kito-cheng added inline comments.
================
Comment at: clang/include/clang/Support/RISCVVIntrinsicUtils.h:100
+ bool MU = false;
+ bool MA = false;
+ bool IntrinsicWithoutMU = false;
----------------
Maybe use enum value for tail/mask policy? *U and *A are mutually exclusive, so
I feel they should just using same variables:
```
bool PolicyNone = false;
enum PolicyType {
Undisturbed,
Agnostic,
Omit, // No policy required.
};
PolicyType TailPolicy = Omit;
PolicyType MastPolicy = Omit;
bool IntrinsicWithoutMU = false;
```
================
Comment at: clang/include/clang/Support/RISCVVIntrinsicUtils.h:436
+
+ assert(false && "unsupport policy");
+ return 0;
----------------
llvm_unreachable
================
Comment at: clang/utils/TableGen/RISCVVEmitter.cpp:167
// We had initialized DefaultPolicy as TU/TUMU in CodeGen function.
- if (RVVI->getDefaultPolicy() != Policy::TU &&
- RVVI->getDefaultPolicy() != Policy::TUMU && !RVVI->hasPassthruOperand()
&&
+ if (RVVI->getDefaultPolicyBits() != 0 && !RVVI->hasPassthruOperand() &&
!RVVI->hasManualCodegen() && RVVI->hasVL())
----------------
Use `!RVVI->isTUPolicy() && !RVVI->isTUMUPolicy()` instead of
`RVVI->getDefaultPolicyBits() != 0`, that's kind of too magical predication.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D139995/new/
https://reviews.llvm.org/D139995
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
