Re: Linux 2.6.32.47

[Ian Campbell]

On Tue, 2011-11-08 at 15:35 -0800, Greg KH wrote:
> On Tue, Nov 08, 2011 at 11:40:00PM +0100, Jiri Slaby wrote:
> > On 11/08/2011 08:22 AM, Ian Campbell wrote:
> > >On Mon, 2011-11-07 at 16:30 -0800, Greg KH wrote:
> > >>On Tue, Nov 08, 2011 at 12:43:49AM +0100, Christoph Biedl wrote:
> > >>>Greg KH wrote...
> > >>>
> > >>>>I'm announcing the release of the 2.6.32.47 kernel.
> > >>>
> > >>>Unfortunately this does not build anymore on ARM (Debian armel):
> > >>>
> > >>>   CHK     include/linux/version.h
> > >>>   SYMLINK include/asm ->  include/asm-arm
> > >>>make[1]: `include/asm-arm/mach-types.h' is up to date.
> > >>>   CHK     include/linux/utsrelease.h
> > >>>   CALL    scripts/checksyscalls.sh
> > >>>   CHK     include/linux/compile.h
> > >>>   GEN     .version
> > >>>   CHK     include/linux/compile.h
> > >>>   UPD     include/linux/compile.h
> > >>>   CC      init/version.o
> > >>>   LD      init/built-in.o
> > >>>   LD      .tmp_vmlinux1
> > >>>drivers/built-in.o: In function `sysdev_resume':
> > >>>ledtrig-default-on.c:(.text+0x2a97c): undefined reference to `irq_pm_syscore_resume'
> > >>>make: *** [.tmp_vmlinux1] Error 1
> > >>>
> > >>>
> > >>>The most likely culprit is
> > >>>
> > >>>     commit 0f12a6ad9fa3a03f2bcee36c9cb704821e244c40
> > >>>     Author: Ian Campbell<ian.campbell@xxxxxxxxxx>
> > >>>     Date:   Mon Oct 3 15:37:00 2011 +0100
> > >>>
> > >>>	genirq: Add IRQF_RESUME_EARLY and resume such IRQs earlier
> > >>>
> > >>>as reverting that commit made the error go away.
> > >>>
> > >>>The .config used is attached. FWIW, the target system is a DockStar.
> > >>>
> > >>>Sorry for not having checked during the review period, I should have
> > >>>done so.
> > >>
> > >>Not good.
> > >>
> > >>Ian, Konrad, any thoughts as to how to fix this?
> > >
> > >Looks like I got the wrong CONFIG symbol for the function, or more
> > >importantly the dummy version of it. Sorry about this.
> > >
> > >Ian.
> > >
> > >8<--------------------------------------------------------------
> > >
> > >>From 33a64cf4a184b47493ef312b08468aca3361e112 Mon Sep 17 00:00:00 2001
> > >From: Ian Campbell<ian.campbell@xxxxxxxxxx>
> > >Date: Tue, 8 Nov 2011 07:19:06 +0000
> > >Subject: [PATCH] irq: fix build error due to missing irq_pm_syscore_resume
> > >
> > >irq_pm_syscore_resume is only available iff CONFIG_PM_SLEEP (kernel/irq/pm.o is
> > >only built if this is true). Move the definition (and the dummy definition)
> > >under that umbrella.
> > >
> > >Introduced by the backport of upstream 9bab0b7fbaceec47d32db51cd9e59c82fb071f5a
> > >as 0f12a6ad9fa3a03f2bcee36c9cb704821e244c40.
> > >
> > >Signed-off-by: Ian Campbell<ian.campbell@xxxxxxxxxx>
> > >Reported-by: Christoph Biedl<linux-kernel.bfrz@xxxxxxxxxxxxxxxxxx>
> > >Reported-by: Antoine Martin<antoine@xxxxxxxxxxxxx>
> > >---
> > >  include/linux/interrupt.h |    4 ++--
> > >  1 files changed, 2 insertions(+), 2 deletions(-)
> > >
> > >diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
> > >index c7e1aa5..a65ef60 100644
> > >--- a/include/linux/interrupt.h
> > >+++ b/include/linux/interrupt.h
> > >@@ -199,16 +199,16 @@ extern void enable_irq(unsigned int irq);
> > >  #ifdef CONFIG_GENERIC_HARDIRQS
> > >  extern void suspend_device_irqs(void);
> > >  extern void resume_device_irqs(void);
> > >-extern void irq_pm_syscore_resume(void);
> > >  #ifdef CONFIG_PM_SLEEP
> > >  extern int check_wakeup_irqs(void);
> > >+extern void irq_pm_syscore_resume(void);
> > >  #else
> > >  static inline int check_wakeup_irqs(void) { return 0; }
> > >+static inline void irq_pm_syscore_resume(void) { };
> > >  #endif
> > >  #else
> > >  static inline void suspend_device_irqs(void) { };
> > >  static inline void resume_device_irqs(void) { };
> > >-static inline void irq_pm_syscore_resume(void) { };
> > >  static inline int check_wakeup_irqs(void) { return 0; }
> > >  #endif
> > 
> > Hmm, sparc doesn't have generic irqs. So:
> > drivers/base/sys.c: In function 'sysdev_resume':
> > drivers/base/sys.c:478: error: implicit declaration of function
> > 'irq_pm_syscore_resume'
> 
> Ugh, what a mess.
> 
> Ian, I'm just going to revert your original patch, and the other 2
> patches that broke the builds, and do a new .32 release.
> 
> Please send me the correct patch that builds on all arches, if you all
> feel it is still needed.

Sorry about this. I should have just looked at where else
check_wakeup_irqs was stubbed out and I'd have gotten this right first
time.

I'm pretty sure this is now correct but to make double sure I've tested
by building defconfig arm amd64 cris i386 ia64 m68k and s390x. I tried a
bunch of other archs too but they failed for other (pre-existing)
reasons but with no mention of irq_pm_syscore_resume (I used "make -k"
so it'll have built as much as it could).

Sorry again for the mess.

Ian.

8<----------------------------------------