qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Richard Henderson <rth@twiddle.net>
To: qemu-devel@nongnu.org
Cc: Blue Swirl <blauwirbel@gmail.com>
Subject: [Qemu-devel] [PATCH 05/12] tcg-sparc: Fix setcond
Date: Wed, 26 Sep 2012 18:55:35 -0700	[thread overview]
Message-ID: <1348710942-3040-6-git-send-email-rth@twiddle.net> (raw)
In-Reply-To: <1348710942-3040-1-git-send-email-rth@twiddle.net>

The set of comparisons that can immediately use the carry are LTU/GEU,
not LTU/LEU.  Don't swap operands when we need a temp register; the
register may already be in use from setcond2.

Signed-off-by: Richard Henderson <rth@twiddle.net>
---
 tcg/sparc/tcg-target.c | 33 +++++++++++++++++++--------------
 1 file changed, 19 insertions(+), 14 deletions(-)

diff --git a/tcg/sparc/tcg-target.c b/tcg/sparc/tcg-target.c
index e55b792..c3e15b9 100644
--- a/tcg/sparc/tcg-target.c
+++ b/tcg/sparc/tcg-target.c
@@ -617,32 +617,37 @@ static void tcg_out_brcond2_i32(TCGContext *s, TCGCond cond,
 static void tcg_out_setcond_i32(TCGContext *s, TCGCond cond, TCGArg ret,
                                 TCGArg c1, TCGArg c2, int c2const)
 {
-    TCGArg t;
-
     /* For 32-bit comparisons, we can play games with ADDX/SUBX.  */
     switch (cond) {
+    case TCG_COND_LTU:
+    case TCG_COND_GEU:
+        /* The result of the comparison is in the carry bit.  */
+        break;
+
     case TCG_COND_EQ:
     case TCG_COND_NE:
+        /* For equality, we can transform to inequality vs zero.  */
         if (c2 != 0) {
             tcg_out_arithc(s, ret, c1, c2, c2const, ARITH_XOR);
         }
         c1 = TCG_REG_G0, c2 = ret, c2const = 0;
-        cond = (cond == TCG_COND_EQ ? TCG_COND_LEU : TCG_COND_LTU);
+        cond = (cond == TCG_COND_EQ ? TCG_COND_GEU : TCG_COND_LTU);
 	break;
 
     case TCG_COND_GTU:
-    case TCG_COND_GEU:
-        if (c2const && c2 != 0) {
-            tcg_out_movi_imm13(s, TCG_REG_T1, c2);
-            c2 = TCG_REG_T1;
-        }
-        t = c1, c1 = c2, c2 = t, c2const = 0;
-        cond = tcg_swap_cond(cond);
-        break;
-
-    case TCG_COND_LTU:
     case TCG_COND_LEU:
-        break;
+        /* If we don't need to load a constant into a register, we can
+           swap the operands on GTU/LEU.  There's no benefit to loading
+           the constant into a temporary register.  */
+        if (!c2const || c2 == 0) {
+            TCGArg t = c1;
+            c1 = c2;
+            c2 = t;
+            c2const = 0;
+            cond = tcg_swap_cond(cond);
+            break;
+        }
+        /* FALLTHRU */
 
     default:
         tcg_out_cmp(s, c1, c2, c2const);
-- 
1.7.11.4

  parent reply	other threads:[~2012-09-27  1:55 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-09-27  1:55 [Qemu-devel] [PATCH 00/12] tcg-sparc fixes and improvements Richard Henderson
2012-09-27  1:55 ` [Qemu-devel] [PATCH 01/12] tcg-sparc: Fix brcond2 Richard Henderson
2012-09-27  1:55 ` [Qemu-devel] [PATCH 02/12] tcg-sparc: Implement movcond Richard Henderson
2012-09-27  1:55 ` [Qemu-devel] [PATCH 03/12] tcg-sparc: Fix setcond2 Richard Henderson
2012-09-27  1:55 ` [Qemu-devel] [PATCH 04/12] tcg-sparc: Fix qemu_st for 32-bit Richard Henderson
2012-09-27  1:55 ` Richard Henderson [this message]
2012-09-27  1:55 ` [Qemu-devel] [PATCH 06/12] tcg-sparc: Fix add2/sub2 Richard Henderson
2012-09-27  1:55 ` [Qemu-devel] [PATCH 07/12] tcg-sparc: Use Z constraint for %g0 Richard Henderson
2012-09-29 11:55   ` Blue Swirl
2012-09-27  1:55 ` [Qemu-devel] [PATCH 08/12] tcg-sparc: Optimize setcond2 equality compare with 0 Richard Henderson
2012-09-27  1:55 ` [Qemu-devel] [PATCH 09/12] tcg-sparc: Drop use of Bicc in favor of BPcc Richard Henderson
2012-09-29 12:07   ` Blue Swirl
2012-09-29 17:21     ` Richard Henderson
2012-09-27  1:55 ` [Qemu-devel] [PATCH 10/12] tcg-sparc: Dump illegal opode contents Richard Henderson
2012-09-27  1:55 ` [Qemu-devel] [PATCH 11/12] tcg-sparc: Emit BPr insns for brcond_i64 Richard Henderson
2012-09-27  1:55 ` [Qemu-devel] [PATCH 12/12] tcg-sparc: Emit MOVR insns for setcond_i64 and movcond_64 Richard Henderson

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=1348710942-3040-6-git-send-email-rth@twiddle.net \
    --to=rth@twiddle.net \
    --cc=blauwirbel@gmail.com \
    --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).