mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2025-02-09 05:47:13 +00:00
240cdafd6c
The v2i64 argument is lowered to a bitcast of v4i32 build_vector. This would then attempt to use the i32-element as the source of the vector truncate. This really would need to collect 2 elements from the build_vector to produce the intended truncate. llvm-svn: 353202