[lld] [LLD] Tombstone LocalTU entry in .debug_names (PR #70701)

David Blaikie via llvm-commits llvm-commits at lists.llvm.org
Mon Oct 30 13:24:53 PDT 2023


================
@@ -918,7 +917,8 @@ void InputSection::relocateNonAlloc(uint8_t *buf, ArrayRef<RelTy> rels) {
       continue;
 
     if (tombstone ||
-        (isDebug && (type == target.symbolicRel || expr == R_DTPREL))) {
+        ((isDebug && (type == target.symbolicRel || expr == R_DTPREL))) ||
----------------
dwblaikie wrote:

While we're waiting for an expert opinion, if youre bored, you could try removing the `type == target.symbolicRel` check and running the lld tests, about 5 tests fail - investigating one fo those failures to see what goes wrong in the absence of that condition might help point the way to understanding how to generalize the condition enough to cover the `.debug_names` situation, without generalizing too far and breaking those tests

https://github.com/llvm/llvm-project/pull/70701


More information about the llvm-commits mailing list