https://llvm.org/bugs/show_bug.cgi?id=27479

Quentin Colombet <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #3 from Quentin Colombet <[email protected]> ---
Fixed CodeGen/AArch64/aarch64-tbz.ll in r267206. When optimizing a branch we
were choosing the W or X variant of the opcode based on the value of the mask
used in the AND we were folding, whereas we should use the opcode of the AND.
Otherwise we may end up feeding 64-bit register into 32-bit variants!

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
llvm-bugs mailing list
[email protected]
http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs

Reply via email to