Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43)
	id 1JzrXR-0001WR-FE
	for qemu-devel@nongnu.org; Sat, 24 May 2008 07:03:49 -0400
Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43)
	id 1JzrXQ-0001Vu-Gt
	for qemu-devel@nongnu.org; Sat, 24 May 2008 07:03:49 -0400
Received: from [199.232.76.173] (port=54795 helo=monty-python.gnu.org)
	by lists.gnu.org with esmtp (Exim 4.43) id 1JzrXQ-0001Vr-A5
	for qemu-devel@nongnu.org; Sat, 24 May 2008 07:03:48 -0400
Received: from ug-out-1314.google.com ([66.249.92.170]:55466)
	by monty-python.gnu.org with esmtp (Exim 4.60)
	(envelope-from <rdsandiford@googlemail.com>) id 1JzrXP-0005JO-Qs
	for qemu-devel@nongnu.org; Sat, 24 May 2008 07:03:48 -0400
Received: by ug-out-1314.google.com with SMTP id j40so37542ugd.4
	for <qemu-devel@nongnu.org>; Sat, 24 May 2008 04:03:46 -0700 (PDT)
From: Richard Sandiford <rdsandiford@googlemail.com>
Date: Sat, 24 May 2008 12:03:41 +0100
Message-ID: <87bq2w6k1e.fsf@firetop.home>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Subject: [Qemu-devel] [PATCH] Fix mov[tf].ps handling for MIPS
Reply-To: qemu-devel@nongnu.org
List-Id: qemu-devel.nongnu.org
List-Unsubscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.gnu.org/pipermail/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: qemu-devel@nongnu.org

    MOVT.PS $fd, $fs, $fccX

is curently treated like MOV*.D and MOV*.S; it moves $fs into $fd
if $fccX is true.  But the instruction actually moves each float
element separately; the lower move is conditional on $fccX and
the upper one is conditional on $fccX+1.  Same for MOVF.PS.

The patch below seems to work for me.  Please install if OK.

(I realise the MOVCF.FMT stuff ought to be TCG-ified eventually,
but hopefully this patch is strict progress.)

Richard


Index: qemu/target-mips/op.c
===================================================================
--- qemu.orig/target-mips/op.c	2008-05-24 09:33:33.000000000 +0100
+++ qemu/target-mips/op.c	2008-05-24 09:35:11.000000000 +0100
@@ -1632,10 +1632,11 @@ FLOAT_OP(movf, s)
 }
 FLOAT_OP(movf, ps)
 {
-    if (!(env->fpu->fcr31 & PARAM1)) {
+    unsigned int mask = GET_FP_COND (env->fpu) >> PARAM1;
+    if (!(mask & 1))
         WT2 = WT0;
+    if (!(mask & 2))
         WTH2 = WTH0;
-    }
     DEBUG_FPU_STATE();
     FORCE_RET();
 }
@@ -1655,10 +1656,11 @@ FLOAT_OP(movt, s)
 }
 FLOAT_OP(movt, ps)
 {
-    if (env->fpu->fcr31 & PARAM1) {
+    unsigned int mask = GET_FP_COND (env->fpu) >> PARAM1;
+    if (mask & 1)
         WT2 = WT0;
+    if (mask & 2)
         WTH2 = WTH0;
-    }
     DEBUG_FPU_STATE();
     FORCE_RET();
 }
Index: qemu/target-mips/translate.c
===================================================================
--- qemu.orig/target-mips/translate.c	2008-05-24 09:35:37.000000000 +0100
+++ qemu/target-mips/translate.c	2008-05-24 11:05:46.000000000 +0100
@@ -5592,7 +5592,6 @@ static void glue(gen_movcf_, fmt) (Disas
 }
 GEN_MOVCF(d);
 GEN_MOVCF(s);
-GEN_MOVCF(ps);
 #undef GEN_MOVCF
 
 static void gen_farith (DisasContext *ctx, uint32_t op1,
@@ -6211,7 +6210,10 @@ static void gen_farith (DisasContext *ct
         GEN_LOAD_FREG_FTN(WTH0, fs);
         GEN_LOAD_FREG_FTN(WT2, fd);
         GEN_LOAD_FREG_FTN(WTH2, fd);
-        gen_movcf_ps(ctx, (ft >> 2) & 0x7, ft & 0x1);
+        if (ft & 0x1)
+            gen_op_float_movt_ps ((ft >> 2) & 0x7);
+        else
+            gen_op_float_movf_ps ((ft >> 2) & 0x7);
         GEN_STORE_FTN_FREG(fd, WT2);
         GEN_STORE_FTN_FREG(fd, WTH2);
         opn = "movcf.ps";