[PATCH] Fix Bug in SROA transformation [PR18615]

[Karthik Bhat]

  Hi Duncan,
  Thanks for having a look at the patch. Added a test case to capture PR18615 in clang/test/CodeGen/ Please let me know if this the right place for this test case as i'm a bit new to llvm.
  Please let me know your inputs about the patch.
  Thanks
  Karthik Bhat

Hi chandlerc,

http://llvm-reviews.chandlerc.com/D2759

CHANGE SINCE LAST DIFF
  http://llvm-reviews.chandlerc.com/D2759?vs=7045&id=7154#toc

Files:
  lib/Transforms/Scalar/SROA.cpp
  test/CodeGen/pr18615.c

Index: lib/Transforms/Scalar/SROA.cpp
===================================================================
--- lib/Transforms/Scalar/SROA.cpp
+++ lib/Transforms/Scalar/SROA.cpp
@@ -524,6 +524,10 @@
     // Insert the use now that we've fixed up the splittable nature.
     insertUse(II, Offset, Size, /*IsSplittable=*/Inserted && Length);
 
+    //If the insert resulted in marking the instruction as dead
+    //i.e. we didn't end up pushing the usage into map then return
+    if (VisitedDeadInsts.count(&II))
+      return;
     // Check that we ended up with a valid index in the map.
     assert(S.Slices[PrevIdx].getUse()->getUser() == &II &&
            "Map index doesn't point back to a slice with this user.");
Index: test/CodeGen/pr18615.c
===================================================================
--- test/CodeGen/pr18615.c
+++ test/CodeGen/pr18615.c
@@ -0,0 +1,18 @@
+// RUN: %clang_cc1 -emit-llvm -O1 %s  -o /dev/null
+int a;
+
+struct S0
+{
+  int f0;
+  int f1;
+  int f2;
+};
+
+void
+fn1 ()
+{
+  int b = -1; 
+  struct S0 f[1];
+  if (a)
+    f[0] = f[b];
+}
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D2759.2.patch
Type: text/x-patch
Size: 1087 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140216/779045ba/attachment.bin>