Message ID | 5710DD1C.4030708@foss.arm.com |
---|---|
State | New |
Headers | show |
On Fri, Apr 15, 2016 at 01:22:52PM +0100, Kyrill Tkachov wrote: > As reported (https://gcc.gnu.org/ml/gcc-patches/2016-04/msg00094.html) > these two tests now fail to perform shrinkwrapping after a regalloc change, > but at least on arm the resulting code is not worse (not clear if better either). > > They have also been reported to fail on powerpc > > One of the proposed solutions for now is to XFAIL them on arm and powerpc, which > is what this patch does. > > Is this ok for GCC 6? Or would you like to remove arm and powerpc from the tested > targets for this altogether? Please add there a comment referencing the PR70681 as the reason for the xfail. Just grep around other xfails to see how is that usually written. > 2016-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com> > > PR rtl-optimization/70681 > * gcc.dg/ira-shrinkwrap-prep-2.c: XFAIL shrinkwrapping > dump scan on arm and powerpc. > * gcc.dg/pr10474.c: Likewise. Ok with that change. > commit 54df4becafdde6676d7af6e868f1ca4bc3844888 > Author: Kyrylo Tkachov <kyrylo.tkachov@arm.com> > Date: Fri Apr 15 12:23:03 2016 +0100 > > [testsuite] XFAIL ira-shrinkwrap-prep-2.c and pr10474.c tests on arm, powerpc > > diff --git a/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c b/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c > index a7f846a..851ced2 100644 > --- a/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c > +++ b/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c > @@ -31,4 +31,4 @@ bar (long a) > > /* { dg-final { scan-rtl-dump "Will split live ranges of parameters" "ira" } } */ > /* { dg-final { scan-rtl-dump "Split live-range of register" "ira" } } */ > -/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" } } */ > +/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" { xfail arm*-*-* powerpc*-*-* } } } */ > diff --git a/gcc/testsuite/gcc.dg/pr10474.c b/gcc/testsuite/gcc.dg/pr10474.c > index ee9edd4..09e865a 100644 > --- a/gcc/testsuite/gcc.dg/pr10474.c > +++ b/gcc/testsuite/gcc.dg/pr10474.c > @@ -12,4 +12,4 @@ void f(int *i) > } > } > > -/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" } } */ > +/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" { xfail arm*-*-* powerpc*-*-* } } } */ Jakub
Hi Jakub, On 15/04/16 13:26, Jakub Jelinek wrote: > On Fri, Apr 15, 2016 at 01:22:52PM +0100, Kyrill Tkachov wrote: >> As reported (https://gcc.gnu.org/ml/gcc-patches/2016-04/msg00094.html) >> these two tests now fail to perform shrinkwrapping after a regalloc change, >> but at least on arm the resulting code is not worse (not clear if better either). >> >> They have also been reported to fail on powerpc >> >> One of the proposed solutions for now is to XFAIL them on arm and powerpc, which >> is what this patch does. >> >> Is this ok for GCC 6? Or would you like to remove arm and powerpc from the tested >> targets for this altogether? > Please add there a comment referencing the PR70681 as the reason for the > xfail. Just grep around other xfails to see how is that usually written. Sorry, grepping around the testsuite I think I'm missing something. The only place where I can add a comment to an xfail is on "dg-xfail-if" standalone directives that xfail the whole test, whereas here we want to xfail just the shrinkwrapping RTL dump, and I don't see that taking a message argument. Did you mean just adding a comment to the code itself, like: /* XFAIL due to PR70681. */ /* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" { xfail arm*-*-* powerpc*-*-* } } } */ ? Kyrill >> 2016-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com> >> >> PR rtl-optimization/70681 >> * gcc.dg/ira-shrinkwrap-prep-2.c: XFAIL shrinkwrapping >> dump scan on arm and powerpc. >> * gcc.dg/pr10474.c: Likewise. > Ok with that change. > >> commit 54df4becafdde6676d7af6e868f1ca4bc3844888 >> Author: Kyrylo Tkachov <kyrylo.tkachov@arm.com> >> Date: Fri Apr 15 12:23:03 2016 +0100 >> >> [testsuite] XFAIL ira-shrinkwrap-prep-2.c and pr10474.c tests on arm, powerpc >> >> diff --git a/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c b/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c >> index a7f846a..851ced2 100644 >> --- a/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c >> +++ b/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c >> @@ -31,4 +31,4 @@ bar (long a) >> >> /* { dg-final { scan-rtl-dump "Will split live ranges of parameters" "ira" } } */ >> /* { dg-final { scan-rtl-dump "Split live-range of register" "ira" } } */ >> -/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" } } */ >> +/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" { xfail arm*-*-* powerpc*-*-* } } } */ >> diff --git a/gcc/testsuite/gcc.dg/pr10474.c b/gcc/testsuite/gcc.dg/pr10474.c >> index ee9edd4..09e865a 100644 >> --- a/gcc/testsuite/gcc.dg/pr10474.c >> +++ b/gcc/testsuite/gcc.dg/pr10474.c >> @@ -12,4 +12,4 @@ void f(int *i) >> } >> } >> >> -/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" } } */ >> +/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" { xfail arm*-*-* powerpc*-*-* } } } */ > > Jakub
