summaryrefslogtreecommitdiffstats
path: root/llvm/lib/Target/README.txt
diff options
context:
space:
mode:
authorNick Lewycky <nicholas@mxc.ca>2009-07-09 04:03:30 +0000
committerNick Lewycky <nicholas@mxc.ca>2009-07-09 04:03:30 +0000
commit4e705e421cd84413359eb7dd21df6990ec133c75 (patch)
tree49638ca87d49beeee1795e38d07752e8ea780c09 /llvm/lib/Target/README.txt
parentdab7b06de971bffb7abced1ad1de9617311e3e8e (diff)
downloadbcm5719-llvm-4e705e421cd84413359eb7dd21df6990ec133c75.tar.gz
bcm5719-llvm-4e705e421cd84413359eb7dd21df6990ec133c75.zip
Move PR4517 to README.txt.
llvm-svn: 75098
Diffstat (limited to 'llvm/lib/Target/README.txt')
-rw-r--r--llvm/lib/Target/README.txt22
1 files changed, 22 insertions, 0 deletions
diff --git a/llvm/lib/Target/README.txt b/llvm/lib/Target/README.txt
index f68cf0e40df..be7bb3a53ec 100644
--- a/llvm/lib/Target/README.txt
+++ b/llvm/lib/Target/README.txt
@@ -1677,3 +1677,25 @@ Instcombine should be able to optimize away the loads (and thus the globals).
//===---------------------------------------------------------------------===//
+
+I saw this constant expression in real code after llvm-g++ -O2:
+
+declare extern_weak i32 @0(i64)
+
+define void @foo() {
+ br i1 icmp eq (i32 zext (i1 icmp ne (i32 (i64)* @0, i32 (i64)* null) to i32),
+i32 0), label %cond_true, label %cond_false
+cond_true:
+ ret void
+cond_false:
+ ret void
+}
+
+That branch expression should be reduced to:
+
+ i1 icmp eq (i32 (i64)* @0, i32 (i64)* null)
+
+It's probably not a perf issue, I just happened to see it while examining
+something else and didn't want to forget about it.
+
+//===---------------------------------------------------------------------===//
OpenPOWER on IntegriCloud