Re: [PATCH v3 1/6] selftests/mm: mseal, self_elf: fix missing __NR_mseal

[John Hubbard <jhubbard@nvidia.com>]

On 6/17/24 11:56 PM, David Hildenbrand wrote:
> On 18.06.24 04:24, John Hubbard wrote:
...
>> diff --git a/tools/testing/selftests/mm/seal_elf.c b/tools/testing/selftests/mm/seal_elf.c
>> index f2babec79bb6..27bf2f84231d 100644
>> --- a/tools/testing/selftests/mm/seal_elf.c
>> +++ b/tools/testing/selftests/mm/seal_elf.c
>> @@ -2,7 +2,7 @@
>>   #define _GNU_SOURCE
>>   #include <sys/mman.h>
>>   #include <stdint.h>
>> -#include <unistd.h>
>> +#include <asm-generic/unistd.h>
>>   #include <string.h>
>>   #include <sys/time.h>
>>   #include <sys/resource.h>
> 
> Still confused. Let's take a look at "microblaze".
> 
> arch/microblaze/include/asm/unistd.h
>   -> #include <uapi/asm/unistd.h>
> 
> arch/microblaze/include/uapi/asm/unistd.h
>   -> #include <asm/unistd_32.h>
>    -> Generated during "make headers"
> 
> usr/include/asm/unistd_32.h is generated via
> arch/microblaze/kernel/syscalls/Makefile with the syshdr command.
> 
> So we never end up including asm-generic/unistd.h directly on microblaze, but rather converts it (IIUC) to something else.
>

Yes.
  
> That will work as expected here?
> 

No. :)

The problem, and the source of confusion here, is that for most user
space programs, the header file inclusion behaves as you've mentioned
above. However, those programs are installed on a single computer that
has a single set of asm and kernel headers installed.

We are quite special here, because we are building a set of user space
programs that:

     a) Mostly avoids using the installed (distro) system header files.

     b) Must build (and run) on all supported CPU architectures

     c) Must occasionally use symbols that have so new that they have not
        yet been included in the distro's header files.

Doing (a) creates a new problem: how to get a set of cross-platform
headers that works in all cases.

Fortunately, asm-generic headers solve that one. Which is why we need to
use them here.

The reason this hasn't really come up yet, is that until now, the
kselftests requirement (which I'm trying to remove) was that "make
headers" must first be run. That allowed the selftests to get a snapshot
of sufficiently new header files that looked just like (and conflict
with) the installed system headers.

I can update the commit description with some of the above, if it helps.

thanks,
-- 
John Hubbard
NVIDIA