[Bug target/86139] [7/8/9 Regression] ICE in in store_constructor, at expr.c:6849 on aarch64-linux-gnu and arm-linux-gnueabihf

[ktkachov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86139

ktkachov at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ktkachov at gcc dot gnu.org,
                   |                            |rguenth at gcc dot gnu.org
                 CC|kyrylo.tkachov at arm dot com      |

--- Comment #1 from ktkachov at gcc dot gnu.org ---
I've reproduced an ICE on arm-none-linux-gnueabihf from GCC 6 onwards but
haven't been able to reproduce it on aarch64-none-linux-gnu.

Furthermore, the ICE on arm appears to me in a different place:
during GIMPLE pass: veclower
full.c: In function ‘x’:
full.c:88:6: internal compiler error: Segmentation fault
 void x() {
      ^
0xbd7faf crash_signal
        $SRC/gcc/gcc/toplev.c:324
0xe4067f contains_struct_check
        $SRC/gcc/gcc/tree.h:3230
0xe4067f expand_vector_parallel
        $SRC/gcc/gcc/tree-vect-generic.c:328
0xe45928 expand_vector_operation
        $SRC/gcc/gcc/tree-vect-generic.c:1007
0xe45928 expand_vector_operations_1
        $SRC/gcc/gcc/tree-vect-generic.c:1801
0xe45928 expand_vector_operations
        $SRC/gcc/gcc/tree-vect-generic.c:1832
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.

when compiled with (the equivalent to the given configure line)  -std=gnu11 -O0
-Wall -Wextra -march=armv7-a -mfpu=vfpv3-d16 -mfloat-abi=hard -mthumb

Note that -mfpu=vfpv3-d16 means there's no NEON available.
With -mfpu=neon the ICE doesn't appear.

Looks like some kind of vec lowering bug when SIMD is not available on a
target?