https://bugs.llvm.org/show_bug.cgi?id=51320
Bug ID: 51320
Summary: __constant__ variable in anonymous namespace is
optimized as if it could not be modified
Product: clang
Version: 12.0
Hardware: PC
OS: Linux
Status: NEW
Severity: normal
Priority: P
Component: CUDA
Assignee: unassignedclangb...@nondot.org
Reporter: kiw...@outlook.com
CC: llvm-bugs@lists.llvm.org
Created attachment 25102
--> https://bugs.llvm.org/attachment.cgi?id=25102&action=edit
code sample showcasing the issue
The best way to understand it is by looking at the attached code sample.
I tested it with: clang++ -L/usr/local/cuda/lib64 -lcudart
--cuda-gpu-arch=sm_30 -O3 main.cu && ./a.out
It works without `-O3`.
Expected output:
good = 10
bad = 20
good = 110
bad = 120
Bad output:
good = 10
bad = 20
good = 110
bad = 20
--
You are receiving this mail because:
You are on the CC list for the bug._______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
