https://sourceware.org/bugzilla/show_bug.cgi?id=23244
--- Comment #5 from Jim Wilson <wilson at tuliptree dot org> --- On Thu, 2018-05-31 at 09:22 +0000, palmer at gcc dot gnu.org wrote: > A similar technique might be possible for R_RISCV_CALL, we could > convert it to > an internal R_RISCV_CALL_ABS relocation and then generate a lui-based > sequence. For jalr we can just use x0 for zero-based addressing. I have an untested prototype patch that modifies the jalr instruction to use x0 and sets the relocation value to the auipc address, so the offset comes out as zero. I don't touch the auipc. This is just 13 lines of code. An R_RISCV_CALL_ABS relocation might be cleaner, but may end up as a larger patch, and we don't need the generality as we only need support for a single value, 0. Zero-page addressing is already handled by relaxation, it is only weak undef 0 that we need to handle here. -- You are receiving this mail because: You are on the CC list for the bug. _______________________________________________ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
