[PATCH] D37655: IR: Represent -ggnu-pubnames with a flag on the DICompileUnit.

[Peter Collingbourne via Phabricator via llvm-commits]

pcc added a comment.

In https://reviews.llvm.org/D37655#867084, @dblaikie wrote:

> Reckon maybe a test case with two units - one using gnu pubnames, one not, would be worth having to demonstrate that all works fine & is appropriately respected?


Added, see `llvm/test/DebugInfo/X86/gnu-public-names-multiple-cus.ll`.



================
Comment at: llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp:412-422
+bool DwarfDebug::hasDwarfPubSections(DwarfCompileUnit &CU) const {
   // Opting in to GNU Pubnames/types overrides the default to ensure these are
   // generated for things like Gold's gdb_index generation.
-  if (GenerateGnuPubSections)
+  if (CU.getCUNode()->getGnuPubnames())
     return true;
 
   if (DwarfPubSections == Default)
----------------
dblaikie wrote:
> Looks like this function could be moved to be a member of DwarfCompileUnit? (& move the DwarfPubSections opt into the DwarfCompileUnit.cpp file?)
> 
> Just a thought - see how you feel about it, if you think it's better here, that's fine.
I think that makes sense, done.

What do you think about removing the `-generate-dwarf-pub-sections` flag entirely? It seems that we only use it in a handful of tests where it looks like it would suffice to use `-debugger-tune` instead. In any event, that can probably be done as a followup.


================
Comment at: llvm/test/DebugInfo/X86/gnu-public-names-gmlt.ll:1
-; RUN: llc -mtriple=x86_64-pc-linux-gnu -filetype=obj < %s -generate-gnu-dwarf-pub-sections | llvm-dwarfdump - | FileCheck --check-prefix=GPUB --check-prefix=CHECK %s
+; RUN: sed -e 's/gnuPubnames: false/gnuPubnames: true/' %s | llc -mtriple=x86_64-pc-linux-gnu -filetype=obj | llvm-dwarfdump - | FileCheck --check-prefix=GPUB --check-prefix=CHECK %s
 ; RUN: llc -mtriple=x86_64-pc-linux-gnu -filetype=obj < %s -generate-dwarf-pub-sections=Enable | llvm-dwarfdump - | FileCheck --check-prefix=PUB --check-prefix=CHECK %s
----------------
dblaikie wrote:
> Is this use of 'sed' in a RUN line portable to windows (I forget exactly which command line tools are available/dependendable in the lit environment there)?
I think so, we are using sed like this in a few places already: http://llvm-cs.pcc.me.uk/?q=RUN%3A+sed


https://reviews.llvm.org/D37655