https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127142
--- Comment #4 from GCC Commits <cvs-commit at gcc dot gnu.org> --- The master branch has been updated by John David Anglin <[email protected]>: https://gcc.gnu.org/g:bc002b9e8439a708cda95e16a918760266fd053b commit r17-5470-gbc002b9e8439a708cda95e16a918760266fd053b Author: John David Anglin <[email protected]> Date: Wed Oct 21 16:17:22 2026 -0400 hppa: Partially revert commit b55ba08f1055d5c8748699a36f419c8078044594 Jeff was correct. GNU ld can't fixup a ldil/addl sequence when we have an unresolved weak symbol. It can only replace the %dp register with %r0 in the addil instruction. This was noticed in the failure of gcc.dg/torture/pr116835.c. Hopefully, the other LRA related improvements that have been recently implemented will fix the %r1 spill failures reported in PR rtl-optimization/127142. If we still get these spill failures, the linker will need fixing to handle ldil/addl sequences. 2026-10-21 John David Anglin <[email protected]> gcc/ChangeLog: PR rtl-optimization/127142 * config/pa/pa.md: Remove ldil/addl alternative from insns to load the high part of a symbolic operand.
