Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758916Ab3HOS2Z (ORCPT ); Thu, 15 Aug 2013 14:28:25 -0400 Received: from moutng.kundenserver.de ([212.227.17.8]:64021 "EHLO moutng.kundenserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753459Ab3HOS2X (ORCPT ); Thu, 15 Aug 2013 14:28:23 -0400 Message-ID: <520D1D9D.2040706@dawncrow.de> Date: Thu, 15 Aug 2013 20:27:41 +0200 From: =?UTF-8?B?QW5kcsOpIEhlbnRzY2hlbA==?= User-Agent: Mozilla/5.0 (X11; Linux i686; rv:17.0) Gecko/20130803 Thunderbird/17.0.8 MIME-Version: 1.0 To: Jonathan Austin CC: "linux-kernel@vger.kernel.org" , Will Deacon , "linux-arch@vger.kernel.org" , "linux-arm-kernel@lists.infradead.org" , Russell King Subject: Re: arm: Only load TLS values when needed References: <51E42E11.1010903@dawncrow.de> <51E5836B.1010904@arm.com> <51E59E8F.1060501@dawncrow.de> <51E67B98.9040101@arm.com> <51E6F532.1030004@dawncrow.de> <520B8F37.4040609@dawncrow.de> <520BAE58.3060600@arm.com> <520BF4E1.8040708@dawncrow.de> <520D0FFA.1060508@arm.com> In-Reply-To: <520D0FFA.1060508@arm.com> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-Provags-ID: V02:K0:MCDWEEco93yDtM+wxLM4v7PNmv7d/9Bhm1QnjyTEpxI 74IpnvBl3yu1LYUOQe/JciQkzdykM8xUtpvN3268+pkWbvJpC0 +4jva7rc8SRrBWaZ+HR1R5BKjHXDsVn9e5nc8gBWa9CHgJPwNe k//9QsLNL9emjaht7T6vygfjRt8Iopv+c9byFQ3j8uMIc67ArF PPJCFtwmIX+CA+TaS01cwdL/8LJY6IJzN7IePifUUpFW4Tn6rw MResK6f5YiqpmxWq7phT0dGtBU271+YTN2ik2OEWXeQiTbUDh4 E/nrDtG029GnEjpvPdDaipWBkT2M8AIkh5BmpUJN7fi3X8CsbO oU0szhrR3f23Lh3rcXnE= Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 6569 Lines: 164 Hi Jonathan, Am 15.08.2013 19:29, schrieb Jonathan Austin: > Hi André > > (I've put RMK and LAKML back on Cc: they got dropped somewhere along the way) > > On 14/08/13 22:21, André Hentschel wrote: >>>>>> [...] >>>>>>>> Now we've only got one instruction between the store and the load >>>>>>>> and risk stalling the pipeline... >>>>>>>> >>>>>>>> Dave M cautiously says "The ancient advice was that one instruction >>>>>>>> was enough" but this is very core dependent... I wonder if anyone >>>>>>>> has a good idea about whether this is an issue here...? >>>>>>> >>>>>>> We could use a ldrd at the top, that'd be nearly what we have right >>>>>>> now, don't we? >>>>>> >>>>>> Yea, that'd be good - as far as I can see from an 1136 TRM, the ldrd *may* be two cycles (depending on alignment of the words) but the ldr and ldrne will always be two cycles. Ahhh, the joys of modifying the fast path ;) >>> >>> Was expecting to see something that reflected this discussion, >> >> Ah ok, i misunderstood that, sry. >> Something like that? >> >> From: André Hentschel >> >> This patch intents to reduce loading instructions when the resulting value is not used. >> It's a follow up on a4780adeefd042482f624f5e0d577bf9cdcbb760 >> > > You caught me just before running away for a week of holiday, but I've managed to squeeze in a bit of testing - hope this helps - I'll try to be clear about what I have/haven't tested so you can judge for yourself how much weight to give this. Thanks for testing it anyway, and happy holiday! > What I've tested: > > * Apply your code to 3.11-rc5 > * Boot on Integrator CP (1176 r1p0) and run a full Debian Wheezy init > > --> It works! > > I also applied the following diff and rebuilt to check that we were really testing the V6 NOT V6K path here: > > diff --git a/arch/arm/include/asm/tls.h b/arch/arm/include/asm/tls.h > index 31743f7..71dfe82 100644 > --- a/arch/arm/include/asm/tls.h > +++ b/arch/arm/include/asm/tls.h > @@ -43,6 +43,7 @@ > #define switch_tls switch_tls_v6 > #elif defined(CONFIG_CPU_32v6K) > #define tls_emu 0 > +#error > #define