[PATCH] Fix PR88440, enable mem* detection at -O[2s]

Richard Biener rguenther@suse.de
Mon May 27 07:38:00 GMT 2019


On Mon, 27 May 2019, Christophe Lyon wrote:

> On Thu, 23 May 2019 at 13:32, Richard Biener <rguenther@suse.de> wrote:
> >
> > On Wed, 22 May 2019, Richard Biener wrote:
> >
> > >
> > > This enables -ftree-loop-distribute-patterns at -O[2s] and also
> > > arranges cold loops to be still processed but for pattern
> > > recognition to save code-size.
> > >
> > > Bootstrap and regtest running on x86_64-unknown-linux-gnu.
> > >
> > > Martin has done extensive compile-time testing on SPEC
> > > identifying only a single regression I'll have a look into.
> >
> > The reason for the compile-time regression is the complexity
> > heuristic in LRA no longer choosing "simple" algorithms and
> > the LIVE problem in particular being awfully slow.
> >
> > Unsurprisingly testing has also revealed loads of testsuite
> > fallout which I deal with in the patch as committed below.
> > Sorry for any further fallout on other targets (which I do
> > expect).
> >
> 
> Hi Richard,http://people.linaro.org/~christophe.lyon/cross-validation/gcc/trunk/271588/report-build-info.html
> 
> Indeed git bisect pointed me to this commit when checking
> the regressions on arm & aarch64 reported at:
> http://people.linaro.org/~christophe.lyon/cross-validation/gcc/trunk/271588/report-build-info.html
> 
> Since I'm a bit later in reporting, I'm not sure you've fixed them already?
> (I didn't notice follow-ups)
> Looking at this patch, it seems adding -fno-tree-loop-distribute-patterns to
> dg-options is the standard way of fixing the regressions?

Yes.  As I wrote above I did expect some target specific fallout and
hoped target maintainers would fix that.

Richard.

> Christophe
> 
> > Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
> >
> > Richard.
> >
> > 2019-05-23  Richard Biener  <rguenther@suse.de>
> >
> >         PR tree-optimization/88440
> >         * opts.c (default_options_table): Enable -ftree-loop-distribute-patterns
> >         at -O[2s]+.
> >         * tree-loop-distribution.c (generate_memset_builtin): Fold the
> >         generated call.
> >         (generate_memcpy_builtin): Likewise.
> >         (distribute_loop): Pass in whether to only distribute patterns.
> >         (prepare_perfect_loop_nest): Also allow size optimization.
> >         (pass_loop_distribution::execute): When optimizing a loop
> >         nest for size allow pattern replacement.
> >
> >         * gcc.dg/tree-ssa/ldist-37.c: New testcase.
> >         * gcc.dg/tree-ssa/ldist-38.c: Likewise.
> >         * gcc.dg/vect/vect.exp: Add -fno-tree-loop-distribute-patterns.
> >         * gcc.dg/tree-ssa/ldist-37.c: Adjust.
> >         * gcc.dg/tree-ssa/ldist-38.c: Likewise.
> >         * g++.dg/tree-ssa/pr78847.C: Likewise.
> >         * gcc.dg/autopar/pr39500-1.c: Likewise.
> >         * gcc.dg/autopar/reduc-1char.c: Likewise.
> >         * gcc.dg/autopar/reduc-7.c: Likewise.
> >         * gcc.dg/tree-ssa/ivopts-lt-2.c: Likewise.
> >         * gcc.dg/tree-ssa/ivopts-lt.c: Likewise.
> >         * gcc.dg/tree-ssa/predcom-dse-1.c: Likewise.
> >         * gcc.dg/tree-ssa/predcom-dse-2.c: Likewise.
> >         * gcc.dg/tree-ssa/predcom-dse-3.c: Likewise.
> >         * gcc.dg/tree-ssa/predcom-dse-4.c: Likewise.
> >         * gcc.dg/tree-ssa/prefetch-7.c: Likewise.
> >         * gcc.dg/tree-ssa/prefetch-8.c: Likewise.
> >         * gcc.dg/tree-ssa/prefetch-9.c: Likewise.
> >         * gcc.dg/tree-ssa/scev-11.c: Likewise.
