[Openmp-commits] [PATCH] D135795: Avoid exporting 80-bit fp functions for architectures other than Intel.

[Malhar via Phabricator via Openmp-commits]

malharJ added inline comments.


================
Comment at: openmp/runtime/src/dllexports:819-820
     # These are specific to x86 and x64
     %ifndef arch_64
     %ifndef arch_aarch64
 
----------------
jlpeyton wrote:
> Replace these two lines with `#ifdef IS_IA_ARCH` (and corresponding `%endif`)
I'm a little unsure if that is having the same semantics ?
If I understand correctly you have suggested this because of the comment on line 818 ?

but I see 
```
%ifdef arch_32
```
on line 839 (and it is inside the ifdef)


================
Comment at: openmp/runtime/src/dllexports:1246-1248
+    %endif
     __kmpc_atomic_float10_max_cpt          2141
     __kmpc_atomic_float10_min_cpt          2142
----------------
jlpeyton wrote:
> Why aren't the _cpt versions guarded as well?
I've not touched these functions as this patch is mainly a fix (for the github issue: #56349) for 
functions affected by one of my earlier patches, D117473.

I can try and make a separate patch later.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D135795/new/

https://reviews.llvm.org/D135795