[PATCH] D66309: Introduce infrastructure for an incremental port of SelectionDAG atomic load/store handling

Wed Aug 28 13:20:41 PDT 2019

reames marked an inline comment as done.
reames added inline comments.

================
Comment at: test/CodeGen/X86/atomic-unordered.ll:4
 ; RUN: llc -O3 < %s -mtriple=x86_64-linux-generic -verify-machineinstrs -mcpu=skylake | FileCheck --check-prefix=CHECK-O3 %s
+; RUN: llc -O3 < %s -mtriple=x86_64-linux-generic -verify-machineinstrs -mcpu=skylake -x86-experimental-unordered-atomic-isel | FileCheck --check-prefix=CHECK-EX %s

----------------
craig.topper wrote:
> xbolva00 wrote:
> > reames wrote:
> > > craig.topper wrote:
> > > > Can we add a common prefix to share with the two O3 run lines? Looks like our codegen is the same for the majority of the test cases.
> > > Not quite following the why behind this request?  Is there some (undocumented) feature of FileCheck which merges the check lines or something?
> > Possibly something like
> > -check-prefixes=CHECK,CHECK-O3
> > -check-prefixes=CHECK,CHECK-EX
> > 
> > (not tested, just saw it many times)
> Yeah that's it. The test update script understands this. FileCheck and the update script also understand "--check-prefix=CHECK --check-prefix=CHECK-O3" on the same command line as meaning the same thing. FileCheck will check any lines with the listed prefix. For the script to work, the common prefix should be placed first and more specific prefix placed last. Then it will favor the common prefix when possible.
Did not know about this, cool.  

Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66309/new/

https://reviews.llvm.org/D66309