https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95627
--- Comment #2 from rguenther at suse dot de <rguenther at suse dot de> --- On June 10, 2020 9:54:39 PM GMT+02:00, "marxin at gcc dot gnu.org" <gcc-bugzi...@gcc.gnu.org> wrote: >https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95627 > > Bug ID: 95627 > Summary: [11 Regression] ICE in rs6000_density_test at > rs6000.c:4992 since > r11-1181-g371cc683371bedb0e53ebcee0c0e89604a1e74b1 > 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: marxin at gcc dot gnu.org > CC: rguenth at gcc dot gnu.org > Target Milestone: --- > Host: x86_64-linux-gnu > Target: ppc64le-linux-gnu > >Since the revision I see: > >$ ./xgcc -B. >/home/mliska/Programming/gcc/gcc/testsuite/gcc.c-torture/compile/20070520-1.c >-fvar-tracking-assignments-toggle -ftree-loop-vectorize -O2 >during GIMPLE pass: vect >/home/mliska/Programming/gcc/gcc/testsuite/gcc.c-torture/compile/20070520-1.c: >In function ‘ff_pred8x8_plane_c’: >/home/mliska/Programming/gcc/gcc/testsuite/gcc.c-torture/compile/20070520-1.c:4:6: >internal compiler error: Segmentation fault > 4 | void ff_pred8x8_plane_c(uint8_t *src, int stride){ > | ^~~~~~~~~~~~~~~~~~ >0xe7e38f crash_signal > ../../gcc/toplev.c:328 >0x1259ce4 rs6000_density_test > ../../gcc/config/rs6000/rs6000.c:4992 >0x125a00c rs6000_finish_cost > ../../gcc/config/rs6000/rs6000.c:5102 >0x1150170 finish_cost > ../../gcc/tree-vectorizer.h:1400 >0x1153821 vect_compute_single_scalar_iteration_cost > ../../gcc/tree-vect-loop.c:1134 >0x11568fb vect_analyze_loop_2 > ../../gcc/tree-vect-loop.c:2041 >0x11586ef vect_analyze_loop(loop*, vec_info_shared*) > ../../gcc/tree-vect-loop.c:2583 >0x119131c try_vectorize_loop_1 > ../../gcc/tree-vectorizer.c:894 >0x1191d69 vectorize_loops() > ../../gcc/tree-vectorizer.c:1128 >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. Just skip debug stmts in the IL walk in the backend or deal with NULL stmt info result for lookup of those. I have public holidays today.
