[llvm-bugs] [Bug 28202] New: [mc] Some F64 insns (v_add_f64 etc.) -- negative inline constants are not supported.

[via llvm-bugs]

https://llvm.org/bugs/show_bug.cgi?id=28202

            Bug ID: 28202
           Summary: [mc] Some F64 insns (v_add_f64 etc.) -- negative
                    inline constants are not supported.
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Backend: AMDGPU
          Assignee: artem.tamazov at amd.com
          Reporter: artem.tamazov at amd.com
                CC: llvm-bugs at lists.llvm.org, nikolay.haustov at amd.com
    Classification: Unclassified

Created attachment 16583
  --> https://llvm.org/bugs/attachment.cgi?id=16583&action=edit
00204.tests_tg_gfx7_asm_lit1_bug.06.08.min.v_add_etc_f64--iconst_encoding.zip

Example of failing instruction:
> v_add_f64 v[0:1], -1, 0

The resulting encoding is always with neg modifier, in spite that we have
suitable negative inline constants.

Note that in the instructions which have both 32- and 64-bit (VOP3) encodings
that would lead to unnecessary usage of VOP3.

Perhaps we need to introduce explicit syntax form of neg modifier which would
allow to distinguish negative constants from neg modifier applied to constant.

Tested with r273139 (git # 4a94acfea8bfd6eb6b5809dd40f03e1fa42e673a)

GFX7 lit tests attached.

-- 
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/20160620/652b680a/attachment.html>