[PATCH] D107994: Making the code compliant to the documentation about Floating Point support default values for C/C++. FPP-MODEL=PRECISE enables FFP-CONTRACT (FMA is enabled).

[Zahira Ammarguellat via Phabricator via cfe-commits]

zahiraam added a comment.

In D107994#3115589 <https://reviews.llvm.org/D107994#3115589>, @thakis wrote:

> This fixes tests on macOS, but I don't know if the space is there intentionally to check for presence of any attributes:
>
>   % git diff
>   diff --git a/clang/test/CodeGen/ffp-contract-option.c b/clang/test/CodeGen/ffp-contract-option.c
>   index 04d23d5f90b7..327722f56f4e 100644
>   --- a/clang/test/CodeGen/ffp-contract-option.c
>   +++ b/clang/test/CodeGen/ffp-contract-option.c
>   @@ -63,7 +63,7 @@
>    // RUN: --check-prefixes=CHECK,CHECK-FPC-ON
>   
>    float mymuladd(float x, float y, float z) {
>   -  // CHECK: define {{.*}} float @mymuladd(float noundef %x, float noundef %y, float noundef %z)
>   +  // CHECK: define{{.*}} float @mymuladd(float noundef %x, float noundef %y, float noundef %z)
>      return x * y + z;
>      // expected-warning{{overriding '-ffp-contract=fast' option with '-ffp-contract=on'}}
>   
>   diff --git a/clang/test/CodeGen/ffp-model.c b/clang/test/CodeGen/ffp-model.c
>   index ee0eedb2718f..98d31106a119 100644
>   --- a/clang/test/CodeGen/ffp-model.c
>   +++ b/clang/test/CodeGen/ffp-model.c
>   @@ -19,7 +19,7 @@
>    // RUN: --check-prefixes CHECK,CHECK-FAST1
>   
>    float mymuladd(float x, float y, float z) {
>   -  // CHECK: define {{.*}} float @mymuladd(float noundef %x, float noundef %y, float noundef %z)
>   +  // CHECK: define{{.*}} float @mymuladd(float noundef %x, float noundef %y, float noundef %z)
>      return x * y + z;
>      // CHECK-FAST: fmul fast float
>
> If this is the right fix, please land it. Else I'll revert again in 30 min or so; bots have been broken for hours by now.





================
Comment at: clang/test/CodeGen/ffp-model.c:4
+// RUN: | FileCheck %s \
+// RUN: --check-prefixes=CHECK,CHECK-FAST --strict-whitespace
+
----------------
zahiraam wrote:
> andrew.w.kaylor wrote:
> > Why did you add the strict-whitespace option?
> Because on some targets, the IR generated for the define is "define {{.*}}float @mymuladd" and on and some others it is 
> "define {{.*}} float @mymuladd" (space between {{.*}} and float.
I was trying to fix the issue with the define function check. 
With this run command:

// RUN: %clang -S -emit-llvm -fno-fast-math -emit-llvm -target x86_64-apple-macosx10.11 %s -o -
I get this IR:

define float @mymuladd(float noundef %x, float noundef %y, float noundef %z)

This wouldn't be represented by

// CHECK: define {{.*}} float @mymuladd

but by

// CHECK: define{{.*}} float @mymuladd

or

// CHECK: define {{.*}}float @mymuladd 
But I get replacing with // CHECK: define{{.*}} float @mymuladd
Should solve the issue. Right?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D107994/new/

https://reviews.llvm.org/D107994