[PATCH] D152540: [MC][MachO]Do not emit DWARF for no-personality case

[Vy Nguyen via Phabricator via llvm-commits]

oontvoo marked an inline comment as not done.
oontvoo added inline comments.


================
Comment at: llvm/lib/Target/AArch64/MCTargetDesc/AArch64AsmBackend.cpp:572
       return CU::UNWIND_ARM64_MODE_FRAMELESS;
-    if (!isDarwinCanonicalPersonality(FI->Personality) &&
+    if (FI->Personality && !isDarwinCanonicalPersonality(FI->Personality) &&
         !Ctxt->emitCompactUnwindNonCanonical())
----------------
jyknight wrote:
> oontvoo wrote:
> > jyknight wrote:
> > > Move the change into the isDarwinCanonicalPersonality function?
> > well, both of these two predicates are *already* in the function. (it is supposed to tell you if a given symbol is a canonical-darwin personality.)
> > BUT if it returns false, then it could be any of the following reason:
> >  - (1) the symbol is NULL
> >  - (2) the symbol is NOT MachO
> >  - (3) the symbol is NOT canonical personality symbol.
> > 
> > We only want to return DWARF-MODE here for reason number (3)
> > 
> > If we want to reuse a util function, then we ought to name it `isNonNullDarwinNonCanonicalPersonality`...
> > That's a bit hard to read.
> I'd say the util function should return true for NULL, because NULL is a canonical personality (it's always given index 0, right?)
> 
> How do you even get a !MachO symbol in here -- all of this is Mach-O-specific, right? Perhaps can just remove that clause?
Good point. 
I've updated the function and replace the isMachO check with just an assert instead. (who knows?!)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152540/new/

https://reviews.llvm.org/D152540