[PATCH] D82698: [NewPM] make parsePassPipeline parse adaptor-wrapped user passes

[Arthur Eubanks via Phabricator via llvm-commits]

aeubanks added inline comments.


================
Comment at: llvm/test/Other/pass-pipeline-parsing.ll:182
+; CHECK-ADAPTORS: Running pass: ModuleToFunctionPassAdaptor<{{.*}}NoOpFunctionPass>
+; CHECK-ADAPTORS: Running pass: ModuleToFunctionPassAdaptor<{{.*}}FunctionToLoopPassAdaptor<{{.*}}NoOpLoopPass>{{.*}}>
+; CHECK-ADAPTORS: Running pass: ModuleToPostOrderCGSCCPassAdaptor<{{.*}}NoOpCGSCCPass>
----------------
ychen wrote:
> aeubanks wrote:
> > ychen wrote:
> > > aeubanks wrote:
> > > > what about a separate check for
> > > > Running pass: FunctionToLoopPassAdaptor<{{.*}}NoOpLoopPass>
> > > > and NoOpLoopPass itself?
> > > I'm happy to do that. What is the motivation?
> > To make sure that the FunctionToLoopPassAdaptor itself isn't skipped for some reason, and the same with the NoOpLoopPass.
> Is line 191 what you refer to?
no, that one is a different no-op-loop in -passes='module(no-op-function,no-op-loop,no-op-cgscc,cgscc(no-op-function,no-op-loop),function(no-op-loop))'.

line 191 refers to the very last no-op-loop, I meant the first one. basically I think every adaptor "Running pass" message should be checked.
1 for the first no-op-function (module -> function), 2 for the first no-op-loop (module -> function -> loop), 1 for the first no-op-cgscc (module -> cgscc), 1 for the second no-op-function (cgscc -> function), 2 for the second no-op-loop (cgscc -> function -> loop), and 1 for the last no-op-loop (function -> loop)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D82698/new/

https://reviews.llvm.org/D82698