From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:54399) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1TXqJG-0004ak-FV for qemu-devel@nongnu.org; Mon, 12 Nov 2012 04:28:05 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1TXqJD-0006bn-DV for qemu-devel@nongnu.org; Mon, 12 Nov 2012 04:28:02 -0500 Received: from mailout3.w1.samsung.com ([210.118.77.13]:29801) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1TXqJD-0006bb-7K for qemu-devel@nongnu.org; Mon, 12 Nov 2012 04:27:59 -0500 Received: from eusync4.samsung.com (mailout3.w1.samsung.com [210.118.77.13]) by mailout3.w1.samsung.com (Oracle Communications Messaging Server 7u4-24.01(7.0.4.24.0) 64bit (built Nov 17 2011)) with ESMTP id <0MDD004CPBNA7B60@mailout3.w1.samsung.com> for qemu-devel@nongnu.org; Mon, 12 Nov 2012 09:28:22 +0000 (GMT) Received: from evvoevodinPC.rnd.samsung.ru ([106.109.8.15]) by eusync4.samsung.com (Oracle Communications Messaging Server 7u4-24.01(7.0.4.24.0) 64bit (built Nov 17 2011)) with ESMTPA id <0MDD00AS6BMG4350@eusync4.samsung.com> for qemu-devel@nongnu.org; Mon, 12 Nov 2012 09:27:56 +0000 (GMT) From: Evgeny Voevodin Date: Mon, 12 Nov 2012 13:27:42 +0400 Message-id: <1352712469-27974-1-git-send-email-e.voevodin@samsung.com> Subject: [Qemu-devel] [PATCH v6 0/7] TCG global variables clean-up List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: Evgeny Voevodin , blauwirbel@gmail.com, kyungmin.park@samsung.com, edgar.iglesias@gmail.com, aurelien@aurel32.net, rth@twiddle.net This set of patches moves global variables to tcg_ctx: gen_opc_ptr gen_opparam_ptr gen_opc_buf gen_opparam_buf Build tested for all targets. Execution tested on Exynos4210 target. After this patchset was aplied, I noticed 0.7% speed-up of code generation. Probably, this is due to better data caching. Here is the test procedure: 1. Boot Linux Kernel 5 times. 2. For each iteration wait while "JIT cycles" is stable for ~10 seconds 3. Write down the "cycles/op" Here are the results (tested on gcc-4.6): Before clean-up: min: 731.5 max: 734.8 avg: 733.0 standard deviation: ~2 ~= 0.2% Average cycles/op = 733 +- 2 After clean-up: min: 725.0 max: 730.5 avg: 727.8 standard deviation: ~3 ~= 0.4% Average cycles/op = 728 +- 3 Speed-up of TCG code generation = 0.7% Changelog: v5->v6: Fixed broken patches. Rebased. v4->v5: Rebased. Fixed authorship. All patches are reviewed-by Richard Henderson v3->v4: Rebased. Added target-cris/translate.c: Code style clean-up v2->v3: Removed tcg_cur_ctx since it gives slow-down on gcc-4.5. Rebased. v1->v2: Introduced TCGContext *tcg_cur_ctx global to use in those places where we don't have an interface to pass pointer to tcg_ctx. Code style clean-up Evgeny Voevodin (7): target-cris/translate.c: Code style clean-up tcg/tcg.h: Duplicate global TCG variables in TCGContext TCG: Use gen_opc_ptr from context instead of global variable. TCG: Use gen_opparam_ptr from context instead of global variable. TCG: Use gen_opc_buf from context instead of global variable. TCG: Use gen_opparam_buf from context instead of global variable. TCG: Remove unused global variables gen-icount.h | 2 +- target-alpha/translate.c | 10 +- target-arm/translate.c | 10 +- target-cris/translate.c | 5040 +++++++++++++++++++++-------------------- target-i386/translate.c | 10 +- target-lm32/translate.c | 13 +- target-m68k/translate.c | 10 +- target-microblaze/translate.c | 13 +- target-mips/translate.c | 11 +- target-openrisc/translate.c | 13 +- target-ppc/translate.c | 11 +- target-s390x/translate.c | 11 +- target-sh4/translate.c | 10 +- target-sparc/translate.c | 10 +- target-unicore32/translate.c | 10 +- target-xtensa/translate.c | 8 +- tcg/optimize.c | 62 +- tcg/tcg-op.h | 324 +-- tcg/tcg.c | 85 +- tcg/tcg.h | 10 +- translate-all.c | 3 - 21 files changed, 2859 insertions(+), 2817 deletions(-) -- 1.7.9.5