From patchwork Tue Feb 19 17:40:01 2013 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Richard Henderson X-Patchwork-Id: 221766 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Received: from lists.gnu.org (lists.gnu.org [208.118.235.17]) (using TLSv1 with cipher AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTPS id 36D882C0077 for ; Wed, 20 Feb 2013 05:36:04 +1100 (EST) Received: from localhost ([::1]:43794 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1U7s2r-0001wy-Mn for incoming@patchwork.ozlabs.org; Tue, 19 Feb 2013 13:36:01 -0500 Received: from eggs.gnu.org ([208.118.235.92]:37506) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1U7s2B-0000Zd-AX for qemu-devel@nongnu.org; Tue, 19 Feb 2013 13:35:29 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1U7rC4-00087Q-9B for qemu-devel@nongnu.org; Tue, 19 Feb 2013 12:41:51 -0500 Received: from mail-da0-f42.google.com ([209.85.210.42]:59429) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1U7rC3-00086o-Td for qemu-devel@nongnu.org; Tue, 19 Feb 2013 12:41:28 -0500 Received: by mail-da0-f42.google.com with SMTP id z17so3107740dal.15 for ; Tue, 19 Feb 2013 09:41:26 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=x-received:sender:from:to:cc:subject:date:message-id:x-mailer :in-reply-to:references; bh=/W+rkCzOM7DXQe5HrXMWFDOBgLRmnvmwNNvtaUEgC9A=; b=NOrzVWOCJMqxnt1Bc37ErgsgUjX6/0ugvzWSW/6Y2wT6KQGL/ZAOd7M88+SRUZrQ40 DcdcJhJFOB+JZUHGEeAh6VceF+YLKDXHhuLWExWRffqXwCqKhdWiT0WYckV1BF5NlGbK aSPmqGYuZQFXYybcE/O9mM36cGnkj5/TgSJpeeG0uKEXBtcNkEqtVi8MdAjYyoba+CqI fPh69jNjj1zU66a1E/Fh5OhkxKHATekJCCHpO7foqaJq6tTfsvAVmIqX29Sen6eCu/35 gvO6UkSmkAmlO9Vo7ythG3i3CAZqveMd5k460SmXYR+0rLZuTJLnCwnfXJv0bG5WwZqT CvmA== X-Received: by 10.66.76.41 with SMTP id h9mr47792145paw.1.1361295686690; Tue, 19 Feb 2013 09:41:26 -0800 (PST) Received: from anchor.twiddle.net (50-194-63-110-static.hfc.comcastbusiness.net. [50.194.63.110]) by mx.google.com with ESMTPS id 1sm18659295pbg.18.2013.02.19.09.41.24 (version=TLSv1.2 cipher=RC4-SHA bits=128/128); Tue, 19 Feb 2013 09:41:25 -0800 (PST) From: Richard Henderson To: qemu-devel@nongnu.org Date: Tue, 19 Feb 2013 09:40:01 -0800 Message-Id: <1361295631-21316-28-git-send-email-rth@twiddle.net> X-Mailer: git-send-email 1.8.1.2 In-Reply-To: <1361295631-21316-1-git-send-email-rth@twiddle.net> References: <1361295631-21316-1-git-send-email-rth@twiddle.net> X-detected-operating-system: by eggs.gnu.org: GNU/Linux 3.x [fuzzy] X-Received-From: 209.85.210.42 Cc: blauwirbel@gmail.com, pbonzini@redhat.com, afaerber@suse.de, aurelien@aurel32.net Subject: [Qemu-devel] [PATCH 27/57] target-i386: optimize setcc instructions X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org Sender: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org From: Paolo Bonzini Reconstruct the arguments for complex conditions involving CC_OP_SUBx (BE, L, LE). In the others do it via setcond and gen_setcc_slow (which is not that slow in many cases). Signed-off-by: Paolo Bonzini Signed-off-by: Richard Henderson --- target-i386/translate.c | 95 +++++++++++++++++++------------------------------ 1 file changed, 37 insertions(+), 58 deletions(-) diff --git