Hi!
On Mon, Nov 29, 2021 at 01:31:48PM +0100, Alejandro Colomar (man-pages) wrote:
> > (a) is always sizeof==time_t+8, and
> > (b) has tv_nsec as __syscall_slong_t
> > *and* !is_same<__syscall_slong_t, long>
> > if using ILP64 syscalls on an LP32 system, i.e. on x32.
> I guess you meant LP64, not ILP64?
> AFAIK, glibc doesn't support ILP64.
Yes, good catch! Def. a typo.
> But this would be a simpler commit if we only had one definition.
I was going to suggest it originally, but held up because, well,
before writing this patchset I didn't even know about
system_data_types.7, and it seemed that most other declarations are
in-line (which is sensible, I think, and in line with the common
use-case).
> Do you agree in changing the whole patch set in that (IMO simpler) way?
I mean, IMO in-line is better to use, but a struct this short and
relatively confusing when you actually unroll the types
makes sense to be banished.
So, just so we're clear on the direction of this before I re-write
a lot of these sections, do these (cherry-picking a few):
clock_getres.2:
The res and tp arguments are timespec(3) structures, as specified in
<time.h>.
mq_receive.3:
This value is an absolute timeout in seconds and nanoseconds since
the Epoch, 1970-01-01 00:00:00 +0000 (UTC),
specified in struct timespec(3).
pthread_tryjoin_np.3:
If the timeout expires before thread terminates, the call returns an
error. The abstime argument is a struct timespec(3), specifying an
absolute time measured since the Epoch (see time(2)).
+ adding timespec(3) to the SEE ALSO sexions
Sound about right?
Best,
наб
Attachment:
signature.asc
Description: PGP signature
