x86_64 remote tlb invalidation

[Michel Lespinasse <walken@google.com>]

Reading through arch/x86/kernel/tlb_64.c, I've been wondering why
the code for native_flush_tlb_others() can't be simplified as follows:

--- linux-2.6.29.2.orig/arch/x86/kernel/tlb_64.c	2009-04-27 10:37:11.000000000 -0700
+++ linux-2.6.29.2/arch/x86/kernel/tlb_64.c	2009-05-05 16:53:07.770085000 -0700
@@ -180,7 +180,7 @@
 
 	f->flush_mm = mm;
 	f->flush_va = va;
-	cpus_or(f->flush_cpumask, cpumask, f->flush_cpumask);
+	f->flush_cpumask = cpumask;
 
 	/*
 	 * Make the above memory operations globally visible before


My reasoning:

* The previous invocation of native_flush_tlb_others() waited for
  f->flush_cpumask to be cleared before it unlocked f->tlbstate_lock;

* sets bits in the flush_cpumask field of union smp_flush_state without
  holding the correwsponding tlbstate_lock;

* Therefore, f->flush_cpumask must still be cleared when the next invocation
  of native_flush_tlb_others() reacquires f->tlbstate_lock.

I don't think I missed anything, did I ?

Thanks,

-- 
Michel "Walken" Lespinasse
A program is never fully debugged until the last user dies.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>