[PATCH] D109597: [DebugInfo] Fix of crash due to DwarfUnit::getOrCreateContextDIE returning NULL

[David Blaikie via Phabricator via llvm-commits]

dblaikie added inline comments.


================
Comment at: llvm/test/DebugInfo/X86/dwarfdump-ContextDICompileUnit.ll:3
+
+; RUN: llc %s -filetype=obj -o - | llvm-dwarfdump - | FileCheck %s
+
----------------
The title of this file seems misleading - if the bug/crash/issue is not in dwarfdump, we wouldn't usually put the word 'dwarfdump' in the test file name (like we don't put opt/llc in the name of tests that use those tools to test some transformation/lowering pass)


================
Comment at: llvm/test/DebugInfo/X86/dwarfdump-ContextDICompileUnit.ll:3
+
+; RUN: llc %s -filetype=obj -o - | llvm-dwarfdump - | FileCheck %s
+
----------------
dblaikie wrote:
> The title of this file seems misleading - if the bug/crash/issue is not in dwarfdump, we wouldn't usually put the word 'dwarfdump' in the test file name (like we don't put opt/llc in the name of tests that use those tools to test some transformation/lowering pass)
Could you include something like the original C or C++ source code (or otherwise) that was used to generate this IR? (if this IR was generated from some flang input - including the instructions of how to recompile/regenerate the IR may be useful for legibility (reading the metadata is a bit painful sometimes, so including some high level description of what's going on can be useful))

It'd be interesting to know/have more details (in this test file, in the patch description, etc) about how the problematic situation arose, and why it hadn't come up until flang/what did flang do differently that triggered this?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D109597/new/

https://reviews.llvm.org/D109597