> >         * gcc.dg/vect/costmodel/i386/costmodel-vect-31.c: Likewise.
> >         * gcc.dg/vect/costmodel/i386/costmodel-vect-33.c: Likewise.
> >         * gcc.dg/vect/costmodel/x86_64/costmodel-vect-31.c: Likewise.
> >         * gcc.dg/vect/costmodel/x86_64/costmodel-vect-33.c: Likewise.
> >         * gcc.target/i386/pr30970.c: Likewise.
> >         * gcc.target/i386/vect-double-1.c: Likewise.
> >         * gcc.target/i386/vect-double-2.c: Likewise.
> >         * gcc.dg/tree-ssa/gen-vect-2.c: Likewise.
> >         * gcc.dg/tree-ssa/gen-vect-26.c: Likewise.
> >         * gcc.dg/tree-ssa/gen-vect-28.c: Likewise.
> >         * gcc.dg/tree-ssa/gen-vect-32.c: Likewise.
> >         * gfortran.dg/vect/vect-5.f90: Likewise.
> >         * gfortran.dg/vect/vect-8.f90: Likewise.
> >
> > Index: gcc/opts.c
> > ===================================================================
> > --- gcc/opts.c  (revision 271513)
> > +++ gcc/opts.c  (working copy)
> > @@ -550,7 +550,7 @@ static const struct default_options defa
> >      { OPT_LEVELS_3_PLUS, OPT_fpredictive_commoning, NULL, 1 },
> >      { OPT_LEVELS_3_PLUS, OPT_fsplit_loops, NULL, 1 },
> >      { OPT_LEVELS_3_PLUS, OPT_fsplit_paths, NULL, 1 },
> > -    { OPT_LEVELS_3_PLUS, OPT_ftree_loop_distribute_patterns, NULL, 1 },
> > +    { OPT_LEVELS_2_PLUS, OPT_ftree_loop_distribute_patterns, NULL, 1 },
> >      { OPT_LEVELS_3_PLUS, OPT_ftree_loop_distribution, NULL, 1 },
> >      { OPT_LEVELS_3_PLUS, OPT_ftree_loop_vectorize, NULL, 1 },
> >      { OPT_LEVELS_3_PLUS, OPT_ftree_partial_pre, NULL, 1 },
> > Index: gcc/testsuite/g++.dg/tree-ssa/pr78847.C
> > ===================================================================
> > --- gcc/testsuite/g++.dg/tree-ssa/pr78847.C     (revision 271513)
> > +++ gcc/testsuite/g++.dg/tree-ssa/pr78847.C     (working copy)
> > @@ -1,6 +1,6 @@
> >  /* { dg-do compile } */
> >  /* { dg-require-effective-target c++14 } */
> > -/* { dg-options "-O3 -fdump-tree-ldist" } */
> > +/* { dg-options "-O3 -fdump-tree-ldist-optimized" } */
> >
> >  #include <stddef.h>
> >  #include <cstring>
> > @@ -23,4 +23,4 @@ void testWithLoopValue(const Foo foo, si
> >        buf_[ptr++] = c;
> >  }
> >
> > -/* { dg-final { scan-tree-dump "memcpy\[^\n\r\]*, 9\\);" "ldist" } } */
> > +/* { dg-final { scan-tree-dump "split to 0 loops and 1 library calls" "ldist" } } */
> > Index: gcc/testsuite/gcc.dg/autopar/pr39500-1.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/autopar/pr39500-1.c    (revision 271513)
> > +++ gcc/testsuite/gcc.dg/autopar/pr39500-1.c    (working copy)
> > @@ -1,7 +1,7 @@
> >  /* pr39500: autopar fails to parallel */
> >  /* origin: nemokingdom@gmail.com(LiFeng) */
> >  /* { dg-do compile } */
> > -/* { dg-options "-O2 -ftree-parallelize-loops=4 -fdump-tree-parloops2-details" } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -ftree-parallelize-loops=4 -fdump-tree-parloops2-details" } */
> >
> >  void abort (void);
> >
> > Index: gcc/testsuite/gcc.dg/autopar/reduc-1char.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/autopar/reduc-1char.c  (revision 271513)
> > +++ gcc/testsuite/gcc.dg/autopar/reduc-1char.c  (working copy)
> > @@ -61,5 +61,5 @@ int main (void)
> >
> >
> >  /* { dg-final { scan-tree-dump-times "Detected reduction" 3 "parloops2" } } */
> > -/* { dg-final { scan-tree-dump-times "SUCCESS: may be parallelized" 4 "parloops2" } } */
> > +/* { dg-final { scan-tree-dump-times "SUCCESS: may be parallelized" 3 "parloops2" } } */
> >
> > Index: gcc/testsuite/gcc.dg/autopar/reduc-7.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/autopar/reduc-7.c      (revision 271513)
> > +++ gcc/testsuite/gcc.dg/autopar/reduc-7.c      (working copy)
> > @@ -85,5 +85,5 @@ int main (void)
> >
> >
> >  /* { dg-final { scan-tree-dump-times "Detected reduction" 2 "parloops2" } } */
