https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124637
--- Comment #17 from Manjunath S Matti <mmatti at linux dot vnet.ibm.com> --- (In reply to Manjunath S Matti from comment #15) > (In reply to Roger Sayle from comment #11) > > Created attachment 64065 [details] > > proposed patch > > > > Here's my guess at a patch. Unfortunately, the only targets I can > > thoroughly test are little-endian, so although this patch appears to > > improve/change code generation on a cross-compiler to powerpc64-linux-gnu, > > testing beyond that is minimal. If someone could confirm that this fixes > > things on real hardware (or a simulator) that would be great, then the next > > step would be to check that it doesn't cause problems on other big-endian > > targets. > > I have a BE machine that I am will to test this patch with. I have built gcc with this patch and the attached testcase works as mentioned. $./tc-comp.sh && ./m-0 && ./m-2 20 1 1 20 1 1 I am running make check-gcc on powerpc64 (BE), still waiting for the results. Will be updating soon.
