| Message ID | 000801d07cd8$42809e90$c781dbb0$@arm.com |
|---|---|
| State | New |
| Headers | show |
On 22/04/15 09:42, Bin Cheng wrote: > Hi, > Case pr42172-1.c failed on pre-armv7 processors because GCC actually > generates better code without ldr instruction. This patch just refines test > case by checking str instead of ldr, makes sure the case passes on all arm > processors. In the end, we need to fix GCC combiner to generate optimal > code on armv7 processors too. PR42172 is kept open for that purpose. > > This is obvious change, is it OK for branches too? For the record, for -mcpu=arm7tdmi we now generate: init_A: mov r3, #8 strb r3, [r0] bx Ok for trunk. Is this test failing on the branches too? Kyrill > > gcc/testsuite/ChangeLog > 2015-04-22 Bin Cheng <bin.cheng@arm.com> > > * gcc.target/arm/pr42172-1.c: Check str instead of ldr. >
On Thu, Apr 23, 2015 at 4:19 PM, Kyrill Tkachov <kyrylo.tkachov@arm.com> wrote: > > On 22/04/15 09:42, Bin Cheng wrote: >> >> Hi, >> Case pr42172-1.c failed on pre-armv7 processors because GCC actually >> generates better code without ldr instruction. This patch just refines >> test >> case by checking str instead of ldr, makes sure the case passes on all arm >> processors. In the end, we need to fix GCC combiner to generate optimal >> code on armv7 processors too. PR42172 is kept open for that purpose. >> >> This is obvious change, is it OK for branches too? > > > For the record, for -mcpu=arm7tdmi we now generate: > init_A: > mov r3, #8 > strb r3, [r0] > bx That's the point, it generates optimal code for pre-armv7 processors, which is treated as failure now. > > > Ok for trunk. Is this test failing on the branches too? Well, at least for gcc-5-branch. > > Kyrill > > >> >> gcc/testsuite/ChangeLog >> 2015-04-22 Bin Cheng <bin.cheng@arm.com> >> >> * gcc.target/arm/pr42172-1.c: Check str instead of ldr. >> >
On 23/04/15 09:53, Bin.Cheng wrote: > On Thu, Apr 23, 2015 at 4:19 PM, Kyrill Tkachov <kyrylo.tkachov@arm.com> wrote: >> On 22/04/15 09:42, Bin Cheng wrote: >>> Hi, >>> Case pr42172-1.c failed on pre-armv7 processors because GCC actually >>> generates better code without ldr instruction. This patch just refines >>> test >>> case by checking str instead of ldr, makes sure the case passes on all arm >>> processors. In the end, we need to fix GCC combiner to generate optimal >>> code on armv7 processors too. PR42172 is kept open for that purpose. >>> >>> This is obvious change, is it OK for branches too? >> >> For the record, for -mcpu=arm7tdmi we now generate: >> init_A: >> mov r3, #8 >> strb r3, [r0] >> bx > That's the point, it generates optimal code for pre-armv7 processors, > which is treated as failure now. > >> >> Ok for trunk. Is this test failing on the branches too? > Well, at least for gcc-5-branch. If the GCC 5 branch is now open for commits (I think it is?), then ok there too. Kyrill > >> Kyrill >> >> >>> gcc/testsuite/ChangeLog >>> 2015-04-22 Bin Cheng <bin.cheng@arm.com> >>> >>> * gcc.target/arm/pr42172-1.c: Check str instead of ldr. >>>
diff --git a/gcc/testsuite/gcc.target/arm/pr42172-1.c b/gcc/testsuite/gcc.target/arm/pr42172-1.c index 207f600..a3de02f 100644 --- a/gcc/testsuite/gcc.target/arm/pr42172-1.c +++ b/gcc/testsuite/gcc.target/arm/pr42172-1.c @@ -16,4 +16,4 @@ void init_A (struct A *this) this->f4 = 0; } -/* { dg-final { scan-assembler-times "ldr" 1 } } */ +/* { dg-final { scan-assembler-times "str" 1 } } */