Fix of test37 subtest 13 (test37m)

[Rene Zatvo]

Hi,

the test37m checks whether setjmp/longjmp restores signal mask. I
don't think that setjmp/longjmp should take
care about them even if minix implementation does. It would be better
to change those functions to sigsetjmp/siglongjmp. Correct?

--
Rene

[Erik van der Kouwe]

Hi,

Thanks for the idea. However, although POSIX leaves signal saving
unspecified for setjmp, MINIX specifies in the man-page that setjmp
does save signals. Not saving signals would be inconsistent with
MINIX' specification of its own API and might break existing MINIX
apps. Therefore, it has to be tested.

However, you are right in pointing out that MINIX does not test
sigsetjmp/siglongjmp yet. I will add tests for these functions, with
both values for savemask (MINIX specfies that it does not save the
signal mask if the parameter is 0, something that POSIX also leaves
unspecified).

With kind regards,
Erik

[Rene Zatvo]

Hi,

On Jan 11, 10:30 pm, Erik van der Kouwe <erik...@gmail.com> wrote:
> Hi,
>
> Thanks for the idea. However, although POSIX leaves signal saving
> unspecified for setjmp, MINIX specifies in the man-page that setjmp
> does save signals. Not saving signals would be inconsistent with
> MINIX' specification of its own API and might break existing MINIX
> apps. Therefore, it has to be tested.

Thanks for clarification. To be honest I didn't check Minix's man page
just followed the opengroup's web.

>
> However, you are right in pointing out that MINIX does not test
> sigsetjmp/siglongjmp yet. I will add tests for these functions, with
> both values for savemask (MINIX specfies that it does not save the
> signal mask if the parameter is 0, something that POSIX also leaves
> unspecified).

Yes, such tests will be ok.

--
Rene

[Erik van der Kouwe]

Revision 5874.