support DPO or FUA
> Mar 26 10:07:04 idefix kernel:  sdc:<3>ata7.00: exception Emask 0x0 SAct
> 0x0 SErr 0x0 action 0x2 frozen
> Mar 26 10:07:04 idefix kernel: ata7.00: cmd
> 20/00:08:00:00:00/00:00:00:00:00/e0 tag 0 pio 4096 in
> Mar 26 10:07:04 idefix kernel:          res
> 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
> Mar 26 10:07:04 idefix kernel: ata7.00: status: { DRDY }
> Mar 26 10:07:04 idefix kernel: ata7: soft resetting link
> Mar 26 10:07:04 idefix kernel: ata7.00: configured for PIO0
> Mar 26 10:07:04 idefix kernel: ata7: EH complete
> Mar 26 10:07:34 idefix kernel: ata7.00: exception Emask 0x0 SAct 0x0
> SErr 0x0 action 0x2 frozen
> Mar 26 10:07:34 idefix kernel: ata7.00: cmd
> 20/00:08:00:00:00/00:00:00:00:00/e0 tag 0 pio 4096 in
> Mar 26 10:07:34 idefix kernel:          res
> 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
> Mar 26 10:07:34 idefix kernel: ata7.00: status: { DRDY }
> Mar 26 10:07:34 idefix kernel: ata7: soft resetting link
> Mar 26 10:07:34 idefix kernel: ata7.00: configured for PIO0
> Mar 26 10:07:34 idefix kernel: ata7: EH complete
> Mar 26 10:08:04 idefix kernel: ata7.00: exception Emask 0x0 SAct 0x0
> SErr 0x0 action 0x2 frozen
> Mar 26 10:08:04 idefix kernel: ata7.00: cmd
> 20/00:08:00:00:00/00:00:00:00:00/e0 tag 0 pio 4096 in
> Mar 26 10:08:04 idefix kernel:          res
> 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
> Mar 26 10:08:04 idefix kernel: ata7.00: status: { DRDY }
> Mar 26 10:08:04 idefix kernel: ata7: soft resetting link
> Mar 26 10:08:04 idefix kernel: ata7.00: configured for PIO0
> Mar 26 10:08:04 idefix kernel: ata7: EH complete
> 
> Any ideas on where this "report" should go?

Does irqpoll help?

-- 
tejun

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Peter F. Patel-Schneider]

From: Tejun Heo <htejun@gmail.com>
Subject: Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure
Date: Wed, 26 Mar 2008 23:44:39 +0900

[Problems with PCMCIA adapter and CF card on ThinkPad T60p.]

> Does irqpoll help?

Yes!

Booting with the irqpoll boot option results in correct behaviour even
with repeated insertions and removals (with or without unmounting from
desktop) and when booting with or without the adapter and card inserted.
Much better (but slower).

Booting without the irqpoll boot option and no adapter and card in at
boot results in the bad behaviour when the adapter and card are first
inserted.  Booting without the irqpoll boot option and the adapter and
card in at boot results in more-or-less correct behaviour (there is an
extra icon on the desktop for the media, but this is a known problem and
probably not related), although in the past the behaviour in this
situation has been somewhat non-deterministic.

Problem solved?  Or not?

> -- 
> tejun

peter

Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure

[Tejun Heo]

Hello,

Peter F. Patel-Schneider wrote:
> From: Tejun Heo <htejun@gmail.com>
> Subject: Re: 2.6.23-rc9-git4: pata_pcmcia, disabling IRQ #9 and Re: libata and legacy ide pcmcia failure
> Date: Wed, 26 Mar 2008 23:44:39 +0900
> 
> [Problems with PCMCIA adapter and CF card on ThinkPad T60p.]
> 
>> Does irqpoll help?
> 
> Yes!
> 
> Booting with the irqpoll boot option results in correct behaviour even
> with repeated insertions and removals (with or without unmounting from
> desktop) and when booting with or without the adapter and card inserted.
> Much better (but slower).
> 
> Booting without the irqpoll boot option and no adapter and card in at
> boot results in the bad behaviour when the adapter and card are first
> inserted.  Booting without the irqpoll boot option and the adapter and
> card in at boot results in more-or-less correct behaviour (there is an
> extra icon on the desktop for the media, but this is a known problem and
> probably not related), although in the past the behaviour in this
> situation has been somewhat non-deterministic.
> 
> Problem solved?  Or not?

Well, we're getting somewhere but not really there yet.  Somehow IRQ 
routing didn't get setup right.  I don't have much idea on how to track 
it down.  Cc'ing linux-pci and linux-pcmcia.

Hello, guys, on tp60, IRQ on PCMCIA CF controller isn't working.  I 
don't know whether this is because IRQ isn't routed properly or someone 
didn't wire the ATA interrupt line.  Any ideas?

Thanks.

-- 
tejun