On 9/21/26 12:51 PM, Dominic P wrote:
The shift count was 31 under an int32 requirement, which left the rule
untested on every other int width.
Tested on x86_64-pc-linux-gnu, aarch64-linux-gnu and arm-none-eabi
(cortex-m0 soft).
Assisted-by: Claude Opus 5 (Anthropic)
gcc/testsuite/ChangeLog:
* gcc.dg/tree-ssa/minmax-swar-1.c: Derive the shift count from
sizeof (int); drop the int32 requirement.
* gcc.dg/tree-ssa/minmax-swar-2.c: Likewise.
Signed-off-by: Dominic P <[email protected]>
OK Jeff
