be more specific when testing for no fuse-ld warnings

[stuij]

Summary

This test, introduced in https://reviews.llvm.org/D87837, broke for our toolchain as this test triggered unrelated warnings. Being more specific about not expecting fuse-ld warnings won't invalidate the test, while playing a bit nicer with possible unrelated features.

[stuij]

slight change in commit message

[dmgreen]

In the past we have usually disabled the downstream warning for similar catch-all warning lines.

clang/test/Driver/fuse-ld.c

5

This line has a = on the end. Would the other warning below also have that? (Negative tests are often tricky).

[dmgreen]

(Not that I'm against this, either way sounds fine to me)

[MaskRay]

clang/test/Driver/fuse-ld.c

15

How does this line trigger unrelated warnings? Can you dump it?

[stuij]

Hi @MaskRay. Yes, so we're seeing a warning specific to our Armcompiler toolchain, so I'm guessing that isn't relevant to OSS LLVM:
armclang: warning: '--target=x86_64-unknown-linux' is not supported.

As David Green pointed out, we have a perfectly fine workaround. But I figured that a similar situation might crop up in OSS LLVM, and this way the test is a bit more future proof, and we might spare some future head-scratching.

However I feel bad already for wasting our time with such a minor change. Feel free to reject it :)

clang/test/Driver/fuse-ld.c

15

see my top-level comment

[stuij]

ping

[stuij]

Abandoned because lack of reaction for such an unimportant issue.

[MaskRay]

In D88566#2317248, @stuij wrote:

Hi @MaskRay. Yes, so we're seeing a warning specific to our Armcompiler toolchain, so I'm guessing that isn't relevant to OSS LLVM:
armclang: warning: '--target=x86_64-unknown-linux' is not supported.

The test uses '-###' and nothing related to X86 backend (-DLLVM_TARGETS_TO_BUILD does not include X86), so the intention may be quetionable.

As David Green pointed out, we have a perfectly fine workaround. But I figured that a similar situation might crop up in OSS LLVM, and this way the test is a bit more future proof, and we might spare some future head-scratching.

If this cannot be reproduced with the OSS LLVM, I am not sure you should adjust such a test.

clang/test/Driver/fuse-ld.c

15

The impoerant bit is that the original message disallows any warning. The new message with an incorrect 'fuse-ld' (instead of '-fuse-ld') seems really questionable.

[stuij]

If this cannot be reproduced with the OSS LLVM, I am not sure you should adjust such a test.

Ok, fair enough. Thanks for the comment.

clang/test/Driver/fuse-ld.c

15

Ai, a typo :(

clang/test/Driver/fuse-ld.c

 /// The absolute path warning is enabled by -Wfuse-ld-path and -Wextra.
 // RUN: %clang %s -### -target x86_64-unknown-linux -Wfuse-ld-path \
 // RUN:   -fuse-ld=/usr/local/bin/or1k-linux-ld 2>&1 | \
 // RUN:   FileCheck %s --check-prefix=CHECK-ABSOLUTE-LD
 // CHECK-ABSOLUTE-LD: warning: '-fuse-ld=' taking a path is deprecated. Use '--ld-path=' instead
 // CHECK-ABSOLUTE-LD: /usr/local/bin/or1k-linux-ld
 
 // RUN: %clang %s -### -target x86_64-unknown-linux -Wextra \
 // RUN:   -fuse-ld=/usr/local/bin/or1k-linux-ld 2>&1 | \
 // RUN:   FileCheck %s --check-prefix=CHECK-ABSOLUTE-LD
 
 // RUN: %clang %s -### -target x86_64-unknown-linux \
 // RUN:   -fuse-ld=/usr/local/bin/or1k-linux-ld 2>&1 | \
 // RUN:   FileCheck %s --check-prefix=CHECK-NO-WARN
-// CHECK-NO-WARN-NOT: warning:
+// CHECK-NO-WARN-NOT: warning: 'fuse-ld'
 
 // RUN: %clang %s -### \
 // RUN:     -target x86_64-unknown-freebsd 2>&1 \
 // RUN:   | FileCheck %s --check-prefix=CHECK-FREEBSD-LD
 // CHECK-FREEBSD-LD: ld
 
 // RUN: %clang %s -### -fuse-ld=bfd \
 // RUN:     --sysroot=%S/Inputs/basic_freebsd_tree \