[LLVMbugs] [Bug 11042] New: Preserve all metadata when removing dead arguments
bugzilla-daemon at llvm.org
bugzilla-daemon at llvm.org
Thu Sep 29 17:44:31 PDT 2011
http://llvm.org/bugs/show_bug.cgi?id=11042
Summary: Preserve all metadata when removing dead arguments
Product: libraries
Version: trunk
Platform: All
OS/Version: All
Status: NEW
Severity: enhancement
Priority: P
Component: Transformation Utilities
AssignedTo: unassignedbugs at nondot.org
ReportedBy: wujingyue at gmail.com
CC: llvmbugs at cs.uiuc.edu
At around Line 257 and Line 835, DAE preserves debug locations after removing
the dead arguments, by copying the debug location from the old call to the new
call. Should we do the same thing for other metadata (e.g. customized metadata)
as well? Would it break anything?
Btw, this issue appears in LLVM 2.9 and some earlier versions as well.
Proposed patch:
Index: lib/Transforms/IPO/DeadArgumentElimination.cpp
===================================================================
--- lib/Transforms/IPO/DeadArgumentElimination.cpp (revision 140827)
+++ lib/Transforms/IPO/DeadArgumentElimination.cpp (working copy)
@@ -254,7 +254,14 @@
if (cast<CallInst>(Call)->isTailCall())
cast<CallInst>(New)->setTailCall();
}
- New->setDebugLoc(Call->getDebugLoc());
+ // by Jingyue
+ // Copy all metadata
+ if (Call->hasMetadata()) {
+ SmallVector<std::pair<unsigned, MDNode*>, 4> TheMDs;
+ Call->getAllMetadata(TheMDs);
+ for (unsigned i = 0, e = TheMDs.size(); i != e; ++i)
+ New->setMetadata(TheMDs[i].first, TheMDs[i].second);
+ }
Args.clear();
@@ -832,7 +839,14 @@
if (cast<CallInst>(Call)->isTailCall())
cast<CallInst>(New)->setTailCall();
}
- New->setDebugLoc(Call->getDebugLoc());
+ // by Jingyue
+ // Copy all metadata
+ if (Call->hasMetadata()) {
+ SmallVector<std::pair<unsigned, MDNode*>, 4> TheMDs;
+ Call->getAllMetadata(TheMDs);
+ for (unsigned i = 0, e = TheMDs.size(); i != e; ++i)
+ New->setMetadata(TheMDs[i].first, TheMDs[i].second);
+ }
Args.clear();
--
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
More information about the llvm-bugs
mailing list