has_tls_reg 1 > #define switch_tls switch_tls_v6k > #else > > What I've *not* tested: > * As this CPU does not have the tls register, I haven't run any tests that try to read/write it, but we wouldn't expect that to work, right? > * Haven't tried a V7 platform, as I understand you've already done that. > * Haven't verified it still builds for v4 (which was the problem for ldrd in the past and now we've put it back, so that's important) It shouldn't matter as v4 would not use the v6 codepath, would it? > * Performance impact I only could test it on v7, would be interesting to see some v6, v5, ... "benchmarks", though. > > I think it'd be good if you can verify the bottom two points - especially the ldrd one: perhaps Russell can tell you which platform it broke on before? > > So, that's a > Tested-by: Jonathan Austin > > If you want it :) sure, thx. > Hope that helps, > > Jonny > >> Signed-off-by: André Hentschel >> >> --- >> This patch is against 28fbc8b6a29c849a3f03a6b05010d4b584055665 >> >> diff --git a/arch/arm/include/asm/tls.h b/arch/arm/include/asm/tls.h >> index 83259b8..31743f7 100644 >> --- a/arch/arm/include/asm/tls.h >> +++ b/arch/arm/include/asm/tls.h >> @@ -3,29 +3,31 @@ >> >> #ifdef __ASSEMBLY__ >> #include >> - .macro switch_tls_none, base, tp, tpuser, tmp1, tmp2 >> + .macro switch_tls_none, prev, next, tp, tpuser, tmp1, tmp2 >> .endm >> >> - .macro switch_tls_v6k, base, tp, tpuser, tmp1, tmp2 >> + .macro switch_tls_v6k, prev, next, tp, tpuser, tmp1, tmp2 >> + ldrd \tp, \tpuser, [\next, #TI_TP_VALUE] @ get the next TLS and user r/w register >> mrc p15, 0, \tmp2, c13, c0, 2 @ get the user r/w register >> mcr p15, 0, \tp, c13, c0, 3 @ set TLS register >> mcr p15, 0, \tpuser, c13, c0, 2 @ and the user r/w register >> - str \tmp2, [\base, #TI_TP_VALUE + 4] @ save it >> + str \tmp2, [\prev, #TI_TP_VALUE + 4] @ save it >> .endm >> >> - .macro switch_tls_v6, base, tp, tpuser, tmp1, tmp2 >> + .macro switch_tls_v6, prev, next, tp, tpuser, tmp1, tmp2 >> + ldrd \tp, \tpuser, [\next, #TI_TP_VALUE] >> ldr \tmp1, =elf_hwcap >> ldr \tmp1, [\tmp1, #0] >> mov \tmp2, #0xffff0fff >> tst \tmp1, #HWCAP_TLS @ hardware TLS available? >> streq \tp, [\tmp2, #-15] @ set TLS value at 0xffff0ff0 >> - mrcne p15, 0, \tmp2, c13, c0, 2 @ get the user r/w register >> + mrcne p15, 0, \tmp2, c13, c0, 2 @ get the previous user r/w register >> mcrne p15, 0, \tp, c13, c0, 3 @ yes, set TLS register >> mcrne p15, 0, \tpuser, c13, c0, 2 @ set user r/w register >> - strne \tmp2, [\base, #TI_TP_VALUE + 4] @ save it >> + strne \tmp2, [\prev, #TI_TP_VALUE + 4] @ save it >> .endm >> >> - .macro switch_tls_software, base, tp, tpuser, tmp1, tmp2 >> + .macro switch_tls_software, prev, next, tp, tpuser, tmp1, tmp2 >> mov \tmp1, #0xffff0fff >> str \tp, [\tmp1, #-15] @ set TLS value at 0xffff0ff0 >> .endm >> diff --git a/arch/arm/kernel/entry-armv.S b/arch/arm/kernel/entry-armv.S >> index d40d0ef..11112de 100644 >> --- a/arch/arm/kernel/entry-armv.S >> +++ b/arch/arm/kernel/entry-armv.S >> @@ -689,12 +689,10 @@ ENTRY(__switch_to) >> THUMB( stmia ip!, {r4 - sl, fp} ) @ Store most regs on stack >> THUMB( str sp, [ip], #4 ) >> THUMB( str lr, [ip], #4 ) >> - ldr r4, [r2, #TI_TP_VALUE] >> - ldr r5, [r2, #TI_TP_VALUE + 4] >> #ifdef CONFIG_CPU_USE_DOMAINS >> ldr r6, [r2, #TI_CPU_DOMAIN] >> #endif >> - switch_tls r1, r4, r5, r3, r7 >> + switch_tls r1, r2, r4, r5, r3, r7 >> #if defined(CONFIG_CC_STACKPROTECTOR) && !defined(CONFIG_SMP) >> ldr r7, [r2, #TI_TASK] >> ldr r8, =__stack_chk_guard >> >> >> > > > -- > To unsubscribe from this list: send the line "unsubscribe linux-arch" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/