* [PATCH] powerpc: Fix deadlock in icswx code [not found] <20110915052313.7127520d@kryten> @ 2011-09-14 19:43 ` Anton Blanchard 2011-09-15 13:09 ` Jimi Xenidis 0 siblings, 1 reply; 3+ messages in thread From: Anton Blanchard @ 2011-09-14 19:43 UTC (permalink / raw) To: benh, paulus; +Cc: linuxppc-dev The icswx code introduced an A-B B-A deadlock: CPU0 CPU1 ---- ---- lock(&anon_vma->mutex); lock(&mm->mmap_sem); lock(&anon_vma->mutex); lock(&mm->mmap_sem); Instead of using the mmap_sem to keep mm_users constant, take the page table spinlock. Signed-off-by: Anton Blanchard <anton@samba.org> Cc: <stable@kernel.org> --- diff --git a/arch/powerpc/mm/mmu_context_hash64.c b/arch/powerpc/mm/mmu_context_hash64.c index 3bafc3d..4ff587e 100644 --- a/arch/powerpc/mm/mmu_context_hash64.c +++ b/arch/powerpc/mm/mmu_context_hash64.c @@ -136,8 +136,8 @@ int use_cop(unsigned long acop, struct mm_struct *mm) if (!mm || !acop) return -EINVAL; - /* We need to make sure mm_users doesn't change */ - down_read(&mm->mmap_sem); + /* The page_table_lock ensures mm_users won't change under us */ + spin_lock(&mm->page_table_lock); spin_lock(mm->context.cop_lockp); if (mm->context.cop_pid == COP_PID_NONE) { @@ -164,7 +164,7 @@ int use_cop(unsigned long acop, struct mm_struct *mm) out: spin_unlock(mm->context.cop_lockp); - up_read(&mm->mmap_sem); + spin_unlock(&mm->page_table_lock); return ret; } @@ -185,8 +185,8 @@ void drop_cop(unsigned long acop, struct mm_struct *mm) if (WARN_ON_ONCE(!mm)) return; - /* We need to make sure mm_users doesn't change */ - down_read(&mm->mmap_sem); + /* The page_table_lock ensures mm_users won't change under us */ + spin_lock(&mm->page_table_lock); spin_lock(mm->context.cop_lockp); mm->context.acop &= ~acop; @@ -213,7 +213,7 @@ void drop_cop(unsigned long acop, struct mm_struct *mm) } spin_unlock(mm->context.cop_lockp); - up_read(&mm->mmap_sem); + spin_unlock(&mm->page_table_lock); } EXPORT_SYMBOL_GPL(drop_cop); ^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH] powerpc: Fix deadlock in icswx code 2011-09-14 19:43 ` [PATCH] powerpc: Fix deadlock in icswx code Anton Blanchard @ 2011-09-15 13:09 ` Jimi Xenidis 2011-09-27 10:04 ` Anton Blanchard 0 siblings, 1 reply; 3+ messages in thread From: Jimi Xenidis @ 2011-09-15 13:09 UTC (permalink / raw) To: Anton Blanchard; +Cc: paulus, linuxppc-dev On Sep 14, 2011, at 2:43 PM, Anton Blanchard wrote: Hi Anton, It would really help me a lot if you could review and maybe merge this = with my earlier patch that splits this file. <http://patchwork.ozlabs.org/patch/109103/> All it does is split.. I promise. You don't have to take the other stuff.. yet :) -JX >=20 > The icswx code introduced an A-B B-A deadlock: >=20 > CPU0 CPU1 > ---- ---- > lock(&anon_vma->mutex); > lock(&mm->mmap_sem); > lock(&anon_vma->mutex); > lock(&mm->mmap_sem); >=20 > Instead of using the mmap_sem to keep mm_users constant, take the > page table spinlock. >=20 > Signed-off-by: Anton Blanchard <anton@samba.org> > Cc: <stable@kernel.org> > --- >=20 > diff --git a/arch/powerpc/mm/mmu_context_hash64.c = b/arch/powerpc/mm/mmu_context_hash64.c > index 3bafc3d..4ff587e 100644 > --- a/arch/powerpc/mm/mmu_context_hash64.c > +++ b/arch/powerpc/mm/mmu_context_hash64.c > @@ -136,8 +136,8 @@ int use_cop(unsigned long acop, struct mm_struct = *mm) > if (!mm || !acop) > return -EINVAL; >=20 > - /* We need to make sure mm_users doesn't change */ > - down_read(&mm->mmap_sem); > + /* The page_table_lock ensures mm_users won't change under us */ > + spin_lock(&mm->page_table_lock); > spin_lock(mm->context.cop_lockp); >=20 > if (mm->context.cop_pid =3D=3D COP_PID_NONE) { > @@ -164,7 +164,7 @@ int use_cop(unsigned long acop, struct mm_struct = *mm) >=20 > out: > spin_unlock(mm->context.cop_lockp); > - up_read(&mm->mmap_sem); > + spin_unlock(&mm->page_table_lock); >=20 > return ret; > } > @@ -185,8 +185,8 @@ void drop_cop(unsigned long acop, struct mm_struct = *mm) > if (WARN_ON_ONCE(!mm)) > return; >=20 > - /* We need to make sure mm_users doesn't change */ > - down_read(&mm->mmap_sem); > + /* The page_table_lock ensures mm_users won't change under us */ > + spin_lock(&mm->page_table_lock); > spin_lock(mm->context.cop_lockp); >=20 > mm->context.acop &=3D ~acop; > @@ -213,7 +213,7 @@ void drop_cop(unsigned long acop, struct mm_struct = *mm) > } >=20 > spin_unlock(mm->context.cop_lockp); > - up_read(&mm->mmap_sem); > + spin_unlock(&mm->page_table_lock); > } > EXPORT_SYMBOL_GPL(drop_cop); >=20 > _______________________________________________ > Linuxppc-dev mailing list > Linuxppc-dev@lists.ozlabs.org > https://lists.ozlabs.org/listinfo/linuxppc-dev ^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] powerpc: Fix deadlock in icswx code 2011-09-15 13:09 ` Jimi Xenidis @ 2011-09-27 10:04 ` Anton Blanchard 0 siblings, 0 replies; 3+ messages in thread From: Anton Blanchard @ 2011-09-27 10:04 UTC (permalink / raw) To: Jimi Xenidis; +Cc: paulus, linuxppc-dev Hi Jimi, > It would really help me a lot if you could review and maybe merge > this with my earlier patch that splits this file. > <http://patchwork.ozlabs.org/patch/109103/> All it does is split.. I > promise. You don't have to take the other stuff.. yet :) Sorry it took so long to get to this. Looking at it now. Anton ^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2011-09-27 10:04 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
[not found] <20110915052313.7127520d@kryten>
2011-09-14 19:43 ` [PATCH] powerpc: Fix deadlock in icswx code Anton Blanchard
2011-09-15 13:09 ` Jimi Xenidis
2011-09-27 10:04 ` Anton Blanchard
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).