> > -/* { dg-final { scan-tree-dump-times "SUCCESS: may be parallelized" 3 "parloops2" } } */
> > +/* { dg-final { scan-tree-dump-times "SUCCESS: may be parallelized" 2 "parloops2" } } */
> >
> > Index: gcc/testsuite/gcc.dg/tree-ssa/gen-vect-2.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/gen-vect-2.c  (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/gen-vect-2.c  (working copy)
> > @@ -1,6 +1,6 @@
> >  /* { dg-do run { target vect_cmdline_needed } } */
> > -/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details -fvect-cost-model=dynamic" } */
> > -/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details -fvect-cost-model=dynamic -mno-sse" { target { i?86-*-* x86_64-*-* } } } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -ftree-vectorize -fdump-tree-vect-details -fvect-cost-model=dynamic" } */
> > +/* { dg-additional-options "-mno-sse" { target { i?86-*-* x86_64-*-* } } } */
> >
> >  #include <stdlib.h>
> >
> > Index: gcc/testsuite/gcc.dg/tree-ssa/gen-vect-26.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/gen-vect-26.c (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/gen-vect-26.c (working copy)
> > @@ -1,6 +1,6 @@
> >  /* { dg-do run { target vect_cmdline_needed } } */
> > -/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details -fvect-cost-model=dynamic" } */
> > -/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details -fvect-cost-model=dynamic -mno-sse" { target { i?86-*-* x86_64-*-* } } } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -ftree-vectorize -fdump-tree-vect-details -fvect-cost-model=dynamic" } */
> > +/* { dg-additional-options "-mno-sse" { target { i?86-*-* x86_64-*-* } } } */
> >
> >  #include <stdlib.h>
> >
> > Index: gcc/testsuite/gcc.dg/tree-ssa/gen-vect-28.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/gen-vect-28.c (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/gen-vect-28.c (working copy)
> > @@ -1,6 +1,6 @@
> >  /* { dg-do run { target vect_cmdline_needed } } */
> > -/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details -fvect-cost-model=dynamic" } */
> > -/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details -fvect-cost-model=dynamic -mno-sse" { target { i?86-*-* x86_64-*-* } } } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -ftree-vectorize -fdump-tree-vect-details -fvect-cost-model=dynamic" } */
> > +/* { dg-additional-options "-mno-sse" { target { i?86-*-* x86_64-*-* } } } */
> >
> >  #include <stdlib.h>
> >
> > Index: gcc/testsuite/gcc.dg/tree-ssa/gen-vect-32.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/gen-vect-32.c (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/gen-vect-32.c (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do run { target vect_cmdline_needed } } */
> > -/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details -fno-vect-cost-model" } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -ftree-vectorize -fdump-tree-vect-details -fno-vect-cost-model" } */
> >  /* { dg-additional-options "-mno-sse" { target { i?86-*-* x86_64-*-* } } } */
> >
> >  #include <stdlib.h>
> > Index: gcc/testsuite/gcc.dg/tree-ssa/ivopts-lt-2.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/ivopts-lt-2.c (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/ivopts-lt-2.c (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do compile } */
> > -/* { dg-options "-O2 -fdump-tree-ivopts" } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -fdump-tree-ivopts" } */
> >  /* { dg-skip-if "PR68644" { hppa*-*-* powerpc*-*-* } } */
> >
> >  void
> > Index: gcc/testsuite/gcc.dg/tree-ssa/ivopts-lt.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/ivopts-lt.c   (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/ivopts-lt.c   (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do compile } */
