[PATCH] D151834: Include math-errno with fast-math

[Zahira Ammarguellat via Phabricator via cfe-commits]

zahiraam added inline comments.


================
Comment at: clang/include/clang/Basic/FPOptions.def:30
 OPTION(BFloat16ExcessPrecision, LangOptions::ExcessPrecisionKind, 2, FPEvalMethod)
+OPTION(MathErrno, bool, 1, BFloat16ExcessPrecision)
 #undef OPTION
----------------
andrew.w.kaylor wrote:
> Does this mean MathErrno is tracked in both LangOpts and FPOptions?
Yes. So that we can check that if it's overridden by a pragma or an attribute. I don't see another way.


================
Comment at: clang/test/CodeGen/math-errno.c:33
+
+__attribute__((optnone))
+float f4(float x) { 
----------------
andrew.w.kaylor wrote:
> Can you add a runline with -O0. That should prevent all instances of the intrinsics, right?
Yes.


================
Comment at: clang/test/CodeGen/math-errno.c:49
+// FAST-LABEL: define dso_local nofpclass(nan inf) float @f4(float noundef nofpclass(nan inf) {{.*}})
+// FAST: call reassoc nnan ninf nsz arcp afn nofpclass(nan inf) float @sqrtf(float noundef nofpclass(nan inf) %0) #[[ATTR0:[0-9]+]]
+
----------------
andrew.w.kaylor wrote:
> I think the 'afn' flag here is a problem. The backend has no concept of errno, so 'afn' will be treated as allowing the function to be replaced.
Added all the commands that are triggered by fast-math in the RUN line command. 


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D151834/new/

https://reviews.llvm.org/D151834