Re: [Minios-devel] [UNIKRAFT PATCH v5 9/9] plat/linuxu: Correct timeout calculation for ukplat_lcpu_halt_to()

[Sharan Santhanam <sharan.santhanam@xxxxxxxxx>]

Hello,

This patch looks fine.

On 08/31/2018 08:57 AM, Florian Schmidt wrote:
> From: Simon Kuenzer <simon.kuenzer@xxxxxxxxx>
>
> time_block_until() in plat/linuxu/lcpu.c was passing the
> until value directly as timeout value for the underlying
> pselect6 system call. This fix substracts the current
> value from the monotonic clock to retrieve the actual
> timeout value.
>
> Signed-off-by: Simon Kuenzer <simon.kuenzer@xxxxxxxxx>
> ---
>   plat/linuxu/lcpu.c | 5 +++++
>   1 file changed, 5 insertions(+)
>
> diff --git a/plat/linuxu/lcpu.c b/plat/linuxu/lcpu.c
> index 8e9ae35..316351a 100644
> --- a/plat/linuxu/lcpu.c
> +++ b/plat/linuxu/lcpu.c
> @@ -61,7 +61,12 @@ void halt(void)
Not related to this patch but why do we use a signed nanosecond as 
against an unsigned nanosecond.
>   void time_block_until(__snsec until)
>   {
>         struct k_timespec timeout;
> +       __nsec now = ukplat_monotonic_clock();
>   
> +	if (until < 0 || (__nsec) until < now)
> +               return; /* timeout expired already */
> +
> +       until -= now;
>         timeout.tv_sec  = until / ukarch_time_sec_to_nsec(1);
>         timeout.tv_nsec = until % ukarch_time_sec_to_nsec(1);
>   
Reviewed-by: Sharan Santhanam <sharan.santhanam@xxxxxxxxx>

Thanks & Regards
Sharan

_______________________________________________
Minios-devel mailing list
Minios-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/minios-devel