[X86][tablgen] Remove predicates for definitions of KMOV b/c due to no pattern, NFCI

[craig.topper]

Could this information be useful to other tools in the future.

[skan]

In D148446#4271665, @craig.topper wrote:

Could this information be useful to other tools in the future.

I'm not very sure, I think there is basically no probability of using it because tools should need CPUID instead of predicate. Predicate does not exactly correspond to CPUID, for example we have predicate NoAVX512, NoAVX and NoVLX_Or_NoVAES. I am trying to remove these predicates b/c they affect the test coverage.

[craig.topper]

In D148446#4271692, @skan wrote:

In D148446#4271665, @craig.topper wrote:

Could this information be useful to other tools in the future.

I'm not very sure, I think there is basically no probability of using it because tools should need CPUID instead of predicate. Predicate does not exactly correspond to CPUID, for example we have predicate NoAVX512, NoAVX and NoVLX_Or_NoVAES. I am trying to remove these predicates b/c they affect the test coverage.

How do they affect the test coverage?

[skan]

In D148446#4271702, @craig.topper wrote:

In D148446#4271692, @skan wrote:

In D148446#4271665, @craig.topper wrote:

Could this information be useful to other tools in the future.

I'm not very sure, I think there is basically no probability of using it because tools should need CPUID instead of predicate. Predicate does not exactly correspond to CPUID, for example we have predicate NoAVX512, NoAVX and NoVLX_Or_NoVAES. I am trying to remove these predicates b/c they affect the test coverage.

How do they affect the test coverage?

No test fails after removing the predicates. So these code is untested.

[craig.topper]

llvm/lib/Target/X86/X86InstrAVX512.td

2885

Aren't there patterns for load/store in avx512_mask_mov?

[craig.topper]

llvm/lib/Target/X86/X86InstrAVX512.td

2885

It probably doesn't fail any tests because the types involved are only legal when the predicate is true or because lowering knew to change the load/store because it wasn't supported.

[skan]

llvm/lib/Target/X86/X86InstrAVX512.td

2885

Ah, you're right. I didn't notice there is pattern in the definition of avx512_mask_mov. Now I believe this change does not make sense.

llvm/lib/Target/X86/X86InstrAVX512.td

def kr : I<opc_kr, MRMSrcReg, (outs KRC:$dst), (ins GRC:$src),
                !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"), []>,
                Sched<[WriteMove]>;
     def rk : I<opc_rk, MRMSrcReg, (outs GRC:$dst), (ins KRC:$src),
                !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"), []>,
                Sched<[WriteMove]>;
   }
 }
 
-let Predicates = [HasDQI] in
-  defm KMOVB : avx512_mask_mov<0x90, 0x90, 0x91, "kmovb", VK8, v8i1, i8mem>,
-               avx512_mask_mov_gpr<0x92, 0x93, "kmovb", VK8, GR32>,
-               VEX, PD;
-
-let Predicates = [HasAVX512] in
-  defm KMOVW : avx512_mask_mov<0x90, 0x90, 0x91, "kmovw", VK16, v16i1, i16mem>,
-               avx512_mask_mov_gpr<0x92, 0x93, "kmovw", VK16, GR32>,
-               VEX, PS;
-
-let Predicates = [HasBWI] in {
-  defm KMOVD : avx512_mask_mov<0x90, 0x90, 0x91, "kmovd", VK32, v32i1,i32mem>,
-               VEX, PD, REX_W;
-  defm KMOVD : avx512_mask_mov_gpr<0x92, 0x93, "kmovd", VK32, GR32>,
-               VEX, XD;
-  defm KMOVQ : avx512_mask_mov<0x90, 0x90, 0x91, "kmovq", VK64, v64i1, i64mem>,
-               VEX, PS, REX_W;
-  defm KMOVQ : avx512_mask_mov_gpr<0x92, 0x93, "kmovq", VK64, GR64>,
-               VEX, XD, REX_W;
-}
 
 // GR from/to mask register
 def : Pat<(v16i1 (bitconvert (i16 GR16:$src))),
           (COPY_TO_REGCLASS (i32 (INSERT_SUBREG (IMPLICIT_DEF), GR16:$src, sub_16bit)), VK16)>;
 def : Pat<(i16 (bitconvert (v16i1 VK16:$src))),
           (EXTRACT_SUBREG (i32 (COPY_TO_REGCLASS VK16:$src, GR32)), sub_16bit)>;
 def : Pat<(i8 (trunc (i16 (bitconvert (v16i1 VK16:$src))))),
           (EXTRACT_SUBREG (i32 (COPY_TO_REGCLASS VK16:$src, GR32)), sub_8bit)>;