[llvm-bugs] [Bug 42590] New: [AMDGPU][MC] Incorrect detection of unsupported instructions on errors

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=42590

            Bug ID: 42590
           Summary: [AMDGPU][MC] Incorrect detection of unsupported
                    instructions on errors
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Backend: AMDGPU
          Assignee: unassignedbugs at nondot.org
          Reporter: dpreobrazhensky at luxoft.com
                CC: llvm-bugs at lists.llvm.org

Assembler often reports an operand error when the actual reason is that
instruction is not supported. For example, the following instruction is not
supported on GFX8, but assembler issues an "invalid operand" error:

    v_swap_b32 v1, 0

A related problem is that when assembler do report "instruction not supported
on this GPU", it may also be incorrect. These errors are caused by unsupported
operands or their combinations. For example:

    v_add_co_ci_u32_e32 v3, vcc_lo, v3, v4, vcc_lo

When this instruction is assembled for GFX10 with wavesize=64, assembler
incorrectly reports that the instruction is not supported which is not the
case.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20190711/bd9a3cef/attachment.html>