[PATCH 4/4] [y2038] lib: cobalt: clock: dispatch clock_settime

[PATCH 4/4] [y2038] lib: cobalt: clock: dispatch clock_settime

[chensong]

If sizeof time_t bigger than 4, which means glibc supports
64bit timespec, and __xn_feat_time64 is available in kernel,
go to clock_settime64.

otherwise, go to original clock_settime.

Signed-off-by: chensong <chensong@tj.kylinos.cn>
---
 lib/cobalt/clock.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/lib/cobalt/clock.c b/lib/cobalt/clock.c
index 8f06320..843ecd4 100644
--- a/lib/cobalt/clock.c
+++ b/lib/cobalt/clock.c
@@ -276,7 +276,13 @@ COBALT_IMPL(int, clock_settime, (clockid_t clock_id, const struct timespec *tp))
 	if (clock_id == CLOCK_REALTIME && !cobalt_use_legacy_tsc())
 		return __STD(clock_settime(CLOCK_REALTIME, tp));
 
+#ifdef __USE_TIME_BITS64
+	ret = -XENOMAI_SYSCALL2((cobalt_features_available(__xn_feat_time64)) ?
+			 sc_cobalt_clock_settime64 : sc_cobalt_clock_settime,
+			 clock_id, tp);
+#else
 	ret = -XENOMAI_SYSCALL2(sc_cobalt_clock_settime, clock_id, tp);
+#endif
 	if (ret) {
 		errno = ret;
 		return -1;
-- 
2.7.4

Re: [PATCH 4/4] [y2038] lib: cobalt: clock: dispatch clock_settime

[florian.bezdeka]

On Thu, 2021-03-11 at 10:47 +0800, chensong wrote:
> If sizeof time_t bigger than 4, which means glibc supports
> 64bit timespec, and __xn_feat_time64 is available in kernel,
> go to clock_settime64.
> 
> otherwise, go to original clock_settime.
> 
> Signed-off-by: chensong <chensong@tj.kylinos.cn>
> ---
>  lib/cobalt/clock.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/lib/cobalt/clock.c b/lib/cobalt/clock.c
> index 8f06320..843ecd4 100644
> --- a/lib/cobalt/clock.c
> +++ b/lib/cobalt/clock.c
> @@ -276,7 +276,13 @@ COBALT_IMPL(int, clock_settime, (clockid_t clock_id, const struct timespec *tp))
>  	if (clock_id == CLOCK_REALTIME && !cobalt_use_legacy_tsc())
>  		return __STD(clock_settime(CLOCK_REALTIME, tp));
>  
> +#ifdef __USE_TIME_BITS64
> +	ret = -XENOMAI_SYSCALL2((cobalt_features_available(__xn_feat_time64)) ?
> +			 sc_cobalt_clock_settime64 : sc_cobalt_clock_settime,
> +			 clock_id, tp);

We don't have a __xn_feat_time64 feature flag. If the syscall is not
available we get ENOSYS which tells the caller that the feature is not
available.

Falling back to sc_cobalt_clock_settime is a bug anyways.

> +#else
>  	ret = -XENOMAI_SYSCALL2(sc_cobalt_clock_settime, clock_id, tp);
> +#endif
>  	if (ret) {
>  		errno = ret;
>  		return -1;