qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Richard Henderson <richard.henderson@linaro.org>
To: qemu-devel@nongnu.org
Cc: Peter Maydell <peter.maydell@linaro.org>
Subject: [PULL 35/48] tcg/i386: Use CMP+SBB in tcg_out_setcond
Date: Wed, 23 Aug 2023 13:23:13 -0700	[thread overview]
Message-ID: <20230823202326.1353645-36-richard.henderson@linaro.org> (raw)
In-Reply-To: <20230823202326.1353645-1-richard.henderson@linaro.org>

Use the carry bit to optimize some forms of setcond.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/i386/tcg-target.c.inc | 50 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)

diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc
index 1542afd94d..4d7b745a52 100644
--- a/tcg/i386/tcg-target.c.inc
+++ b/tcg/i386/tcg-target.c.inc
@@ -1531,6 +1531,56 @@ static void tcg_out_setcond(TCGContext *s, int rexw, TCGCond cond,
                             TCGArg dest, TCGArg arg1, TCGArg arg2,
                             int const_arg2)
 {
+    bool inv = false;
+
+    switch (cond) {
+    case TCG_COND_NE:
+        inv = true;
+        /* fall through */
+    case TCG_COND_EQ:
+        /* If arg2 is 0, convert to LTU/GEU vs 1. */
+        if (const_arg2 && arg2 == 0) {
+            arg2 = 1;
+            goto do_ltu;
+        }
+        break;
+
+    case TCG_COND_LEU:
+        inv = true;
+        /* fall through */
+    case TCG_COND_GTU:
+        /* If arg2 is a register, swap for LTU/GEU. */
+        if (!const_arg2) {
+            TCGReg t = arg1;
+            arg1 = arg2;
+            arg2 = t;
+            goto do_ltu;
+        }
+        break;
+
+    case TCG_COND_GEU:
+        inv = true;
+        /* fall through */
+    case TCG_COND_LTU:
+    do_ltu:
+        /*
+         * Relying on the carry bit, use SBB to produce -1 if LTU, 0 if GEU.
+         * We can then use NEG or INC to produce the desired result.
+         * This is always smaller than the SETCC expansion.
+         */
+        tcg_out_cmp(s, arg1, arg2, const_arg2, rexw);
+        tgen_arithr(s, ARITH_SBB, dest, dest);              /* T:-1 F:0 */
+        if (inv) {
+            tgen_arithi(s, ARITH_ADD, dest, 1, 0);          /* T:0  F:1 */
+        } else {
+            tcg_out_modrm(s, OPC_GRP3_Ev, EXT3_NEG, dest);  /* T:1  F:0 */
+        }
+        return;
+
+    default:
+        break;
+    }
+
     tcg_out_cmp(s, arg1, arg2, const_arg2, rexw);
     tcg_out_modrm(s, OPC_SETCC | tcg_cond_to_jcc[cond], 0, dest);
     tcg_out_ext8u(s, dest, dest);
-- 
2.34.1



  parent reply	other threads:[~2023-08-23 20:27 UTC|newest]

