From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:56665) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dHvFK-0004Km-Lz for qemu-devel@nongnu.org; Mon, 05 Jun 2017 12:52:53 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1dHvFI-0003jT-Fz for qemu-devel@nongnu.org; Mon, 05 Jun 2017 12:52:50 -0400 Received: from mail-qt0-x244.google.com ([2607:f8b0:400d:c0d::244]:36125) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1dHvFI-0003hu-5C for qemu-devel@nongnu.org; Mon, 05 Jun 2017 12:52:48 -0400 Received: by mail-qt0-x244.google.com with SMTP id s33so9906426qtg.3 for ; Mon, 05 Jun 2017 09:52:47 -0700 (PDT) Sender: Richard Henderson From: Richard Henderson Date: Mon, 5 Jun 2017 09:52:14 -0700 Message-Id: <20170605165233.4135-8-rth@twiddle.net> In-Reply-To: <20170605165233.4135-1-rth@twiddle.net> References: <20170605165233.4135-1-rth@twiddle.net> Subject: [Qemu-devel] [PULL 07/26] target/arm: optimize indirect branches List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: peter.maydell@linaro.org, "Emilio G. Cota" From: "Emilio G. Cota" Speed up indirect branches by jumping to the target if it is valid. Softmmu measurements (see later commit for user-mode results): Note: baseline (i.e. speedup == 1x) is QEMU v2.9.0. - Impact on Boot time | setup | ARM debian jessie boot+shutdown time | stddev | |--------+--------------------------------------+--------| | v2.9.0 | 8.84 | 0.07 | | +cross | 8.85 | 0.03 | | +jr | 8.83 | 0.06 | - NBench, arm-softmmu (debian jessie guest). Host: Intel i7-4790K @ 4.00GHz 1.3x +-+-------------------------------------------------------------------------------------------------------------+-+ | | | cross #### | 1.25x +cross+jr..........................................................#++#.........................................+-+ | #### # # | | +++# # # # | | +++ **** # # # | 1.2x +-+...................................####............*..*..#......#..#.........................................+-+ | **** # * * # # # #### | | * * # * * # # # # # | 1.15x +-+................................*..*..#............*..*..#......#..#.....#..#................................+-+ | * * # * * # # # # # | | * * # #### * * # # # # # | | * * # # # * * # # # # # #### | 1.1x +-+................................*..*..#......#..#..*..*..#......#..#.....#..#.........................#..#...+-+ | * * # # # * * # # # # # # # | | * * # # # * * # # # # # # # | 1.05x +-+..........................####..*..*..#......#..#..*..*..#......#..#.....#..#......+++............*****..#...+-+ | ***** # * * # # # * * # ***** # # # +++ | ****### * * # | | *+++* # * * # # # * * # *+++* # **** # *****### * * # * * # | | *****### +++#### * * # * * # ***** # * * # * * # * * # * | *++# * * # * * # | 1x +-++-+*+++*-+#++****++#++*+-+*++#+-*++*++#-+*+++*-+#++*++*++#++*+-+*++#+-*++*++#-+*+++*-+#++*++*++#++*+-+*++#+-++-+ | * * # * * # * * # * * # * * # * * # * * # * * # * * # * * # * * # | | * * # * * # * * # * * # * * # * * # * * # * * # * * # * * # * * # | 0.95x +-+---*****###--****###--*****###--****###--*****###--****###--*****###--****###--*****###--****###--*****###---+-+ ASSIGNMENT BITFIELD FOURFP EMULATION HUFFMAN LU DECOMPOSITIONEURAL NNUMERIC SOSTRING SORT hmean png: http://imgur.com/eOLmZNR NB. 'cross' represents the previous commit. Signed-off-by: Emilio G. Cota Message-Id: <1493263764-18657-8-git-send-email-cota@braap.org> [rth: Replace gen_jr global variable with DISAS_EXIT state.] Signed-off-by: Richard Henderson --- target/arm/translate.c | 25 ++++++++++++++++--------- target/arm/translate.h | 4 ++++ 2 files changed, 20 insertions(+), 9 deletions(-) diff --git a/target/arm/translate.c b/target/arm/translate.c index d72953f..0862f9e 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -1182,7 +1182,7 @@ static void gen_exception_internal_insn(DisasContext *s, int offset, int excp) gen_set_condexec(s); gen_set_pc_im(s, s->pc - offset); gen_exception_internal(excp); - s->is_jmp = DISAS_JUMP; + s->is_jmp = DISAS_EXC; } static void gen_exception_insn(DisasContext *s, int offset, int excp, @@ -1191,14 +1191,14 @@ static void gen_exception_insn(DisasContext *s, int offset, int excp, gen_set_condexec(s); gen_set_pc_im(s, s->pc - offset); gen_exception(excp, syn, target_el); - s->is_jmp = DISAS_JUMP; + s->is_jmp = DISAS_EXC; } /* Force a TB lookup after an instruction that changes the CPU state. */ static inline void gen_lookup_tb(DisasContext *s) { tcg_gen_movi_i32(cpu_R[15], s->pc & ~1); - s->is_jmp = DISAS_JUMP; + s->is_jmp = DISAS_EXIT; } static inline void gen_hlt(DisasContext *s, int imm) @@ -4150,19 +4150,23 @@ static inline bool use_goto_tb(DisasContext *s, target_ulong dest) #endif } -static inline void gen_goto_tb(DisasContext *s, int n, target_ulong dest) +static void gen_goto_ptr(void) +{ + TCGv addr = tcg_temp_new(); + tcg_gen_extu_i32_tl(addr, cpu_R[15]); + tcg_gen_lookup_and_goto_ptr(addr); + tcg_temp_free(addr); +} + +static void gen_goto_tb(DisasContext *s, int n, target_ulong dest) { if (use_goto_tb(s, dest)) { tcg_gen_goto_tb(n); gen_set_pc_im(s, dest); tcg_gen_exit_tb((uintptr_t)s->tb + n); } else { - TCGv addr = tcg_temp_new(); - gen_set_pc_im(s, dest); - tcg_gen_extu_i32_tl(addr, cpu_R[15]); - tcg_gen_lookup_and_goto_ptr(addr); - tcg_temp_free(addr); + gen_goto_ptr(); } } @@ -12095,11 +12099,14 @@ void gen_intermediate_code(CPUARMState *env, TranslationBlock *tb) gen_set_pc_im(dc, dc->pc); /* fall through */ case DISAS_JUMP: + gen_goto_ptr(); + break; default: /* indicate that the hash table must be used to find the next TB */ tcg_gen_exit_tb(0); break; case DISAS_TB_JUMP: + case DISAS_EXC: /* nothing more to generate */ break; case DISAS_WFI: diff --git a/target/arm/translate.h b/target/arm/translate.h index 6b2cc34..15d383d 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -139,6 +139,10 @@ static void disas_set_insn_syndrome(DisasContext *s, uint32_t syn) * custom end-of-TB code) */ #define DISAS_BX_EXCRET 11 +/* For instructions which want an immediate exit to the main loop, + * as opposed to attempting to use lookup_and_goto_ptr. + */ +#define DISAS_EXIT 12 #ifdef TARGET_AARCH64 void a64_translate_init(void); -- 2.9.4