| Message ID | 0DA23CC379F5F945ACB41CF394B98277210EEC76@LEMAIL01.le.imgtec.org |
|---|---|
| State | New |
| Headers | show |
Andrew Bennett <Andrew.Bennett@imgtec.com> writes: > The stack-1.c testcase fails when being compiled for micromips with the > -O0 optimization level. The reason is the testcase is expecting the > following sequence at the end of the function: > > addiu $sp,$sp,16 > jrc $31 > > But for micromips it generates the following: > > jraddiusp 16 > > > As the failure only happens at one optimization level I have decided to > just change the expected output rather than creating a separate > micromips testcase. I'm not sure this is the right approach here. If we get a jraddiusp then the problem that the test is trying to cover can't possibly happen anyway. (The test is checking if a load and final stack adjustment are ever re-ordered from what I can see.) I'd just mark the test as NOCOMPRESSION instead of just NOMIPS16 and update the comment to say that it is avoiding SAVE, RESTORE and JRADDIUSP. Thanks, Matthew
> -----Original Message----- > From: Matthew Fortune [mailto:Matthew.Fortune@imgtec.com] > Sent: Monday, July 06, 2015 11:00 AM > To: Andrew Bennett; gcc-patches@gcc.gnu.org > Cc: Moore, Catherine > Subject: RE: [PATCH] MIPS: Update stack-1.c testcase to match micromips > jraddiusp instruction. > > Andrew Bennett <Andrew.Bennett@imgtec.com> writes: > > The stack-1.c testcase fails when being compiled for micromips with > > the > > -O0 optimization level. The reason is the testcase is expecting the > > following sequence at the end of the function: > > > > addiu $sp,$sp,16 > > jrc $31 > > > > But for micromips it generates the following: > > > > jraddiusp 16 > > > > > > As the failure only happens at one optimization level I have decided > > to just change the expected output rather than creating a separate > > micromips testcase. > > I'm not sure this is the right approach here. If we get a jraddiusp then the > problem that the test is trying to cover can't possibly happen anyway. > (The test is checking if a load and final stack adjustment are ever re-ordered > from what I can see.) > > I'd just mark the test as NOCOMPRESSION instead of just NOMIPS16 and > update the comment to say that it is avoiding SAVE, RESTORE and > JRADDIUSP. > Another approach would be to add the micromips testcase variant and skip the test if code-quality (ie. -O0). Catherine
diff --git a/gcc/testsuite/gcc.target/mips/stack-1.c b/gcc/testsuite/gcc.target/mips/stack-1.c index a28e4bf..2249a3b 100644 --- a/gcc/testsuite/gcc.target/mips/stack-1.c +++ b/gcc/testsuite/gcc.target/mips/stack-1.c @@ -1,4 +1,4 @@ -/* { dg-final { scan-assembler "\td?addiu\t(\\\$sp,)?\\\$sp,\[1-9\]" } } */ +/* { dg-final { scan-assembler "\t((d?addiu\t(\\\$sp,)?\\\$sp,)|jraddiusp\t)\[1-9\]" } } */ /* { dg-final { scan-assembler "\tlw\t" } } */ /* { dg-final { scan-assembler-not "\td?addiu\t(\\\$sp,)?\\\$sp,\[1-9\].*\tlw\t" } } */
Hi, The stack-1.c testcase fails when being compiled for micromips with the -O0 optimization level. The reason is the testcase is expecting the following sequence at the end of the function: addiu $sp,$sp,16 jrc $31 But for micromips it generates the following: jraddiusp 16 As the failure only happens at one optimization level I have decided to just change the expected output rather than creating a separate micromips testcase. I have tested this on the mips-mti-elf target using mips32r2/{-mno-micromips/-mmicromips} test options and there are no new regressions. The patch and ChangeLog are below. Ok to commit? Many thanks, Andrew testsuite/ * gcc.target/mips/stack-1.c: Allow testcase to match jraddiusp instruction.