Thread overview: 50+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-08-23 20:22 [PULL 00/48] tcg patch queue Richard Henderson
2023-08-23 20:22 ` [PULL 01/48] accel/kvm: Widen pc/saved_insn for kvm_sw_breakpoint Richard Henderson
2023-08-23 20:22 ` [PULL 02/48] accel/hvf: Widen pc/saved_insn for hvf_sw_breakpoint Richard Henderson
2023-08-23 20:22 ` [PULL 03/48] sysemu/kvm: Use vaddr for kvm_arch_[insert|remove]_hw_breakpoint Richard Henderson
2023-08-23 20:22 ` [PULL 04/48] sysemu/hvf: Use vaddr for hvf_arch_[insert|remove]_hw_breakpoint Richard Henderson
2023-08-23 20:22 ` [PULL 05/48] include/exec: Replace target_ulong with abi_ptr in cpu_[st|ld]*() Richard Henderson
2023-08-23 20:22 ` [PULL 06/48] include/exec: typedef abi_ptr to vaddr in softmmu Richard Henderson
2023-08-23 20:22 ` [PULL 07/48] include/exec: Widen tlb_hit/tlb_hit_page() Richard Henderson
2023-08-23 20:22 ` [PULL 08/48] accel/tcg: Widen address arg in tlb_compare_set() Richard Henderson
2023-08-23 20:22 ` [PULL 09/48] accel/tcg: Update run_on_cpu_data static assert Richard Henderson
2023-08-23 20:22 ` [PULL 10/48] target/m68k: Use tcg_gen_deposit_i32 in gen_partset_reg Richard Henderson
2023-08-23 20:22 ` [PULL 11/48] tcg/i386: Drop BYTEH deposits for 64-bit Richard Henderson
2023-08-23 20:22 ` [PULL 12/48] tcg: Fold deposit with zero to and Richard Henderson
2023-08-23 20:22 ` [PULL 13/48] tcg/i386: Allow immediate as input to deposit_* Richard Henderson
2023-08-23 20:22 ` [PULL 14/48] docs/devel/tcg-ops: Bury mentions of trunc_shr_i64_i32() Richard Henderson
2023-08-23 20:22 ` [PULL 15/48] tcg: Unify TCG_TARGET_HAS_extr[lh]_i64_i32 Richard Henderson
2023-08-23 20:22 ` [PULL 16/48] tcg: Introduce negsetcond opcodes Richard Henderson
2023-08-23 20:22 ` [PULL 17/48] tcg: Use tcg_gen_negsetcond_* Richard Henderson
2023-08-23 20:22 ` [PULL 18/48] target/alpha: Use tcg_gen_movcond_i64 in gen_fold_mzero Richard Henderson
2023-08-23 20:22 ` [PULL 19/48] target/arm: Use tcg_gen_negsetcond_* Richard Henderson
2023-08-23 20:22 ` [PULL 20/48] target/m68k: " Richard Henderson
2023-08-23 20:22 ` [PULL 21/48] target/openrisc: " Richard Henderson
2023-08-23 20:23 ` [PULL 22/48] target/ppc: " Richard Henderson
2023-08-23 20:23 ` [PULL 23/48] target/sparc: Use tcg_gen_movcond_i64 in gen_edge Richard Henderson
2023-08-23 20:23 ` [PULL 24/48] target/tricore: Replace gen_cond_w with tcg_gen_negsetcond_tl Richard Henderson
2023-08-23 20:23 ` [PULL 25/48] tcg/ppc: Implement negsetcond_* Richard Henderson
2023-08-23 20:23 ` [PULL 26/48] tcg/ppc: Use the Set Boolean Extension Richard Henderson
2023-08-23 20:23 ` [PULL 27/48] tcg/aarch64: Implement negsetcond_* Richard Henderson
2023-08-23 20:23 ` [PULL 28/48] tcg/arm: Implement negsetcond_i32 Richard Henderson
2023-08-23 20:23 ` [PULL 29/48] tcg/riscv: Implement negsetcond_* Richard Henderson
2023-08-23 20:23 ` [PULL 30/48] tcg/s390x: " Richard Henderson
2023-08-23 20:23 ` [PULL 31/48] tcg/sparc64: " Richard Henderson
2023-08-23 20:23 ` [PULL 32/48] tcg/i386: Merge tcg_out_brcond{32,64} Richard Henderson
2023-08-23 20:23 ` [PULL 33/48] tcg/i386: Merge tcg_out_setcond{32,64} Richard Henderson
2023-08-23 20:23 ` [PULL 34/48] tcg/i386: Merge tcg_out_movcond{32,64} Richard Henderson
2023-08-23 20:23 ` Richard Henderson [this message]
2023-08-23 20:23 ` [PULL 36/48] tcg/i386: Clear dest first in tcg_out_setcond if possible Richard Henderson
2023-08-23 20:23 ` [PULL 37/48] tcg/i386: Use shift in tcg_out_setcond Richard Henderson
2023-08-23 20:23 ` [PULL 38/48] tcg/i386: Implement negsetcond_* Richard Henderson
2023-08-23 20:23 ` [PULL 39/48] tcg/tcg-op: Document bswap16_i32() byte pattern Richard Henderson
2023-08-23 20:23 ` [PULL 40/48] tcg/tcg-op: Document bswap16_i64() " Richard Henderson
2023-08-23 20:23 ` [PULL 41/48] tcg/tcg-op: Document bswap32_i32() " Richard Henderson
2023-08-23 20:23 ` [PULL 42/48] tcg/tcg-op: Document bswap32_i64() " Richard Henderson
2023-08-23 20:23 ` [PULL 43/48] tcg/tcg-op: Document bswap64_i64() " Richard Henderson
2023-08-23 20:23 ` [PULL 44/48] tcg/tcg-op: Document hswap_i32/64() " Richard Henderson
2023-08-23 20:23 ` [PULL 45/48] tcg/tcg-op: Document wswap_i64() " Richard Henderson
2023-08-23 20:23 ` [PULL 46/48] target/cris: Fix a typo in gen_swapr() Richard Henderson
2023-08-23 20:23 ` [PULL 47/48] docs/devel/tcg-ops: fix missing newlines in "Host vector operations" Richard Henderson
2023-08-23 20:23 ` [PULL 48/48] tcg: spelling fixes Richard Henderson
2023-08-24 14:05 ` [PULL 00/48] tcg patch queue Stefan Hajnoczi

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230823202326.1353645-36-richard.henderson@linaro.org \
    --to=richard.henderson@linaro.org \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).