Skip to content
This repository was archived by the owner on Mar 28, 2020. It is now read-only.

Commit 9ef1010

Browse files
committed
Add some release notes about the fix for PR26774
As suggested by Chandler on the review thread for D18634. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@267239 91177308-0d34-0410-b5e6-96231b3b80d8
1 parent dcbc34e commit 9ef1010

File tree

1 file changed

+10
-0
lines changed

1 file changed

+10
-0
lines changed

docs/ReleaseNotes.rst

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -87,6 +87,16 @@ Changes to the LLVM IR
8787
``llvm.masked.gather`` and ``llvm.masked.scatter`` were introduced to the
8888
LLVM IR to allow selective memory access for vector data types.
8989

90+
Changes to LLVM's IPO model
91+
---------------------------
92+
93+
LLVM no longer does inter-procedural analysis and optimization (except
94+
inlining) on functions with comdat linkage. Doing IPO over such
95+
functions is unsound because the implementation the linker chooses at
96+
link-time may be differently optimized than the one what was visible
97+
during optimization, and may have arbitrarily different observable
98+
behavior. See `PR26774 <http://llvm.org/PR26774>`_ for more details.
99+
90100
Changes to the ARM Backend
91101
--------------------------
92102

0 commit comments

Comments
 (0)