a/target-i386/translate.c b/target-i386/translate.c index fea43c7..5c9211f 100644 --- a/target-i386/translate.c +++ b/target-i386/translate.c @@ -1090,55 +1090,55 @@ static void gen_setcc_slow(DisasContext *s, int jcc_op, TCGv reg, bool inv) } } -/* return true if setcc_slow is not needed (WARNING: must be kept in - sync with gen_jcc1) */ -static int is_fast_jcc_case(DisasContext *s, int b) +/* perform a conditional store into register 'reg' according to jump opcode + value 'b'. In the fast case, T0 is guaranted not to be used. */ +static inline void gen_setcc1(DisasContext *s, int b, TCGv reg) { - int jcc_op; + int inv, jcc_op, size, cond; + TCGv t0; + + inv = b & 1; jcc_op = (b >> 1) & 7; - switch(s->cc_op) { - /* we optimize the cmp/jcc case */ + + switch (s->cc_op) { + /* we optimize relational operators for the cmp/jcc case */ case CC_OP_SUBB: case CC_OP_SUBW: case CC_OP_SUBL: case CC_OP_SUBQ: - if (jcc_op == JCC_O || jcc_op == JCC_P) - goto slow_jcc; - break; - - /* some jumps are easy to compute */ - case CC_OP_ADDB: - case CC_OP_ADDW: - case CC_OP_ADDL: - case CC_OP_ADDQ: - - case CC_OP_LOGICB: - case CC_OP_LOGICW: - case CC_OP_LOGICL: - case CC_OP_LOGICQ: - - case CC_OP_INCB: - case CC_OP_INCW: - case CC_OP_INCL: - case CC_OP_INCQ: + size = s->cc_op - CC_OP_SUBB; + switch (jcc_op) { + case JCC_BE: + cond = inv ? TCG_COND_GTU : TCG_COND_LEU; + tcg_gen_add_tl(cpu_tmp4, cpu_cc_dst, cpu_cc_src); + gen_extu(size, cpu_tmp4); + t0 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, false); + tcg_gen_setcond_tl(cond, reg, cpu_tmp4, t0); + break; - case CC_OP_DECB: - case CC_OP_DECW: - case CC_OP_DECL: - case CC_OP_DECQ: + case JCC_L: + cond = inv ? TCG_COND_GE : TCG_COND_LT; + goto fast_jcc_l; + case JCC_LE: + cond = inv ? TCG_COND_GT : TCG_COND_LE; + fast_jcc_l: + tcg_gen_add_tl(cpu_tmp4, cpu_cc_dst, cpu_cc_src); + gen_exts(size, cpu_tmp4); + t0 = gen_ext_tl(cpu_tmp0, cpu_cc_src, size, true); + tcg_gen_setcond_tl(cond, reg, cpu_tmp4, t0); + break; - case CC_OP_SHLB: - case CC_OP_SHLW: - case CC_OP_SHLL: - case CC_OP_SHLQ: - if (jcc_op != JCC_Z && jcc_op != JCC_S) + default: goto slow_jcc; + } break; + default: slow_jcc: - return 0; + /* gen_setcc_slow actually generates good code for JC, JZ and JS */ + gen_setcc_slow(s, jcc_op, reg, inv); + break; } - return 1; } /* generate a conditional jump to label 'l1' according to jump opcode @@ -2487,28 +2487,7 @@ static inline void gen_jcc(DisasContext *s, int b, static void gen_setcc(DisasContext *s, int b) { - int inv, jcc_op, l1; - TCGv t0; - - if (is_fast_jcc_case(s, b)) { - /* nominal case: we use a jump */ - /* XXX: make it faster by adding new instructions in TCG */ - t0 = tcg_temp_local_new(); - tcg_gen_movi_tl(t0, 0); - l1 = gen_new_label(); - gen_jcc1(s, b ^ 1, l1); - tcg_gen_movi_tl(t0, 1); - gen_set_label(l1); - tcg_gen_mov_tl(cpu_T[0], t0); - tcg_temp_free(t0); - } else { - /* slow case: it is more efficient not to generate a jump, - although it is questionnable whether this optimization is - worth to */ - inv = b & 1; - jcc_op = (b >> 1) & 7; - gen_setcc_slow(s, jcc_op, cpu_T[0], inv); - } + gen_setcc1(s, b, cpu_T[0]); } static inline void gen_op_movl_T0_seg(int seg_reg)