Re: [PATCH v4] selftests/mm: add process_madvise() tests

[Zi Yan <ziy@nvidia.com>]

On 10 Jul 2025, at 9:42, Mark Brown wrote:

> On Thu, Jul 10, 2025 at 07:22:49PM +0800, wang lian wrote:
>
>> Add tests for process_madvise(), focusing on verifying behavior under
>> various conditions including valid usage and error cases.
>
>> --- a/tools/testing/selftests/mm/guard-regions.c
>> +++ b/tools/testing/selftests/mm/guard-regions.c
>
>> -static void handle_fatal(int c)
>> -{
>> -	if (!signal_jump_set)
>> -		return;
>> -
>> -	siglongjmp(signal_jmp_buf, c);
>> -}
>
> I see from looking later in the patch that you're factoring this out of
> the guard regions test into vm_util.c so that it can be used by your new
> test.  This is good and sensible but it's a bit surprising, especially
> since your changelog only said you were adding a new test.  It would be
> better to split this out into a separate refactoring patch that just
> does the code motion, as covered in submitting-patches.rst it's better
> if changes just do one thing.
>
>> +#include <linux/pidfd.h>
>> +#include <linux/uio.h>
>
> Does this work without 'make headers_install' for the systems that were
> affectd by missing headers?  Lorenzo mentioned that we shouldn't depend
> on that for the mm tests (I'm not enthusiastic about that approach
> myself, but if it's what mm needs).

No. “make headers_install” is still needed. I tried to get it compiled
without it but failed. It seems that a lot of files will need to be
copied to tools/include from “make headers”.

Best Regards,
Yan, Zi