From mboxrd@z Thu Jan 1 00:00:00 1970 From: mingo@elte.hu (Ingo Molnar) Date: Thu, 30 Apr 2009 16:10:50 +0200 Subject: [ltt-dev] [PATCH] Fix dirty page accounting in redirty_page_for_writepage() In-Reply-To: References: <20090429232546.GB15782@Krystal> <20090430024303.GB19875@Krystal> <20090430062140.GA9559@elte.hu> <20090430063306.GA27431@Krystal> <20090430065055.GA16277@elte.hu> Message-ID: <20090430141050.GC14696@elte.hu> * Christoph Lameter wrote: > On Thu, 30 Apr 2009, Ingo Molnar wrote: > > > > I see however that it's only guaranteed to be atomic wrt preemption. > > > > That's really only true for the non-x86 fallback defines. If we so > > decide, we could make the fallbacks in asm-generic/percpu.h irq-safe > > The fallbacks have different semantics and therefore we cannot > rely on irq safeness in the core code when using the x86 cpu ops. Well it's irq and preempt safe on x86. It's preempt-safe on other architectures - but the fallback is not irq-safe on other architectures. That is remedied easily via the patch below. (Note: totally untested) Ingo diff --git a/include/linux/percpu.h b/include/linux/percpu.h index 1581ff2..6b3984a 100644 --- a/include/linux/percpu.h +++ b/include/linux/percpu.h @@ -139,17 +139,23 @@ static inline void free_percpu(void *p) #ifndef percpu_read # define percpu_read(var) \ ({ \ + unsigned long flags; \ typeof(per_cpu_var(var)) __tmp_var__; \ - __tmp_var__ = get_cpu_var(var); \ - put_cpu_var(var); \ + \ + local_irq_save(flags); \ + __tmp_var__ = __get_cpu_var(var); \ + local_irq_restore(flags); \ __tmp_var__; \ }) #endif #define __percpu_generic_to_op(var, val, op) \ do { \ - get_cpu_var(var) op val; \ - put_cpu_var(var); \ + unsigned long flags; \ + \ + local_irq_save(flags); \ + op val; \ + local_irq_restore(flags); \ } while (0) #ifndef percpu_write