This fixes a useless filecheck and wrong comment for always-inline.ll. Testing
has been done using ninja check-llvm and llvm-lit always-inline.ll --show-all.
Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Updating D127815: [test][AlwaysInline]:Correct comment and file check for always-inline.ll
llvm/test/Transforms/Inline/always-inline.ll | ||
---|---|---|
116 | Should we have check like below after the label? ; CHECK: call void @inner6 |
llvm/test/Transforms/Inline/always-inline.ll | ||
---|---|---|
116 | Yes, that can be added but do you think the functionality that it will check gets already checked at line 131 and that is sufficient, or this CHECK will indeed check for something new? |
LGTM, thanks.
llvm/test/Transforms/Inline/always-inline.ll | ||
---|---|---|
116 | The check at line 131 is for a different function scope, i.e, outer6. Note that the check at line 130 signals that the checker is in outer6 scope when it is there. I was thinking about adding a check inside the inner6 scope. On the second thought, I might misunderstand the original author's intention, which may be to check inner6 was not inlined into outer6. To that extent the current change is good. |
llvm/test/Transforms/Inline/always-inline.ll | ||
---|---|---|
116 | I think the intent behind the checks on inner functions is: if they are inlined, are they also successfully eliminated from IR? ; CHECK-NOT: @inner4( for example. If that's correct, then agreed this can remain as is. |
Should we have check like below after the label?