On Sat, Oct 6, 2018 at 2:22 PM H.J. Lu <hjl.to...@gmail.com> wrote: > > On Sat, Oct 6, 2018 at 12:48 AM Uros Bizjak <ubiz...@gmail.com> wrote: > > > > On Fri, Oct 5, 2018 at 1:43 PM H.J. Lu <hjl.to...@gmail.com> wrote: > > > > > > On Sat, Sep 29, 2018 at 11:02 AM Uros Bizjak <ubiz...@gmail.com> wrote: > > > > > > > > On Sat, Sep 29, 2018 at 6:36 PM H.J. Lu <hjl.to...@gmail.com> wrote: > > > > > > > > > > When passing and returning BLKmode values in 2 integer registers, use > > > > > 1 TImode register instead of 2 DImode registers. Otherwise, V1TImode > > > > > may be used to move and store such BLKmode values, which prevent RTL > > > > > optimizations. > > > > > > > > > > Tested on x86-64. OK for trunk? > > > > > > > > > > Thanks. > > > > > > > > > > H.J. > > > > > --- > > > > > gcc/ > > > > > > > > > > PR target/87370 > > > > > * config/i386/i386.c (construct_container): Use TImode for > > > > > BLKmode values in 2 integer registers. > > > > > > > > > > gcc/testsuite/ > > > > > > > > > > PR target/87370 > > > > > * gcc.target/i386/pr87370.c: New test. > > > > > > > > OK. > > > > > > I'd like to backport it to release branches. Is that OK? > > > > This is a minor optimization bug, not a wrong-code bug, so unless the > > patch is super-safe, I think we should leave release branches > > unpatched. > > > > This is a regression against GCC 6. I think we should fix it for > GCC 7 and 8.
OK then, but please wait a couple of days for possible regressions on trunk. Thanks, Uros. > -- > H.J.