> > -/* { dg-options "-O2 -fdump-tree-ivopts" } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -fdump-tree-ivopts" } */
> >  /* { dg-require-effective-target stdint_types } */
> >
> >  #include "stdint.h"
> > Index: gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-1.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-1.c       (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-1.c       (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do run } */
> > -/* { dg-options "-O2 -fno-inline -fpredictive-commoning -fdump-tree-pcom-details" } */
> > +/* { dg-options "-O2 -fno-inline -fno-tree-loop-distribute-patterns -fpredictive-commoning -fdump-tree-pcom-details" } */
> >
> >  int arr[105] = {2, 3, 5, 7, 11};
> >  int result0[10] = {2, 3, 5, 7, 11};
> > Index: gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-2.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-2.c       (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-2.c       (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do run } */
> > -/* { dg-options "-O2 -fno-inline -fpredictive-commoning -fdump-tree-pcom-details" } */
> > +/* { dg-options "-O2 -fno-inline -fno-tree-loop-distribute-patterns -fpredictive-commoning -fdump-tree-pcom-details" } */
> >
> >  int arr[105] = {2, 3, 5, 7, 11};
> >  int result0[10] = {2, 3, 5, 7, 11};
> > Index: gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-3.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-3.c       (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-3.c       (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do run } */
> > -/* { dg-options "-O2 -fno-inline -fpredictive-commoning -fdump-tree-pcom-details" } */
> > +/* { dg-options "-O2 -fno-inline -fno-tree-loop-distribute-patterns -fpredictive-commoning -fdump-tree-pcom-details" } */
> >
> >  int arr1[105] = {2, 3, 5, 7, 11, 13, 0};
> >  int arr2[105] = {2, 3, 5, 7, 11, 13, 0};
> > Index: gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-4.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-4.c       (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/predcom-dse-4.c       (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do run } */
> > -/* { dg-options "-O2 -fno-inline -fpredictive-commoning -fdump-tree-pcom-details" } */
> > +/* { dg-options "-O2 -fno-inline -fno-tree-loop-distribute-patterns -fpredictive-commoning -fdump-tree-pcom-details" } */
> >
> >  int arr[105] = {2, 3, 5, 7, 11};
> >  int result0[10] = {2, 3, 5, 7, 11};
> > Index: gcc/testsuite/gcc.dg/tree-ssa/prefetch-7.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/prefetch-7.c  (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/prefetch-7.c  (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do compile { target { i?86-*-* x86_64-*-* } } } */
> > -/* { dg-options "-O2 -fprefetch-loop-arrays -march=amdfam10 --param simultaneous-prefetches=100 -fdump-tree-aprefetch-details -fdump-tree-optimized" } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -fprefetch-loop-arrays -march=amdfam10 --param simultaneous-prefetches=100 -fdump-tree-aprefetch-details -fdump-tree-optimized" } */
> >
> >  #define K 1000000
> >  int a[K];
> > Index: gcc/testsuite/gcc.dg/tree-ssa/prefetch-8.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/prefetch-8.c  (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/prefetch-8.c  (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do compile { target { i?86-*-* x86_64-*-* } } } */
> > -/* { dg-options "-O2 -fprefetch-loop-arrays -march=amdfam10 --param simultaneous-prefetches=100 -fdump-tree-aprefetch-details -fdump-tree-optimized" } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -fprefetch-loop-arrays -march=amdfam10 --param simultaneous-prefetches=100 -fdump-tree-aprefetch-details -fdump-tree-optimized" } */
