From mboxrd@z Thu Jan 1 00:00:00 1970 From: Shan Wei Subject: Re: [PATCH v5 3/9] net: xfrm: use __this_cpu_read per-cpu helper Date: Tue, 13 Nov 2012 20:36:00 +0800 Message-ID: <50A23EB0.1060808@gmail.com> References: <50A1A7C9.3060703@gmail.com> <20121113072101.GG22290@secunet.com> <50A213DF.1040103@gmail.com> <20121113104839.GJ22290@secunet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Christoph Lameter , David Miller , NetDev , Herbert Xu , Kernel-Maillist To: Steffen Klassert Return-path: In-Reply-To: <20121113104839.GJ22290@secunet.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Steffen Klassert said, at 2012/11/13 18:48: >=20 > Ok, so please add a commit message to describe your changes. >=20 > Thanks. >=20 [PATCH v5] net: xfrm: use __this_cpu_read per-cpu helper this_cpu_ptr/this_cpu_read is faster than per_cpu_ptr(p, smp_processor_= id())=20 and can reduce memory accesses. The latter helper needs to find the offset for current cpu, and needs more assembler instructions which objdump shows in following.= =20 this_cpu_ptr relocates and address. this_cpu_read() relocates the addre= ss and performs the fetch. this_cpu_read() saves you more instructions since it can do the relocation and the fetch in one instruction. per_cpu_ptr(p, smp_processor_id())=EF=BC=9A 1e: 65 8b 04 25 00 00 00 00 mov %gs:0x0,%eax 26: 48 98 cltq 28: 31 f6 xor %esi,%esi 2a: 48 c7 c7 00 00 00 00 mov $0x0,%rdi 31: 48 8b 04 c5 00 00 00 00 mov 0x0(,%rax,8),%rax 39: c7 44 10 04 14 00 00 00 movl $0x14,0x4(%rax,%rdx,1) this_cpu_ptr(p) 1e: 65 48 03 14 25 00 00 00 00 add %gs:0x0,%rdx 27: 31 f6 xor %esi,%esi 29: c7 42 04 14 00 00 00 movl $0x14,0x4(%rdx) 30: 48 c7 c7 00 00 00 00 mov $0x0,%rdi Signed-off-by: Shan Wei --- net/xfrm/xfrm_ipcomp.c | 8 +++----- 1 files changed, 3 insertions(+), 5 deletions(-) diff --git a/net/xfrm/xfrm_ipcomp.c b/net/xfrm/xfrm_ipcomp.c index e5246fb..2906d52 100644 --- a/net/xfrm/xfrm_ipcomp.c +++ b/net/xfrm/xfrm_ipcomp.c @@ -276,18 +276,16 @@ static struct crypto_comp * __percpu *ipcomp_allo= c_tfms(const char *alg_name) struct crypto_comp * __percpu *tfms; int cpu; =20 - /* This can be any valid CPU ID so we don't need locking. */ - cpu =3D raw_smp_processor_id(); =20 list_for_each_entry(pos, &ipcomp_tfms_list, list) { struct crypto_comp *tfm; =20 - tfms =3D pos->tfms; - tfm =3D *per_cpu_ptr(tfms, cpu); + /* This can be any valid CPU ID so we don't need locking. */ + tfm =3D __this_cpu_read(*pos->tfms); =20 if (!strcmp(crypto_comp_name(tfm), alg_name)) { pos->users++; - return tfms; + return pos->tfms; } } =20 --=20 1.7.1