Patchwork [ARM] VFP index range

login
register
mail settings
Submitter Chung-Lin Tang
Date Dec. 17, 2010, 5:14 a.m.
Message ID <4D0AF1C4.9000706@codesourcery.com>
Download mbox | patch
Permalink /patch/75835/
State New
Headers show

Comments

Chung-Lin Tang - Dec. 17, 2010, 5:14 a.m.
Hi Richard,
this is a VFP load/store index range patch that I submitted last
year[1], which you approved at that time, but never got committed.

I've added the Thumb-2 bits I missed then, corrected some stuff in the
testcase, and re-submitting now. Re-tested without regressions. Is this
ok for trunk?

Thanks,
Chung-Lin

[1] http://gcc.gnu.org/ml/gcc-patches/2009-07/msg01844.html


2010-12-17  Chung-Lin Tang  <cltang@codesourcery.com>

	gcc/
	* config/arm/arm.c
	(arm_legitimate_index_p): Add VFP load/store index range case.
	Change to SF/DFmode tests to avoid capturing HFmode.
	(thumb2_legitimate_index_p): Same.

	gcc/testsuite/
	* gcc.target/arm/vfp-1.c (test_ldst): New test for VFP
	load/store immediate indexes.
Richard Earnshaw - Dec. 17, 2010, 9:24 a.m.
OK.

R.

On Fri, 2010-12-17 at 13:14 +0800, Chung-Lin Tang wrote:
> Hi Richard,
> this is a VFP load/store index range patch that I submitted last
> year[1], which you approved at that time, but never got committed.
> 
> I've added the Thumb-2 bits I missed then, corrected some stuff in the
> testcase, and re-submitting now. Re-tested without regressions. Is this
> ok for trunk?
> 
> Thanks,
> Chung-Lin
> 
> [1] http://gcc.gnu.org/ml/gcc-patches/2009-07/msg01844.html
> 
> 
> 2010-12-17  Chung-Lin Tang  <cltang@codesourcery.com>
> 
> 	gcc/
> 	* config/arm/arm.c
> 	(arm_legitimate_index_p): Add VFP load/store index range case.
> 	Change to SF/DFmode tests to avoid capturing HFmode.
> 	(thumb2_legitimate_index_p): Same.
> 
> 	gcc/testsuite/
> 	* gcc.target/arm/vfp-1.c (test_ldst): New test for VFP
> 	load/store immediate indexes.
Ramana Radhakrishnan - Jan. 15, 2011, 4:03 p.m.
On Fri, Dec 17, 2010 at 5:14 AM, Chung-Lin Tang <cltang@codesourcery.com> wrote:
> Hi Richard,
> this is a VFP load/store index range patch that I submitted last
> year[1], which you approved at that time, but never got committed.
>
> I've added the Thumb-2 bits I missed then, corrected some stuff in the
> testcase, and re-submitting now. Re-tested without regressions. Is this
> ok for trunk?

This caused http://gcc.gnu.org/PR47246


cheers
Ramana

>
> Thanks,
> Chung-Lin
>
> [1] http://gcc.gnu.org/ml/gcc-patches/2009-07/msg01844.html
>
>
> 2010-12-17  Chung-Lin Tang  <cltang@codesourcery.com>
>
>        gcc/
>        * config/arm/arm.c
>        (arm_legitimate_index_p): Add VFP load/store index range case.
>        Change to SF/DFmode tests to avoid capturing HFmode.
>        (thumb2_legitimate_index_p): Same.
>
>        gcc/testsuite/
>        * gcc.target/arm/vfp-1.c (test_ldst): New test for VFP
>        load/store immediate indexes.
>

Patch

Index: config/arm/arm.c
===================================================================
--- config/arm/arm.c	(revision 167900)
+++ config/arm/arm.c	(working copy)
@@ -5647,8 +5647,8 @@ 
 
   /* Standard coprocessor addressing modes.  */
   if (TARGET_HARD_FLOAT
-      && (TARGET_FPA || TARGET_MAVERICK)
-      && (GET_MODE_CLASS (mode) == MODE_FLOAT
+      && (TARGET_VFP || TARGET_FPA || TARGET_MAVERICK)
+      && (mode == SFmode || mode == DFmode
 	  || (TARGET_MAVERICK && mode == DImode)))
     return (code == CONST_INT && INTVAL (index) < 1024
 	    && INTVAL (index) > -1024
@@ -5768,8 +5768,8 @@ 
   /* ??? Combine arm and thumb2 coprocessor addressing modes.  */
   /* Standard coprocessor addressing modes.  */
   if (TARGET_HARD_FLOAT
-      && (TARGET_FPA || TARGET_MAVERICK)
-      && (GET_MODE_CLASS (mode) == MODE_FLOAT
+      && (TARGET_VFP || TARGET_FPA || TARGET_MAVERICK)
+      && (mode == SFmode || mode == DFmode
 	  || (TARGET_MAVERICK && mode == DImode)))
     return (code == CONST_INT && INTVAL (index) < 1024
 	    && INTVAL (index) > -1024
Index: testsuite/gcc.target/arm/vfp-1.c
===================================================================
--- testsuite/gcc.target/arm/vfp-1.c	(revision 167900)
+++ testsuite/gcc.target/arm/vfp-1.c	(working copy)
@@ -125,3 +125,15 @@ 
   d1 = u1;
 }
 
+void test_ldst (float f[], double d[]) {
+  /* { dg-final { scan-assembler "flds.+ \\\[r0, #1020\\\]" } } */
+  /* { dg-final { scan-assembler "flds.+ \\\[r0, #-1020\\\]" } } */
+  /* { dg-final { scan-assembler "add.+ r0, #1024" } } */
+  /* { dg-final { scan-assembler "fsts.+ \\\[r0, #0\\\]$" } } */
+  f[256] = f[255] + f[-255];
+
+  /* { dg-final { scan-assembler "fldd.+ \\\[r1, #1016\\\]" } } */
+  /* { dg-final { scan-assembler "fldd.+ \\\[r1, #-1016\\\]" } } */
+  /* { dg-final { scan-assembler "fstd.+ \\\[r1, #256\\\]" } } */
+  d[32] = d[127] + d[-127];
+}