Re: [PATCH v8 01/10] Make mcheck tests conditional on GLIBC_2.24 or earlier

[Carlos O'Donell <carlos@redhat.com>]

On 7/13/21 3:38 AM, Siddhesh Poyarekar wrote:
> Targets with base versions later than 2.24 won't have
> __malloc_initialize_hook because of which the tests will essentially
> be the same as the regular malloc tests.  Avoid running them instead
> and save time.

Use have-GLIBC_2.23 for the release before the deprecation of a symbol.

Post v9 please and I'll ACK it quickly.

Tested-by: Carlos O'Donell <carlos@redhat.com>

> ---
>  malloc/Makefile | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/malloc/Makefile b/malloc/Makefile
> index 47369b6084..d15729569b 100644
> --- a/malloc/Makefile
> +++ b/malloc/Makefile
> @@ -78,6 +78,8 @@ tests-exclude-malloc-check = tst-malloc-check tst-malloc-usable \
>  # Run all tests with MALLOC_CHECK_=3
>  tests-malloc-check = $(filter-out $(tests-exclude-malloc-check),$(tests))
>  
> +# -lmcheck needs __malloc_initialize_hook, which was deprecated in 2.24.

OK. Wow, 5 years ago.

> +ifeq ($(have-GLIBC_2.24)$(build-shared),yesyes)

Should be have-GLIBC_2.23, since if it was deprecated in 2.24, a new arch in 2.24
would not have it and it should be tested. You always want the conditional to
trigger on the last release to have the symbol.

>  # Tests that don't play well with mcheck.  They are either bugs in mcheck or
>  # the tests expect specific internal behavior that is changed due to linking to
>  # libmcheck.a.
> @@ -100,6 +102,7 @@ tests-exclude-mcheck = tst-mallocstate \
>  	tst-reallocarray
>  
>  tests-mcheck = $(filter-out $(tests-exclude-mcheck), $(tests))
> +endif

OK.

>  
>  routines = malloc morecore mcheck mtrace obstack reallocarray \
>    scratch_buffer_dupfree \
> 


-- 
Cheers,
Carlos.