[Bug tree-optimization/98928] New: [11 regression] ICE when build 638.imagick_s

[Bug tree-optimization/98928] New: [11 regression] ICE when build 638.imagick_s

[crazylht at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98928

            Bug ID: 98928
           Summary: [11 regression] ICE when build 638.imagick_s
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: crazylht at gmail dot com
  Target Milestone: ---
              Host: x86_64-pc-linux-gnu

I can't reproduce it with small testcase, the regression may happen between
20201211 and 20201215, ICE still can be reproduced in the latest trunk with gcc
option: -march=skylake-avx512 -Ofast -funroll-loops -mfpmath=sse -Ofast
-fno-associative-math -flto -fopenmp

ice dump:

spec@shgcc100:~/workspace/CPU2017_SHGCC100/result/benchspec/CPU/638.imagick_s/build/build_base_march_native_ofast.0000$
/export/liuhongt/install/gcc11_trunk/bin/gcc -m64   -Wl,-z,muldefs  
-march=native -Ofast -funroll-loops -mfpmath=sse -Ofast -fno-associative-math
-funroll-loops -flto -fopenmp -DSPEC_OPENMP                       coders/tga.o
filters/analyze.o magick/accelerate.o magick/animate.o magick/annotate.o
magick/artifact.o magick/attribute.o magick/blob.o magick/cache-view.o
magick/cache.o magick/channel.o magick/cipher.o magick/client.o magick/coder.o
magick/color.o magick/colormap.o magick/colorspace.o magick/compare.o
magick/composite.o magick/compress.o magick/configure.o magick/constitute.o
magick/decorate.o magick/delegate.o magick/display.o magick/distort.o
magick/distribute-cache.o magick/draw.o magick/effect.o magick/enhance.o
magick/exception.o magick/feature.o magick/fourier.o magick/fx.o magick/gem.o
magick/geometry.o magick/hashmap.o magick/histogram.o magick/identify.o
magick/image.o magick/layer.o magick/list.o magick/locale.o magick/log.o
magick/magic.o magick/magick.o magick/matrix.o magick/memory.o magick/mime.o
magick/module.o magick/monitor.o magick/montage.o magick/morphology.o
magick/option.o magick/paint.o magick/pixel.o magick/policy.o magick/prepress.o
magick/profile.o magick/property.o magick/quantize.o magick/quantum-export.o
magick/quantum-import.o magick/quantum.o magick/random.o magick/registry.o
magick/resample.o magick/resize.o magick/resource.o magick/segment.o
magick/semaphore.o magick/shear.o magick/signature.o magick/splay-tree.o
magick/static.o magick/statistic.o magick/stream.o magick/string.o
magick/threshold.o magick/timer.o magick/token.o magick/transform.o
magick/type.o magick/utility.o magick/version.o magick/xml-tree.o
utilities/convert.o wand/convert.o wand/drawing-wand.o wand/magick-image.o
wand/magick-wand.o wand/mogrify.o wand/pixel-wand.o wand/magick-property.o
wand/pixel-iterator.o wand/wand.o magick/deprecate.o             -lm    
-L/home/spec/workspace/CPU2017_SHGCC100/speccpu2017/jemalloc/intel64/lib/
-ljemalloc -lpthread -ldl         -o imagick_sIn function ‘CopyMagickMemory’,
    inlined from ‘ConvertUTF16ToUTF8’ at magick/xml-tree.c:1341:14,
    inlined from ‘NewXMLTree.part.0’ at magick/xml-tree.c:1973:8:
magick/memory.c:702:10: warning: ‘__builtin_memcpy’ writing one too many bytes
into a region of a size that depends on ‘strlen’ [-Wstringop-overflow=]
  702 |   return(memmove(destination,source,size));
      |          ^
magick/xml-tree.c: In function ‘NewXMLTree.part.0’:
magick/memory.c:464:10: note: destination object of size [0,
9223372036854775805] allocated by ‘malloc’
  464 |   memory=memory_methods.acquire_memory_handler(size == 0 ? 1UL : size);
      |          ^
during GIMPLE pass: vect
magick/pixel.c: In function ‘InterpolateMagickPixelPacket.constprop.isra’:
magick/pixel.c:4244:32: internal compiler error: in
vect_get_vec_defs_for_operand, at tree-vect-stmts.c:1450
 4244 | MagickExport MagickBooleanType InterpolateMagickPixelPacket(const Image
*image,
      |                                ^
0x75bc56 vect_get_vec_defs_for_operand(vec_info*, _stmt_vec_info*, unsigned
int, tree_node*, vec<tree_node*, va_heap, vl_ptr>*, tree_node*)
        ../../../gcc/git_trunk/gcc/tree-vect-stmts.c:1450
0x10295e3 vect_get_vec_defs(vec_info*, _stmt_vec_info*, _slp_tree*, unsigned
int, tree_node*, vec<tree_node*, va_heap, vl_ptr>*, tree_node*, tree_node*,
vec<tree_node*, va_heap, vl_ptr>*, tree_node*, tree_node*, vec<tree_node*,
va_heap, vl_ptr>*, tree_node*, tree_node*, vec<tree_node*, va_heap, vl_ptr>*,
tree_node*)
        ../../../gcc/git_trunk/gcc/tree-vect-stmts.c:1482
0x10296cc vect_get_vec_defs(vec_info*, _stmt_vec_info*, _slp_tree*, unsigned
int, tree_node*, vec<tree_node*, va_heap, vl_ptr>*, tree_node*, vec<tree_node*,
va_heap, vl_ptr>*, tree_node*, vec<tree_node*, va_heap, vl_ptr>*, tree_node*,
vec<tree_node*, va_heap, vl_ptr>*)
        ../../../gcc/git_trunk/gcc/tree-vect-stmts.c:1504
0x102a1a5 vectorizable_conversion
        ../../../gcc/git_trunk/gcc/tree-vect-stmts.c:4963
0x1045e38 vect_transform_stmt(vec_info*, _stmt_vec_info*,
gimple_stmt_iterator*, _slp_tree*, _slp_instance*)
        ../../../gcc/git_trunk/gcc/tree-vect-stmts.c:10831
0x104b7ba vect_transform_loop_stmt
        ../../../gcc/git_trunk/gcc/tree-vect-loop.c:9204
0x10659c7 vect_transform_loop(_loop_vec_info*, gimple*)
        ../../../gcc/git_trunk/gcc/tree-vect-loop.c:9640
0x109393c try_vectorize_loop_1
        ../../../gcc/git_trunk/gcc/tree-vectorizer.c:1103
0x1094629 vectorize_loops()
        ../../../gcc/git_trunk/gcc/tree-vectorizer.c:1242
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.
lto-wrapper: fatal error: /export/liuhongt/install/gcc11_trunk/bin/gcc returned
1 exit status
compilation terminated.
/usr/bin/ld: error: lto-wrapper failed

