diff options
| author | Andreas Krebbel <krebbel@linux.ibm.com> | 2024-06-10 09:09:10 +0200 |
|---|---|---|
| committer | Andreas Krebbel <krebbel@linux.ibm.com> | 2024-06-10 09:55:49 +0200 |
| commit | 21fd8c67ad297212e3cb885883cc8df8611f3040 (patch) | |
| tree | 1b0d09286afc780de734d23cc6d74c3f52d13a3f /gcc/ada | |
| parent | 48d6d8c9e91018a625a797d50ac4def88376a515 (diff) | |
| download | gcc-21fd8c67ad297212e3cb885883cc8df8611f3040.zip gcc-21fd8c67ad297212e3cb885883cc8df8611f3040.tar.gz gcc-21fd8c67ad297212e3cb885883cc8df8611f3040.tar.bz2 | |
IBM Z: Fix ICE in expand_perm_as_replicate
The current implementation assumes to always be invoked with register
operands. For memory operands we even have an instruction
though (vlrep). With the patch we try this first and only if it fails
force the input into a register and continue.
vec_splats generation fails for single element 128bit types which are
allowed for vec_splat. This is something to sort out with another
patch I guess.
gcc/ChangeLog:
* config/s390/s390.cc (expand_perm_as_replicate): Handle memory
operands.
* config/s390/vx-builtins.md (vec_splats<mode>): Turn into parameterized expander.
(@vec_splats<mode>): New expander.
gcc/testsuite/ChangeLog:
* g++.dg/torture/vshuf-mem.C: New test.
Diffstat (limited to 'gcc/ada')
0 files changed, 0 insertions, 0 deletions