From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:42021) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1cEXmf-0000Xg-Us for qemu-devel@nongnu.org; Wed, 07 Dec 2016 03:41:02 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1cEXmb-0003vf-SR for qemu-devel@nongnu.org; Wed, 07 Dec 2016 03:41:01 -0500 Received: from mail-wm0-f43.google.com ([74.125.82.43]:35208) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1cEXmb-0003va-DW for qemu-devel@nongnu.org; Wed, 07 Dec 2016 03:40:57 -0500 Received: by mail-wm0-f43.google.com with SMTP id a197so156688105wmd.0 for ; Wed, 07 Dec 2016 00:40:57 -0800 (PST) References: <20161206205627.8443-1-bobby.prani@gmail.com> <20161206205627.8443-2-bobby.prani@gmail.com> From: Alex =?utf-8?Q?Benn=C3=A9e?= In-reply-to: <20161206205627.8443-2-bobby.prani@gmail.com> Date: Wed, 07 Dec 2016 08:39:54 +0000 Message-ID: <87a8c8yvxh.fsf@linaro.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit Subject: Re: [Qemu-devel] [RFC PATCH 1/3] tcg: Release tb_lock in the order acquired List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Pranith Kumar Cc: qemu-devel@nongnu.org, rth@twiddle.net, mst@redhat.com, cota@braap.org Pranith Kumar writes: > We acquire mmap lock and tb lock in one order and release them in a > different order. This does not need to be that way. > > This patch was inspired by a previous patch by Emilio G. Cota > (https://lists.gnu.org/archive/html/qemu-devel/2016-08/msg03785.html). > > Signed-off-by: Pranith Kumar > --- > cpu-exec.c | 12 +++--------- > 1 file changed, 3 insertions(+), 9 deletions(-) > > diff --git a/cpu-exec.c b/cpu-exec.c > index aa8318d864..f4a00f5047 100644 > --- a/cpu-exec.c > +++ b/cpu-exec.c > @@ -318,7 +318,6 @@ static inline TranslationBlock *tb_find(CPUState *cpu, > TranslationBlock *tb; > target_ulong cs_base, pc; > uint32_t flags; > - bool have_tb_lock = false; > > /* we record a subset of the CPU state. It will > always be the same before a given translated block > @@ -336,7 +335,6 @@ static inline TranslationBlock *tb_find(CPUState *cpu, > */ > mmap_lock(); > tb_lock(); > - have_tb_lock = true; > > /* There's a chance that our desired tb has been translated while > * taking the locks so we check again inside the lock. > @@ -347,6 +345,7 @@ static inline TranslationBlock *tb_find(CPUState *cpu, > tb = tb_gen_code(cpu, pc, cs_base, flags, 0); > } > > + tb_unlock(); > mmap_unlock(); > } > > @@ -364,17 +363,12 @@ static inline TranslationBlock *tb_find(CPUState *cpu, > #endif > /* See if we can patch the calling TB. */ > if (last_tb && !qemu_loglevel_mask(CPU_LOG_TB_NOCHAIN)) { > - if (!have_tb_lock) { > - tb_lock(); > - have_tb_lock = true; > - } > if (!tb->invalid) { > + tb_lock(); > tb_add_jump(last_tb, tb_exit, tb); > + tb_unlock(); > } > } > - if (have_tb_lock) { > - tb_unlock(); > - } > return tb; > } Do you have any numbers for this? The main reason being we are trying to avoid bouncing the lock too much and while this is cleaner it could cause more contention. -- Alex Bennée