On Fri, Apr 15, 2016 at 01:38:57PM +0100, Kyrill Tkachov wrote: > Hi Jakub, > > On 15/04/16 13:26, Jakub Jelinek wrote: > >On Fri, Apr 15, 2016 at 01:22:52PM +0100, Kyrill Tkachov wrote: > >>As reported (https://gcc.gnu.org/ml/gcc-patches/2016-04/msg00094.html) > >>these two tests now fail to perform shrinkwrapping after a regalloc change, > >>but at least on arm the resulting code is not worse (not clear if better either). > >> > >>They have also been reported to fail on powerpc > >> > >>One of the proposed solutions for now is to XFAIL them on arm and powerpc, which > >>is what this patch does. > >> > >>Is this ok for GCC 6? Or would you like to remove arm and powerpc from the tested > >>targets for this altogether? > >Please add there a comment referencing the PR70681 as the reason for the > >xfail. Just grep around other xfails to see how is that usually written. > > Sorry, grepping around the testsuite I think I'm missing something. > The only place where I can add a comment to an xfail is on > "dg-xfail-if" standalone directives that xfail the whole test, whereas > here we want to xfail just the shrinkwrapping RTL dump, and I don't see that > taking a message argument. Sorry, I've swapped xfail and dg-skip-if in my mind, for the latter we have comments. > Did you mean just adding a comment to the code itself, like: > > /* XFAIL due to PR70681. */ > /* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" { xfail arm*-*-* powerpc*-*-* } } } */ But yeah, this LGTM. Thanks. Jakub
On 15/04/16 13:40, Jakub Jelinek wrote: > On Fri, Apr 15, 2016 at 01:38:57PM +0100, Kyrill Tkachov wrote: >> Hi Jakub, >> >> On 15/04/16 13:26, Jakub Jelinek wrote: >>> On Fri, Apr 15, 2016 at 01:22:52PM +0100, Kyrill Tkachov wrote: >>>> As reported (https://gcc.gnu.org/ml/gcc-patches/2016-04/msg00094.html) >>>> these two tests now fail to perform shrinkwrapping after a regalloc change, >>>> but at least on arm the resulting code is not worse (not clear if better either). >>>> >>>> They have also been reported to fail on powerpc >>>> >>>> One of the proposed solutions for now is to XFAIL them on arm and powerpc, which >>>> is what this patch does. >>>> >>>> Is this ok for GCC 6? Or would you like to remove arm and powerpc from the tested >>>> targets for this altogether? >>> Please add there a comment referencing the PR70681 as the reason for the >>> xfail. Just grep around other xfails to see how is that usually written. >> Sorry, grepping around the testsuite I think I'm missing something. >> The only place where I can add a comment to an xfail is on >> "dg-xfail-if" standalone directives that xfail the whole test, whereas >> here we want to xfail just the shrinkwrapping RTL dump, and I don't see that >> taking a message argument. > Sorry, I've swapped xfail and dg-skip-if in my mind, for the latter we have > comments. > >> Did you mean just adding a comment to the code itself, like: >> >> /* XFAIL due to PR70681. */ >> /* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" { xfail arm*-*-* powerpc*-*-* } } } */ > But yeah, this LGTM. Thanks. Thanks, I've commited with comment added to the testcases. Kyrill > Jakub
commit 54df4becafdde6676d7af6e868f1ca4bc3844888 Author: Kyrylo Tkachov <kyrylo.tkachov@arm.com> Date: Fri Apr 15 12:23:03 2016 +0100 [testsuite] XFAIL ira-shrinkwrap-prep-2.c and pr10474.c tests on arm, powerpc diff --git a/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c b/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c index a7f846a..851ced2 100644 --- a/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c +++ b/gcc/testsuite/gcc.dg/ira-shrinkwrap-prep-2.c @@ -31,4 +31,4 @@ bar (long a) /* { dg-final { scan-rtl-dump "Will split live ranges of parameters" "ira" } } */ /* { dg-final { scan-rtl-dump "Split live-range of register" "ira" } } */ -/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" } } */ +/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" { xfail arm*-*-* powerpc*-*-* } } } */ diff --git a/gcc/testsuite/gcc.dg/pr10474.c b/gcc/testsuite/gcc.dg/pr10474.c index ee9edd4..09e865a 100644 --- a/gcc/testsuite/gcc.dg/pr10474.c +++ b/gcc/testsuite/gcc.dg/pr10474.c @@ -12,4 +12,4 @@ void f(int *i) } } -/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" } } */ +/* { dg-final { scan-rtl-dump "Performing shrink-wrapping" "pro_and_epilogue" { xfail arm*-*-* powerpc*-*-* } } } */