Re: [patch, testsuite] Skip gnu-ifunc tests if building the testcase fails

[Tom Tromey <tom at tromey dot com>]

>>>>> "Sandra" == Sandra Loosemore <sandra@codesourcery.com> writes:

Sandra> Here's a revised patch that adapts the same compile test that gcc
Sandra> uses. Is this one OK?

Thanks for the patch.

Sandra> +# Return true if the IFUNC feature is unsupported.
Sandra> +gdb_caching_proc skip_ifunc_tests {
Sandra> +    # Set up, compile, and execute a test program.
Sandra> +    # Include the current process ID in the file names to prevent conflicts
Sandra> +    # with invocations for multiple testsuites.
Sandra> +    set src [standard_temp_file ifunc[pid].c]
Sandra> +    set obj [standard_temp_file ifunc[pid].o]
Sandra> +
Sandra> +    verbose -log "checking for ifunc support"
Sandra> +    gdb_produce_source $src {
Sandra> +	extern void f_ ();
Sandra> +	typedef void F (void);
Sandra> +	F* g (void) { return &f_; }
Sandra> +	void f () __attribute__ ((ifunc ("g")));
Sandra> +    }
Sandra> +
Sandra> +    set lines [gdb_compile $src $obj object {quiet}]
Sandra> +    file delete $src
Sandra> +    file delete $obj


I wonder whether this could use gdb_can_simple_compile.  A recent patch
changed a bunch of other caching procs to use this helper function
instead.

I didn't examine it in detail, though, so if there is some reason it
can't be used, that's fine.

But if it can be used, this is ok with that change.

Tom