The recent addition of (Colibri) PXA3xx specific declarations breaks the 
building of the Colibri 270 with similar features.

CC: Daniel Mack <daniel@caiaq.de>
CC: Eric Miao <eric.y.miao@gmail.com>
Signed-off-by <jakob.viketoft@bitsim.com>

--- a/arch/arm/mach-pxa/include/mach/colibri.h	2010-03-15 
17:09:39.000000000 +0100
+++ b/arch/arm/mach-pxa/include/mach/colibri.h	2010-04-16 
11:06:15.000000000 +0200
@@ -7,6 +7,7 @@
   * common settings for all modules
   */

+#if defined(CONFIG_MACH_COLIBRI300) || defined(CONFIG_MACH_COLIBRI320)
  #if defined(CONFIG_MMC_PXA) || defined(CONFIG_MMC_PXA_MODULE)
  extern void colibri_pxa3xx_init_mmc(mfp_cfg_t *pins, int len, int 
detect_pin);
  #else
@@ -28,6 +29,7 @@
  #else
  static inline void colibri_pxa3xx_init_nand(void) {}
  #endif
+#endif

  /* physical memory regions */
  #define COLIBRI_SDRAM_BASE	0xa0000000      /* SDRAM region */

--

[ message continues ]

Hmm, I can't see any problem here. Which error did you get? Which boards
did you select in your config?

--

[ message continues ]

Hmmm, it appears I was slightly too fast on the trigger there. I 
encountered the problem some time ago when migrating fron 2.6.29 to 
2.6.33.1 and a quick look made me think there was a linking problem. I'm 
finishing up this project now and wanted to push any relevant fixes 
upstream. However, the problem I seem to be getting is something 
different though:

   CC      drivers/pcmcia/pxa2xx_colibri.o
In file included from drivers/pcmcia/pxa2xx_colibri.c:25:
arch/arm/mach-pxa/include/mach/colibri.h:13: error: expected ')' before 
'*' token
make[2]: *** [drivers/pcmcia/pxa2xx_colibri.o] Error 1
make[1]: *** [drivers/pcmcia] Error 2
make: *** [drivers] Error 2

Is this an error on my part somehow?

	/Jakob

--

[ message continues ]

Adding a "#include <mach/mfp.h>" seem to cure this problem. If you agree 
I can supply a new patch with the correct solution this time.

	/Jakob

--

[ message continues ]

Ok. Add it to colibri.h, and while at it, you can kick the ax88796.h
inclusion. That's a leftover and isn't needed.

Thanks,
Daniel
--

[ message continues ]

And Cc: stable@kernel.org for the backmerge to 2.6.33.

Thanks,
Daniel
--

[ message continues ]

Jakob, any update on this?
--

[ message continues ]

Jakob provided an updated patch for this:

  http://lkml.org/lkml/2010/4/16/100


--

[ message continues ]

Looks it didn't make into -stable. And I'm not able to find
drivers/pcmcia/pxa2xx_colibri.c as well.

Daniel, any idea?
--

[ message continues ]

Greg, stable@kernel.org was in Cc: for the patch in the URL above.

Erm - how is that related to the patch in question?

Daniel
--

[ message continues ]

The error seems to be (as in previous mail):

 CC      drivers/pcmcia/pxa2xx_colibri.o
In file included from drivers/pcmcia/pxa2xx_colibri.c:25:
arch/arm/mach-pxa/include/mach/colibri.h:13: error: expected ')'
before '*' token
make[2]: *** [drivers/pcmcia/pxa2xx_colibri.o] Error 1
make[1]: *** [drivers/pcmcia] Error 2
make: *** [drivers] Error 2

And actually, inclusion of <mach/mfp.h> is better to be placed where it's
--

[ message continues ]

In general I agree, but in this case, mfp_t is used in this header file,
so every piece of code which includes colibri.h will need both
<mach/mfp.h> and <net/ax88796.h>. And as this is a rule for all users, I
think the right place for the inclusion is the header file?

Apart from that, this fix will work for both mainline and -stable, which
makes things easier :)

Daniel

--

[ message continues ]

I see.


--

[ message continues ]

Hello!

I didn't see this first as the e-mail address jakob.viketoft@bitsim.com 
is being shut down (changing jobs).


The pcmcia-patch isn't my work, but that of Alex A. Mihaylov 
(minimumlaw@rambler.ru). I was hoping that he would be pushing those 
bits upstream separately, but I think some things needs to me shifted 


Sounds good!

	/Jakob

--

[ message continues ]

Is it in Linus's tree?  If so, what is the git commit id of it?  If you
properly tag the patch with "cc: stable <stable@kernel.org>" like you
did in the patch, then it will be automatically sent to me when it hits
Linus's tree.

I don't see this one in my queue at the moment.

thanks,

greg k-h
--

[ message continues ]

I see - thought you were picking them up from the mails.

It's not there yet. Eric picked it, but I'm not sure whether it will
still make it to .34. But ok, let's just wait then.

Thanks,
Daniel
--

[ message continues ]

I _CAN_NOT_ take patches into the stable tree before they go into
Linus's tree, so even if I was picking them up from emails, this could
not be accepted.

Please read the file, Documentation/stable_kernel_rules.txt for how the
process works.

thanks,

greg k-h
--

[ message continues ]

It seems that mfp_cfg_t is unknown. Does including <mach/mfp.h> from
pxa2xx_colibri.c help? However, I can't see any problem on 2.6.34-rc4,
with or without CONFIG_MMC set, so that might only be related to 2.6.33.
Any change of trying this on a cutting edge tree, just to make sure?

Daniel
--

[ message continues ]