Details
Details
Diff Detail
Diff Detail
- Repository
- rL LLVM
Event Timeline
lib/Target/AArch64/AArch64Subtarget.cpp | ||
---|---|---|
176 ↗ | (On Diff #107022) | The TT.isOSBinFormatCOFF() part feels a little superfluous here |
test/CodeGen/AArch64/arm64-platform-reg.ll | ||
1 ↗ | (On Diff #107022) | Unrelated to this particular patch, but shouldn't we drop the -mattr=+reserve-x18 here, to also test that darwin actuall does this right by default? (I'm not sure if there's any other tests that explicitly test that?) |
28 ↗ | (On Diff #107022) | Why the custom test prefix for windows/coff? Doesn't the existing CHECK-RESERVE-X18 also work for aarch64-windows? |
test/CodeGen/AArch64/arm64-platform-reg.ll | ||
---|---|---|
28 ↗ | (On Diff #107022) | fp is an alias for x29; I think that part of the test just makes sure that nothing tries to use that register for the copying that the test does (especially when x18 isn't available), since that register shouldn't be used for unrelated things. |