On 10/2/19 5:47 AM, Richard Sandiford wrote:
> Jeff pointed out that gcc.target/mips/call-clobbered-4.c started
> failing after the function-abi series. This is because IRA used
> to treat partly call-clobbered registers as hard conflicts and
> so wouldn't consider them for -fcaller-saves. Now th
Jeff pointed out that gcc.target/mips/call-clobbered-4.c started
failing after the function-abi series. This is because IRA used
to treat partly call-clobbered registers as hard conflicts and
so wouldn't consider them for -fcaller-saves. Now that we treat
call clobbers the same way regardless of
