Re: [PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-30 Thread Randy Dunlap
On Thu, 30 Nov 2006 14:27:19 -0800 Andrew Morton wrote: > On Thu, 30 Nov 2006 14:11:40 -0800 > Randy Dunlap <[EMAIL PROTECTED]> wrote: > > > On Thu, 30 Nov 2006 08:00:00 +0100 Ingo Molnar wrote: > > > > > On Wed, 2006-11-29 at 17:45 -0800, Andrew Morton wrote: > > > > No, I think this patch is r

Re: [PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-30 Thread Randy Dunlap
Andrew Morton wrote: On Thu, 30 Nov 2006 14:11:40 -0800 Randy Dunlap <[EMAIL PROTECTED]> wrote: On Thu, 30 Nov 2006 08:00:00 +0100 Ingo Molnar wrote: On Wed, 2006-11-29 at 17:45 -0800, Andrew Morton wrote: No, I think this patch is right - the declaration of the CONFIG_SMP smp_call_function_

Re: [PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-30 Thread Andrew Morton
On Thu, 30 Nov 2006 14:11:40 -0800 Randy Dunlap <[EMAIL PROTECTED]> wrote: > On Thu, 30 Nov 2006 08:00:00 +0100 Ingo Molnar wrote: > > > On Wed, 2006-11-29 at 17:45 -0800, Andrew Morton wrote: > > > No, I think this patch is right - the declaration of the CONFIG_SMP > > > smp_call_function_single

Re: [PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-30 Thread Randy Dunlap
On Thu, 30 Nov 2006 08:00:00 +0100 Ingo Molnar wrote: > On Wed, 2006-11-29 at 17:45 -0800, Andrew Morton wrote: > > No, I think this patch is right - the declaration of the CONFIG_SMP > > smp_call_function_single() is in linux/smp.h so the !CONFIG_SMP > > declaration > > or definition should be th

Re: [PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-30 Thread Andrew Morton
On Thu, 30 Nov 2006 10:22:20 +0100 Ingo Molnar <[EMAIL PROTECTED]> wrote: > > > > btw, does anyone know why the SMP versions of this function use > > spin_lock_bh(&call_lock)? > > that makes no sense (neither the get_cpu()/put_cpu() gymnastics) if this > is called with irqs disabled all the time

Re: [PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-30 Thread Ingo Molnar
On Wed, 2006-11-29 at 23:54 -0800, Andrew Morton wrote: > which is somewhat unpleasant. I added a WARN_ON(irqs_disabled()) to > the > out-of-line SMP version. ok. > > btw, does anyone know why the SMP versions of this function use > spin_lock_bh(&call_lock)? that makes no sense (neither the get

Re: [PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-29 Thread Andrew Morton
On Thu, 30 Nov 2006 08:00:00 +0100 Ingo Molnar <[EMAIL PROTECTED]> wrote: > On Wed, 2006-11-29 at 17:45 -0800, Andrew Morton wrote: > > No, I think this patch is right - the declaration of the CONFIG_SMP > > smp_call_function_single() is in linux/smp.h so the !CONFIG_SMP > > declaration > > or def

Re: [PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-29 Thread Ingo Molnar
On Wed, 2006-11-29 at 17:45 -0800, Andrew Morton wrote: > No, I think this patch is right - the declaration of the CONFIG_SMP > smp_call_function_single() is in linux/smp.h so the !CONFIG_SMP > declaration > or definition should be there too. > > It's still buggy though. It should disable local i

Re: [PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-29 Thread Andrew Morton
On Wed, 29 Nov 2006 17:01:11 -0800 Randy Dunlap <[EMAIL PROTECTED]> wrote: > From: Randy Dunlap <[EMAIL PROTECTED]> > > smp_call_function_single() needs to be visible in non-SMP builds, to fix: > > arch/x86_64/kernel/vsyscall.c:283: warning: implicit declaration of function > 'smp_call_function

[PATCH -mm] x86_64 UP needs smp_call_function_single

2006-11-29 Thread Randy Dunlap
From: Randy Dunlap <[EMAIL PROTECTED]> smp_call_function_single() needs to be visible in non-SMP builds, to fix: arch/x86_64/kernel/vsyscall.c:283: warning: implicit declaration of function 'smp_call_function_single' The (other/trivial) fix (instead of this one) is to add: #include to linux-2.