[PATCH] D144154: [X86]Use Class to refactor ArithMetic td file in X86

[Kan Shengchen via Phabricator via llvm-commits]

skan added inline comments.


================
Comment at: llvm/lib/Target/X86/X86InstrArithmetic.td:804-812
+def Xi16x : X86TypeInfo<i16, "w", GR16, loadi16, i16mem,
+                       Imm16, i16imm, imm, i16i8imm, i16immSExt8,
+                       1, OpSize16, 0>;
+def Xi32x : X86TypeInfo<i32, "l", GR32, loadi32, i32mem,
+                       Imm32, i32imm, imm, i32i8imm, i32immSExt8,
+                       1, OpSize32, 0>;
+def Xi64x : X86TypeInfo<i64, "q", GR64, loadi64, i64mem,
----------------
The only difference between Xi64x and Xi64 is the "small use of immediate", which is defined in X86InstrInfo.td. e,g `imm_su` <-> `imm`

`imm_su` is not used here b/c IMUL are not binary operations. 

so the suffix x here is not appropriate.

I would suggest you add one field `SDPatternOperator ImmNoSuOperator` in `X86TypeInfo` and use `Xi16`, `Xi32`, `Xi64` for IMUL too.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144154/new/

https://reviews.llvm.org/D144154