Details
Diff Detail
- Repository
- rL LLVM
Event Timeline
lld/test/ELF/aarch64-relocs.s | ||
---|---|---|
242 ↗ | (On Diff #209662) | You can consider moving this chunk to a separate test and using llvm-objdump -d --print-imm-hex, then you don't have to explain -65537 = 0xfffffffffffeffff. x86-64-reloc-*.s and ppc32-reloc-*.s are organized this way. |
Functionally looks good. To my eyes it matches what is in the Arm ARM and ELF for the 64-bit architecture:
https://developer.arm.com/docs/ihi0056/latest/elf-for-the-arm-64-bit-architecture-aarch64-abi-2019q2-documentation
https://developer.arm.com/docs/ddi0487/latest/arm-architecture-reference-manual-armv8-for-armv8-a-architecture-profile
lld/ELF/Arch/AArch64.cpp | ||
---|---|---|
257 ↗ | (On Diff #209662) | Really trivial nit, we use lower case in the comment before the function but upper case in the body. Is it worth choosing one? The rest of the file seems to mostly use lower case although there are some uses of MOVZ and MOVK in relaxTlsIeToLE. |
262 ↗ | (On Diff #209662) | To save a lookup in the Arm ARM, is it worth a quick comment along the lines: |
LGTM
lld/ELF/Arch/AArch64.cpp | ||
---|---|---|
260 ↗ | (On Diff #209662) | As a single-letter variable, l is perhaps not a good name because it looks like 1. Maybe loc? |
lld/test/ELF/aarch64-relocs.s | ||
---|---|---|
242 ↗ | (On Diff #209662) | I think I'll leave this as is. The switch to --print-imm-hex would probably be better done as a cleanup over this entire file since there are already a number of instances of the same pattern here. |