Bug in pcmcia-core

Bug in pcmcia-core

[James Courtier-Dutton]

Hi,

I have tried conacting the mailing list for the PCMCIA subsystem in
Linux, but no-one seems to respond.

PCMCIA SUBSYSTEM
L:      http://lists.infradead.org/mailman/listinfo/linux-pcmcia
S:      Unmaintained

I am trying to write a Linux ALSA driver for the Creative Audigy 2 NX
Notebook PCMCIA card.
This is a cardbus card, that uses ioports.
When it is inserted into the laptop, the entry appears in "lspci -vv "
showing ioports used by the card.
As soon as my driver uses "outb()" to anything in the address range
shown in "lspci -vv" , the PC hangs.

I can only conclude from this that ioport resources are not being
allocated correctly to the PCMCIA card.

Can anybody help me track this down. If someone could tell me which
PCMCIA and PCI registers should be set for it to work, I could then find
out which pcmcia registers have not been set correctly, and fix the bug.

It seems that the PCMCIA specification is not open and free, so I cannot
refer to it in order to fix this myself.

Can anybody help me?

James

Re: Bug in pcmcia-core

[Wakko Warner]

James Courtier-Dutton wrote:
> I am trying to write a Linux ALSA driver for the Creative Audigy 2 NX
> Notebook PCMCIA card.
> This is a cardbus card, that uses ioports.
> When it is inserted into the laptop, the entry appears in "lspci -vv "
> showing ioports used by the card.
> As soon as my driver uses "outb()" to anything in the address range
> shown in "lspci -vv" , the PC hangs.
> 
> I can only conclude from this that ioport resources are not being
> allocated correctly to the PCMCIA card.
> 
> Can anybody help me track this down. If someone could tell me which
> PCMCIA and PCI registers should be set for it to work, I could then find
> out which pcmcia registers have not been set correctly, and fix the bug.
> 
> It seems that the PCMCIA specification is not open and free, so I cannot
> refer to it in order to fix this myself.

I thought drivers for the cardbus cards were the same as standard PCI cards. 
I know that as far as networking goes, the same driver runs a cardbus 3com
3c575 and the pci 3c905.  Same with netgear's cardbus FA510 and PCI FA310.

I'm not a kernel developer, but this is what I've understood.

-- 
 Lab tests show that use of micro$oft causes cancer in lab animals

Re: Bug in pcmcia-core

[James Courtier-Dutton]

Wakko Warner wrote:
> James Courtier-Dutton wrote:
> 
>>I am trying to write a Linux ALSA driver for the Creative Audigy 2 NX
>>Notebook PCMCIA card.
>>This is a cardbus card, that uses ioports.
>>When it is inserted into the laptop, the entry appears in "lspci -vv "
>>showing ioports used by the card.
>>As soon as my driver uses "outb()" to anything in the address range
>>shown in "lspci -vv" , the PC hangs.
>>
>>I can only conclude from this that ioport resources are not being
>>allocated correctly to the PCMCIA card.
>>
>>Can anybody help me track this down. If someone could tell me which
>>PCMCIA and PCI registers should be set for it to work, I could then find
>>out which pcmcia registers have not been set correctly, and fix the bug.
>>
>>It seems that the PCMCIA specification is not open and free, so I cannot
>>refer to it in order to fix this myself.
> 
> 
> I thought drivers for the cardbus cards were the same as standard PCI cards. 
> I know that as far as networking goes, the same driver runs a cardbus 3com
> 3c575 and the pci 3c905.  Same with netgear's cardbus FA510 and PCI FA310.
> 
> I'm not a kernel developer, but this is what I've understood.
> 

That is also what I thought. But I think that the cardbus 3com 3c575
uses memory for io and not ioports. I think the problem is related to
the use of ioports on an cardbus card.

