Hi! On Sat, Sep 13, 2025 at 12:07:33AM -0300, Alexandre Oliva wrote: > The use of the TLS register in a TOC/GOT address computation was > probably a cut&pasto or a thinko. It causes a linker warning and, > because the TLS access in the test is incomplete, may cause > significant confusion. Adjust to use the TOC/GOT register as base.
Yeah, it does look wrong. > Tested manually with cross binutils for ppc64-linux-gnu and > ppc64-vxworks7r2. "Tested manually", what does that mean? You compiled <something> <some way> with and without the patch, and looked at the eventual machine code <some way>, and <something changed>. Please fill in the blanks. Or "<something> didn't work but now it does". Whatever will give us the warm fuzzies and make us feel confident your patch is an improvement over the status quo. > Unless there's any objection in the next few days, I > plan to put this in along with other changes. You cannot do that, ever. The patch itself looks fine, but the commit message doesn't, as said, and the changelog not either: > for gcc/ChangeLog > > * configure.ac: Adjust base register in linker test for large > toc support. > * configure: Rebuilt. You should never use passive tense in changelogs. Oh, and changelog lines are 80 character positions long, please don't wrap early. And "TOC" is an initialism, spelled in all capitals. Segher