> >
> >  #define K 1000000
> >  int a[K];
> > Index: gcc/testsuite/gcc.dg/tree-ssa/prefetch-9.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/prefetch-9.c  (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/prefetch-9.c  (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do compile { target { i?86-*-* x86_64-*-* } } } */
> > -/* { dg-options "-O2 -fprefetch-loop-arrays -march=amdfam10 --param simultaneous-prefetches=100 -fdump-tree-aprefetch-details -fdump-tree-optimized" } */
> > +/* { dg-options "-O2 -fno-tree-loop-distribute-patterns -fprefetch-loop-arrays -march=amdfam10 --param simultaneous-prefetches=100 -fdump-tree-aprefetch-details -fdump-tree-optimized" } */
> >
> >  #define K 1000000
> >  int a[K], b[K];
> > Index: gcc/testsuite/gcc.dg/tree-ssa/scev-11.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/tree-ssa/scev-11.c     (revision 271513)
> > +++ gcc/testsuite/gcc.dg/tree-ssa/scev-11.c     (working copy)
> > @@ -15,7 +15,7 @@ foo (int n)
> >      {
> >        unsigned char uc = (unsigned char)i;
> >        a[i] = i;
> > -      b[uc] = 0;
> > +      b[uc] = 1;
> >      }
> >
> >    bar (a);
> > Index: gcc/testsuite/gcc.dg/vect/costmodel/i386/costmodel-vect-31.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/vect/costmodel/i386/costmodel-vect-31.c        (revision 271513)
> > +++ gcc/testsuite/gcc.dg/vect/costmodel/i386/costmodel-vect-31.c        (working copy)
> > @@ -1,4 +1,5 @@
> >  /* { dg-require-effective-target vect_int } */
> > +/* { dg-additional-options "-fno-tree-loop-distribute-patterns" } */
> >
> >  #include <stdarg.h>
> >  #include "../../tree-vect.h"
> > Index: gcc/testsuite/gcc.dg/vect/costmodel/i386/costmodel-vect-33.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/vect/costmodel/i386/costmodel-vect-33.c        (revision 271513)
> > +++ gcc/testsuite/gcc.dg/vect/costmodel/i386/costmodel-vect-33.c        (working copy)
> > @@ -1,5 +1,6 @@
> >  /* { dg-do compile } */
> >  /* { dg-require-effective-target vect_int } */
> > +/* { dg-additional-options "-fno-tree-loop-distribute-patterns" } */
> >
> >  #include <stdarg.h>
> >  #include "../../tree-vect.h"
> > Index: gcc/testsuite/gcc.dg/vect/costmodel/x86_64/costmodel-vect-31.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/vect/costmodel/x86_64/costmodel-vect-31.c      (revision 271513)
> > +++ gcc/testsuite/gcc.dg/vect/costmodel/x86_64/costmodel-vect-31.c      (working copy)
> > @@ -1,4 +1,5 @@
> >  /* { dg-require-effective-target vect_int } */
> > +/* { dg-additional-options "-fno-tree-loop-distribute-patterns" } */
> >
> >  #include <stdarg.h>
> >  #include "../../tree-vect.h"
> > Index: gcc/testsuite/gcc.dg/vect/costmodel/x86_64/costmodel-vect-33.c
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/vect/costmodel/x86_64/costmodel-vect-33.c      (revision 271513)
> > +++ gcc/testsuite/gcc.dg/vect/costmodel/x86_64/costmodel-vect-33.c      (working copy)
> > @@ -1,5 +1,6 @@
> >  /* { dg-do compile } */
> >  /* { dg-require-effective-target vect_int } */
> > +/* { dg-additional-options "-fno-tree-loop-distribute-patterns" } */
> >
> >  #include <stdarg.h>
> >  #include "../../tree-vect.h"
> > Index: gcc/testsuite/gcc.dg/vect/vect.exp
> > ===================================================================
> > --- gcc/testsuite/gcc.dg/vect/vect.exp  (revision 271513)
> > +++ gcc/testsuite/gcc.dg/vect/vect.exp  (working copy)
> > @@ -45,7 +45,7 @@ if ![check_vect_support_and_set_flags] {
> >  }
> >
> >  # These flags are used for all targets.
> > -lappend DEFAULT_VECTCFLAGS "-ftree-vectorize" "-fno-vect-cost-model" "-fno-common"
> > +lappend DEFAULT_VECTCFLAGS "-ftree-vectorize" "-fno-tree-loop-distribute-patterns" "-fno-vect-cost-model" "-fno-common"
> >
> >  # Initialize `dg'.
