[Bug tree-optimization/117202] SLP permutation for VLA vectors broken

2025-01-28 Thread rsandifo at gcc dot gnu.org via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117202 --- Comment #5 from Richard Sandiford --- FWIW, gcc.target/riscv/rvv/autovec/struct/mask_struct_store_run-3.c seems to produce similar VEC_PERM_EXPRs for SVE, but it works there. The idea is that we're unpacking one vector of [16,16] chars into

[Bug tree-optimization/117202] SLP permutation for VLA vectors broken

2025-01-28 Thread rguenth at gcc dot gnu.org via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117202 --- Comment #4 from Richard Biener --- Created attachment 60309 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=60309&action=edit refreshed patch I picked this from a branch where it was updated.

[Bug tree-optimization/117202] SLP permutation for VLA vectors broken

2024-10-18 Thread rguenth at gcc dot gnu.org via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117202 --- Comment #3 from Richard Biener --- I think it goes wrong when computing vperm[], specifically active_lane[] but I'm a bit lost so leaving to Richard S. for now.

[Bug tree-optimization/117202] SLP permutation for VLA vectors broken

2024-10-18 Thread rguenth at gcc dot gnu.org via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117202 --- Comment #2 from Richard Biener --- Smaller testcase: void __attribute__ ((noinline, noclone)) foo (signed char *__restrict dest, signed char *__restrict src, signed char *__restrict cond, signed char bias, long

[Bug tree-optimization/117202] SLP permutation for VLA vectors broken

2024-10-18 Thread rguenth at gcc dot gnu.org via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117202 Richard Biener changed: What|Removed |Added Keywords||wrong-code CC|