[Bug tree-optimization/98928] [11 regression] ICE when build 638.imagick_s

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98928

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
     Ever confirmed|0                           |1
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org
   Target Milestone|---                         |11.0
   Last reconfirmed|                            |2021-02-02

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
I will investigate

[Bug tree-optimization/98928] [11 regression] ICE when build 638.imagick_s

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98928

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |marxin at gcc dot gnu.org

--- Comment #2 from Martin Liška <marxin at gcc dot gnu.org> ---
I'm reducing a test-case.

[Bug tree-optimization/98928] [11 regression] ICE when build 638.imagick_s

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98928

--- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
Reduced test-case:

$ cat ice.i
typedef float MagickRealType;
typedef short Quantum;
float InterpolateMagickPixelPacket_alpha[1];
int InterpolateMagickPixelPacket_i;

void InterpolateMagickPixelPacket();

void main() { InterpolateMagickPixelPacket(); }

typedef struct {
  MagickRealType red, green, blue, opacity, index;
} MagickPixelPacket;
typedef struct {
  Quantum blue, green, red, opacity;
} PixelPacket;
struct _Image {
  int colorspace;
  int matte;
} GetMagickPixelPacket(MagickPixelPacket *pixel) {
  pixel->red = pixel->green = pixel->blue = 0.0;
}
int AlphaBlendMagickPixelPacket(struct _Image *image, PixelPacket *color,
                            Quantum *indexes, MagickPixelPacket *pixel,
                            MagickRealType *alpha) {
  if (image->matte) {
    *alpha = pixel->red = pixel->green = pixel->blue = pixel->opacity =
        color->opacity;
    pixel->index = 0.0;
    if (image->colorspace)
      pixel->index = *indexes;
    return 0;
  }
  *alpha = 1.0 / 0.2;
  pixel->red = *alpha * color->red;
  pixel->green = *alpha * color->green;
  pixel->blue = *alpha * color->blue;
  pixel->opacity = pixel->index = 0.0;
  if (image->colorspace && indexes)
    pixel->index = *indexes;
}
MagickPixelPacket InterpolateMagickPixelPacket_pixels[1];
PixelPacket InterpolateMagickPixelPacket_p;