> >  dg-init
> > Index: gcc/testsuite/gcc.target/i386/pr30970.c
> > ===================================================================
> > --- gcc/testsuite/gcc.target/i386/pr30970.c     (revision 271513)
> > +++ gcc/testsuite/gcc.target/i386/pr30970.c     (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do compile }
> > -/* { dg-options "-msse2 -O2 -ftree-vectorize -mtune=generic" } */
> > +/* { dg-options "-msse2 -O2 -fno-tree-loop-distribute-patterns -ftree-vectorize -mtune=generic" } */
> >
> >  #define N 256
> >  int b[N];
> > Index: gcc/testsuite/gcc.target/i386/vect-double-1.c
> > ===================================================================
> > --- gcc/testsuite/gcc.target/i386/vect-double-1.c       (revision 271513)
> > +++ gcc/testsuite/gcc.target/i386/vect-double-1.c       (working copy)
> > @@ -1,6 +1,6 @@
> >  /* { dg-do compile } */
> >  /* { dg-skip-if "" { *-*-* } { "-march=*" } { "-march=core2" } } */
> > -/* { dg-options "-O2 -ftree-vectorize -mfpmath=sse -march=core2 -fdump-tree-vect-stats" } */
> > +/* { dg-options "-O2 -ftree-vectorize -fno-tree-loop-distribute-patterns -mfpmath=sse -march=core2 -fdump-tree-vect-stats" } */
> >  /* { dg-add-options bind_pic_locally } */
> >
> >  extern void abort (void);
> > Index: gcc/testsuite/gcc.target/i386/vect-double-2.c
> > ===================================================================
> > --- gcc/testsuite/gcc.target/i386/vect-double-2.c       (revision 271513)
> > +++ gcc/testsuite/gcc.target/i386/vect-double-2.c       (working copy)
> > @@ -1,5 +1,5 @@
> >  /* { dg-do compile } */
> > -/* { dg-options "-O2 -ftree-vectorize -mfpmath=sse -msse2 -mtune=atom -fdump-tree-vect-stats" } */
> > +/* { dg-options "-O2 -ftree-vectorize -fno-tree-loop-distribute-patterns -mfpmath=sse -msse2 -mtune=atom -fdump-tree-vect-stats" } */
> >
> >  extern void abort (void);
> >
> > Index: gcc/testsuite/gfortran.dg/vect/vect-5.f90
> > ===================================================================
> > --- gcc/testsuite/gfortran.dg/vect/vect-5.f90   (revision 271513)
> > +++ gcc/testsuite/gfortran.dg/vect/vect-5.f90   (working copy)
> > @@ -1,5 +1,5 @@
> >  ! { dg-require-effective-target vect_int }
> > -! { dg-additional-options "--param vect-max-peeling-for-alignment=0" }
> > +! { dg-additional-options "-fno-tree-loop-distribute-patterns --param vect-max-peeling-for-alignment=0" }
> >
> >          Subroutine foo (N, M)
> >          Integer N
> > Index: gcc/testsuite/gfortran.dg/vect/vect-8.f90
> > ===================================================================
> > --- gcc/testsuite/gfortran.dg/vect/vect-8.f90   (revision 271513)
> > +++ gcc/testsuite/gfortran.dg/vect/vect-8.f90   (working copy)
> > @@ -1,6 +1,6 @@
> >  ! { dg-do compile }
> >  ! { dg-require-effective-target vect_double }
> > -! { dg-additional-options "-finline-matmul-limit=0" }
> > +! { dg-additional-options "-fno-tree-loop-distribute-patterns -finline-matmul-limit=0" }
> >
> >  module lfk_prec
> >   integer, parameter :: dp=kind(1.d0)
> > Index: gcc/tree-loop-distribution.c
> > ===================================================================
> > --- gcc/tree-loop-distribution.c        (revision 271513)
> > +++ gcc/tree-loop-distribution.c        (working copy)
> > @@ -115,6 +115,7 @@ along with GCC; see the file COPYING3.