Re: Bug in pcmcia-core

[Lee Revell]

On Fri, 2005-06-17 at 08:35 +0100, James Courtier-Dutton wrote:
> > 
> > I thought drivers for the cardbus cards were the same as standard PCI cards. 
> > I know that as far as networking goes, the same driver runs a cardbus 3com
> > 3c575 and the pci 3c905.  Same with netgear's cardbus FA510 and PCI FA310.
> > 
> > I'm not a kernel developer, but this is what I've understood.
> > 
> 
> That is also what I thought. But I think that the cardbus 3com 3c575
> uses memory for io and not ioports. I think the problem is related to
> the use of ioports on an cardbus card.

Any update on this?  Is the PCMCIA layer really unmaintained?  Are we
hosed?

A quick grep of the subjects in my LKML folder for "PCMCIA" is certainly
not encouraging.  I see lots of questions along the lines of "did ANYONE
EVER get this to work?" with no replies...

Lee

Re: Bug in pcmcia-core

[Dominik Brodowski]

Hi,

On Tue, Jun 21, 2005 at 11:43:14AM -0400, Lee Revell wrote:
> On Fri, 2005-06-17 at 08:35 +0100, James Courtier-Dutton wrote:
> > > I thought drivers for the cardbus cards were the same as standard PCI cards. 
> > > I know that as far as networking goes, the same driver runs a cardbus 3com
> > > 3c575 and the pci 3c905.  Same with netgear's cardbus FA510 and PCI FA310.
> > > 
> > > I'm not a kernel developer, but this is what I've understood.
> > > 
> > 
> > That is also what I thought. But I think that the cardbus 3com 3c575
> > uses memory for io and not ioports. I think the problem is related to
> > the use of ioports on an cardbus card.
> 
> Any update on this?  Is the PCMCIA layer really unmaintained?  Are we
> hosed?

Check the linux-pcmcia list, check -mm and you'll see that the PCMCIA layer
is in the process of being updated.

> A quick grep of the subjects in my LKML folder for "PCMCIA" is certainly
> not encouraging.  I see lots of questions along the lines of "did ANYONE
> EVER get this to work?" with no replies...

Well, some of them are handled in the background, some are simply unsolvable
(no real reason for hardware to behave the way it does, no driver support,
extremely strange setups, ...), some bugs are caused elsewhere (e.g., in
current -mm, PCI makes life difficult for PCMCIA), and some bugs are not
handled at all because the few developers taking care of PCMCIA do have too
few time. So, some help in maintaining PCMCIA is much appreciated.

	Dominik

Re: Bug in pcmcia-core

[Lee Revell]

On Tue, 2005-06-21 at 18:53 +0200, Dominik Brodowski wrote:
> Check the linux-pcmcia list, check -mm and you'll see that the PCMCIA layer
> is in the process of being updated.
> 

OK thanks for the info, sounds like we should try -mm.  It's certainly
possible that the hardware is buggy.

Lee

Re: Bug in pcmcia-core

[James Courtier-Dutton]

Lee Revell wrote:
> On Tue, 2005-06-21 at 18:53 +0200, Dominik Brodowski wrote:
> 
>>Check the linux-pcmcia list, check -mm and you'll see that the PCMCIA layer
>>is in the process of being updated.
>>
> 
> 
> OK thanks for the info, sounds like we should try -mm.  It's certainly
> possible that the hardware is buggy.
> 
> Lee
> 

I would try -mm if I could, but I can't.

linux-2.6.12-rc6-mm1 fails to even boot on my system. See previous email
for details.

I have the same problem with 2 different laptops, and those laptops had
different pcmcia chips, so that is why I suspect the pcmcia-core.

James

Re: Bug in pcmcia-core

[Lee Revell]

On Tue, 2005-06-21 at 19:03 +0100, James Courtier-Dutton wrote:
> linux-2.6.12-rc6-mm1 fails to even boot on my system. See previous
> email for details.
> 
> I have the same problem with 2 different laptops, and those laptops
> had different pcmcia chips, so that is why I suspect the pcmcia-core.

James,

Did you ever try 2.6.13-rcx?  I see that lots of PCMCIA fixes went in
since you posted this.