void
InterpolateMagickPixelPacket(struct _Image *image) {
  Quantum *indexes;
  for (; InterpolateMagickPixelPacket_i; InterpolateMagickPixelPacket_i++) {
    GetMagickPixelPacket(InterpolateMagickPixelPacket_pixels +
                         InterpolateMagickPixelPacket_i);
    AlphaBlendMagickPixelPacket(
        image, &InterpolateMagickPixelPacket_p +
InterpolateMagickPixelPacket_i,
        indexes + InterpolateMagickPixelPacket_i,
        InterpolateMagickPixelPacket_pixels + InterpolateMagickPixelPacket_i,
        InterpolateMagickPixelPacket_alpha + InterpolateMagickPixelPacket_i);
  }
}

one needs the following options: gcc -Ofast -march=skylake-avx512 ice.i
-fwhole-program

[Bug tree-optimization/98928] [11 regression] ICE when build 638.imagick_s since r11-5969-g3ed472af6bc9f83b

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98928

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tnfchris at gcc dot gnu.org
             Blocks|                            |26163
      Known to work|                            |10.2.0
      Known to fail|                            |11.0
           Priority|P3                          |P1
            Summary|[11 regression] ICE when    |[11 regression] ICE when
                   |build 638.imagick_s         |build 638.imagick_s since
                   |                            |r11-5969-g3ed472af6bc9f83b

--- Comment #4 from Martin Liška <marxin at gcc dot gnu.org> ---
Started with r11-5969-g3ed472af6bc9f83b.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26163
[Bug 26163] [meta-bug] missed optimization in SPEC (2k17, 2k and 2k6 and 95)

[Bug tree-optimization/98928] [11 regression] ICE when build 638.imagick_s since r11-5969-g3ed472af6bc9f83b

[tnfchris at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98928

--- Comment #5 from Tamar Christina <tnfchris at gcc dot gnu.org> ---
Hmm looks like the SLP unwinding code is accidentally cancelling a pattern it
shouldn't have. Checking why...

[Bug tree-optimization/98928] [11 regression] ICE when build 638.imagick_s since r11-5969-g3ed472af6bc9f83b

[tnfchris at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98928

Tamar Christina <tnfchris at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|rguenth at gcc dot gnu.org         |tnfchris at gcc dot gnu.org

--- Comment #6 from Tamar Christina <tnfchris at gcc dot gnu.org> ---
testing patch.

[Bug tree-optimization/98928] [11 regression] ICE when build 638.imagick_s since r11-5969-g3ed472af6bc9f83b

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98928

--- Comment #7 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Tamar Christina <tnfchris@gcc.gnu.org>:

https://gcc.gnu.org/g:5e606ed90a1bed878071b6b5a3ef9b97b3d99838

commit r11-7049-g5e606ed90a1bed878071b6b5a3ef9b97b3d99838
Author: Tamar Christina <tamar.christina@arm.com>
Date:   Wed Feb 3 08:06:11 2021 +0000

    slp: Split out patterns away from using SLP_ONLY into their own flag

    Previously the SLP pattern matcher was using STMT_VINFO_SLP_VECT_ONLY as a
way
    to dissolve the SLP only patterns during SLP cancellation.  However it
seems
    like the semantics for STMT_VINFO_SLP_VECT_ONLY are slightly different than
what
    I expected.

    Namely that the non-SLP path can still use a statement marked
    STMT_VINFO_SLP_VECT_ONLY.  One such example is masked loads which are used
both
    in the SLP and non-SLP path.

    To fix this I now introduce a new flag STMT_VINFO_SLP_VECT_ONLY_PATTERN
which is
    used only by the pattern matcher.

    gcc/ChangeLog:

            PR tree-optimization/98928
            * tree-vect-loop.c (vect_analyze_loop_2): Change
            STMT_VINFO_SLP_VECT_ONLY to STMT_VINFO_SLP_VECT_ONLY_PATTERN.
            * tree-vect-slp-patterns.c (complex_pattern::build): Likewise.
            * tree-vectorizer.h (STMT_VINFO_SLP_VECT_ONLY_PATTERN): New.
            (class _stmt_vec_info): Add slp_vect_pattern_only_p.

    gcc/testsuite/ChangeLog:

            PR tree-optimization/98928
            * gcc.target/i386/pr98928.c: New test.

[Bug tree-optimization/98928] [11 regression] ICE when build 638.imagick_s since r11-5969-g3ed472af6bc9f83b

[tnfchris at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98928

Tamar Christina <tnfchris at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|ASSIGNED                    |RESOLVED

--- Comment #8 from Tamar Christina <tnfchris at gcc dot gnu.org> ---
Fixed in trunk