> >  #include "params.h"
> >  #include "tree-vectorizer.h"
> >  #include "tree-eh.h"
> > +#include "gimple-fold.h"
> >
> >
> >  #define MAX_DATAREFS_NUM \
> > @@ -1028,6 +1029,7 @@ generate_memset_builtin (struct loop *lo
> >    fn = build_fold_addr_expr (builtin_decl_implicit (BUILT_IN_MEMSET));
> >    fn_call = gimple_build_call (fn, 3, mem, val, nb_bytes);
> >    gsi_insert_after (&gsi, fn_call, GSI_CONTINUE_LINKING);
> > +  fold_stmt (&gsi);
> >
> >    if (dump_file && (dump_flags & TDF_DETAILS))
> >      {
> > @@ -1071,6 +1073,7 @@ generate_memcpy_builtin (struct loop *lo
> >    fn = build_fold_addr_expr (builtin_decl_implicit (kind));
> >    fn_call = gimple_build_call (fn, 3, dest, src, nb_bytes);
> >    gsi_insert_after (&gsi, fn_call, GSI_CONTINUE_LINKING);
> > +  fold_stmt (&gsi);
> >
> >    if (dump_file && (dump_flags & TDF_DETAILS))
> >      {
> > @@ -2769,7 +2772,8 @@ finalize_partitions (struct loop *loop,
> >
> >  static int
> >  distribute_loop (struct loop *loop, vec<gimple *> stmts,
> > -                control_dependences *cd, int *nb_calls, bool *destroy_p)
> > +                control_dependences *cd, int *nb_calls, bool *destroy_p,
> > +                bool only_patterns_p)
> >  {
> >    ddrs_table = new hash_table<ddr_hasher> (389);
> >    struct graph *rdg;
> > @@ -2843,7 +2847,7 @@ distribute_loop (struct loop *loop, vec<
> >
> >    /* If we are only distributing patterns but did not detect any,
> >       simply bail out.  */
> > -  if (!flag_tree_loop_distribution
> > +  if (only_patterns_p
> >        && !any_builtin)
> >      {
> >        nbp = 0;
> > @@ -2855,7 +2859,7 @@ distribute_loop (struct loop *loop, vec<
> >       a loop into pieces, separated by builtin calls.  That is, we
> >       only want no or a single loop body remaining.  */
> >    struct partition *into;
> > -  if (!flag_tree_loop_distribution)
> > +  if (only_patterns_p)
> >      {
> >        for (i = 0; partitions.iterate (i, &into); ++i)
> >         if (!partition_builtin_p (into))
> > @@ -3085,7 +3089,6 @@ prepare_perfect_loop_nest (struct loop *
> >          && loop_outer (outer)
> >          && outer->inner == loop && loop->next == NULL
> >          && single_exit (outer)
> > -        && optimize_loop_for_speed_p (outer)
> >          && !chrec_contains_symbols_defined_in_loop (niters, outer->num)
> >          && (niters = number_of_latch_executions (outer)) != NULL_TREE
> >          && niters != chrec_dont_know)
> > @@ -3139,9 +3142,11 @@ pass_loop_distribution::execute (functio
> >       walking to innermost loops.  */
> >    FOR_EACH_LOOP (loop, LI_ONLY_INNERMOST)
> >      {
> > -      /* Don't distribute multiple exit edges loop, or cold loop.  */
> > +      /* Don't distribute multiple exit edges loop, or cold loop when
> > +         not doing pattern detection.  */
> >        if (!single_exit (loop)
> > -         || !optimize_loop_for_speed_p (loop))
> > +         || (!flag_tree_loop_distribute_patterns
> > +             && !optimize_loop_for_speed_p (loop)))
> >         continue;
> >
> >        /* Don't distribute loop if niters is unknown.  */
> > @@ -3169,9 +3174,10 @@ pass_loop_distribution::execute (functio
> >
> >           bool destroy_p;
> >           int nb_generated_loops, nb_generated_calls;
> > -         nb_generated_loops = distribute_loop (loop, work_list, cd,
> > -                                               &nb_generated_calls,
> > -                                               &destroy_p);
> > +         nb_generated_loops
> > +           = distribute_loop (loop, work_list, cd, &nb_generated_calls,
> > +                              &destroy_p, (!optimize_loop_for_speed_p (loop)
> > +                                           || !flag_tree_loop_distribution));
> >           if (destroy_p)
> >             loops_to_be_destroyed.safe_push (loop);
> >
> 

-- 
Richard Biener <rguenther@suse.de>
SUSE Linux GmbH, Maxfeldstrasse 5, 90409 Nuernberg, Germany;
GF: Felix Imendörffer, Mary Higgins, Sri Rasiah; HRB 21284 (AG Nürnberg)


More information about the Gcc-patches mailing list