On Mon, Aug 20, 2018 at 09:26:39AM -0700, Matthew Wilcox wrote: > On Mon, Aug 20, 2018 at 05:56:50PM +0200, Peter Zijlstra wrote: > > Yeah, _acquire should be retained; sorry about loosing that. I'm neck > > deep into tlb invalidate stuff and wrote this without much thinking > > involved. > > NP. Here's the current version I've got, with some updated likely() > hints. > > From 337298a88266f7b21492f893c2bf05409a5392c8 Mon Sep 17 00:00:00 2001 > From: Matthew Wilcox <willy@xxxxxxxxxxxxx> > Date: Mon, 20 Aug 2018 10:19:14 -0400 > Subject: [PATCH] locking: Remove an insn from spin and write locks > > Both spin locks and write locks currently do: > > f0 0f b1 17 lock cmpxchg %edx,(%rdi) > 85 c0 test %eax,%eax > 75 05 jne [slowpath] > > This 'test' insn is superfluous; the cmpxchg insn sets the Z flag > appropriately. Peter pointed out that using atomic_try_cmpxchg_acquire() > will let the compiler know this is true. Comparing before/after > disassemblies show the only effect is to remove this insn. > > Take this opportunity to make the spin & write lock code resemble each > other more closely and have similar likely() hints. > > Suggested-by: Peter Zijlstra <peterz@xxxxxxxxxxxxx> > Signed-off-by: Matthew Wilcox <willy@xxxxxxxxxxxxx> > --- > include/asm-generic/qrwlock.h | 7 ++++--- > include/asm-generic/qspinlock.h | 17 ++++++++++------- > 2 files changed, 14 insertions(+), 10 deletions(-) Shouldn't make any difference on arm64, so: Acked-by: Will Deacon <will.deacon@xxxxxxx> Will
