Closed dpaoliello closed 2 months ago
r? @Amanieu
rustbot has assigned @Amanieu. They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.
Use r?
to explicitly pick a reviewer
mov
is an alias for ins
when moving vector elements, so it's just the disassembler choosing to disassemble things differently. The actual generated code is the same. uxtll2
is not a valid instruction, the correct name is uxtl2
, so that's just the disassembler being wrong.
It may be worth using llvm-objdump
for disassembly on all targets for consistency.
mov
is an alias forins
when moving vector elements, so it's just the disassembler choosing to disassemble things differently. The actual generated code is the same.uxtll2
is not a valid instruction, the correct name isuxtl2
, so that's just the disassembler being wrong.
Ah, thanks! This reduced the failures to 1.
It may be worth using
llvm-objdump
for disassembly on all targets for consistency.
I tried this, but it didn't work since llvm-objdump
doesn't seem to read in the PDB files, and so the disassembly didn't contain any function names. Let me see if I can disable split-debuginfo for tests on Windows and try again...
It may be worth using
llvm-objdump
for disassembly on all targets for consistency.I tried this, but it didn't work since
llvm-objdump
doesn't seem to read in the PDB files, and so the disassembly didn't contain any function names. Let me see if I can disable split-debuginfo for tests on Windows and try again...
Except that split-debuginfo = "off"
isn't supported by MSVC targets... so never mind.
Enables testing for AArch64 Windows as "no run", as there are no AArch64 GitHub runners available.
GITHUB_ENV
for Windows (use Bash instead of PowerShell).shll
instructions to handle dumpbin's format.ins
asmov
for MSVC.dup
instead ofmov
for MSVC.