Lee  

Re: Bug in pcmcia-core

[Adam Belay]

On Thu, Jun 16, 2005 at 11:37:30PM +0100, James Courtier-Dutton wrote:
> Hi,
> 
> I have tried conacting the mailing list for the PCMCIA subsystem in
> Linux, but no-one seems to respond.
> 
> PCMCIA SUBSYSTEM
> L:      http://lists.infradead.org/mailman/listinfo/linux-pcmcia
> S:      Unmaintained
> 
> I am trying to write a Linux ALSA driver for the Creative Audigy 2 NX
> Notebook PCMCIA card.
> This is a cardbus card, that uses ioports.
> When it is inserted into the laptop, the entry appears in "lspci -vv "
> showing ioports used by the card.
> As soon as my driver uses "outb()" to anything in the address range
> shown in "lspci -vv" , the PC hangs.
> 
> I can only conclude from this that ioport resources are not being
> allocated correctly to the PCMCIA card.

It's possible.

> 
> Can anybody help me track this down. If someone could tell me which
> PCMCIA and PCI registers should be set for it to work, I could then find
> out which pcmcia registers have not been set correctly, and fix the bug.
> 
> It seems that the PCMCIA specification is not open and free, so I cannot
> refer to it in order to fix this myself.
> 
> Can anybody help me?
> 
> James

Please provide more information.  /proc/ioports, lspci -vv, the ranges
assigned to your driver, and your driver code if it's available.  I'll try
to look into the problem.

Thanks,
Adam

Re: Bug in pcmcia-core

[James Courtier-Dutton]

Adam Belay wrote:
> On Thu, Jun 16, 2005 at 11:37:30PM +0100, James Courtier-Dutton wrote:
> 
>>Hi,
>>
>>I have tried conacting the mailing list for the PCMCIA subsystem in
>>Linux, but no-one seems to respond.
>>
>>PCMCIA SUBSYSTEM
>>L:      http://lists.infradead.org/mailman/listinfo/linux-pcmcia
>>S:      Unmaintained
>>
>>I am trying to write a Linux ALSA driver for the Creative Audigy 2 NX
>>Notebook PCMCIA card.
>>This is a cardbus card, that uses ioports.
>>When it is inserted into the laptop, the entry appears in "lspci -vv "
>>showing ioports used by the card.
>>As soon as my driver uses "outb()" to anything in the address range
>>shown in "lspci -vv" , the PC hangs.
>>
>>I can only conclude from this that ioport resources are not being
>>allocated correctly to the PCMCIA card.
> 
> 
> It's possible.
> 
> 
>>Can anybody help me track this down. If someone could tell me which
>>PCMCIA and PCI registers should be set for it to work, I could then find
>>out which pcmcia registers have not been set correctly, and fix the bug.
>>
>>It seems that the PCMCIA specification is not open and free, so I cannot
>>refer to it in order to fix this myself.
>>
>>Can anybody help me?
>>
>>James
> 
> 
> Please provide more information.  /proc/ioports, lspci -vv, the ranges
> assigned to your driver, and your driver code if it's available.  I'll try
> to look into the problem.
> 
> Thanks,
> Adam
> 
> 

Please see bug#5057. I have placed all the information there.

http://bugzilla.kernel.org/show_bug.cgi?id=5057

James

Bug in pcmcia-core

[James Courtier-Dutton]

Hi,

I have tried conacting the mailing list for the PCMCIA subsystem in 
Linux, but no-one seems to respond.

PCMCIA SUBSYSTEM
L:      http://lists.infradead.org/mailman/listinfo/linux-pcmcia
S:      Unmaintained

I am trying to write a Linux ALSA driver for the Creative Audigy 2 NX 
Notebook PCMCIA card.
This is a cardbus card, that uses ioports.
When it is inserted into the laptop, the entry appears in "lspci -vv " 
showing ioports used by the card.
As soon as my driver uses "outb()" to anything in the address range 
shown in "lspci -vv" , the PC hangs.

I can only conclude from this that ioport resources are not being 
allocated correctly to the PCMCIA card.

Can anybody help me track this down. If someone could tell me which 
PCMCIA and PCI registers should be set for it to work, I could then find 
out which pcmcia registers have not been set correctly, and fix the bug.

It seems that the PCMCIA specification is not open and free, so I cannot 
refer to it in order to fix this myself.

Can anybody help me?

James