[PATCH] D63732: [UpdateTestChecks][NFC] Remove entries with same prefix

Jinsong Ji via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Jun 24 12:45:02 PDT 2019


jsji marked 2 inline comments as done.
jsji added inline comments.


================
Comment at: llvm/utils/UpdateTestChecks/asm.py:293
       'i686': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
       'x86': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
       'i386': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
----------------
lebedev.ri wrote:
> Is `x86` a real triple?
> I've never seen it used.
Good question. 
enum `Triple::x86` is a valid enum, but it actually maps to `"i386", "i486", "i586", "i686", "i786", "i886", "i986"` in triple strings.

However, in this script, we will return `x86` in `get_triple_from_march` for unsupported `-march` when there is not triple specified in IR or command line.

So `x86` is a valid entry here.



================
Comment at: llvm/utils/UpdateTestChecks/asm.py:294
       'x86': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
       'i386': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
       'aarch64': (scrub_asm_arm_eabi, ASM_FUNCTION_AARCH64_RE),
----------------
Since `x86` is not a real triple, we are actually missing support for `i486`, `i586`,...

There is only one `i586` triple case found in current LLVM test.
`llvm/test/CodeGen/X86/constructor.ll`
So this is not a big issue.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D63732/new/

https://reviews.llvm.org/D63732





More information about the llvm-commits mailing list