Eh, this is probably wrong, or the wrong place:
this function filters the clocks above which
a timer can be set ...


On Jun 26 15:15:48, [email protected] wrote:
> Hi,
> 
> the linux clock compatibility code in
> sys/compat/linux/common/linux_time.c
> seems to be outdated:
> 
> - we do have CLOCK_PROCESS_CPUTIME_ID under that name,
>   so there is no need to map it
> - we do have CLOCK_THREAD_CPUTIME_ID, so recognize that
> - CLOCK_VIRTUAL is not CLOCK_PROCESS_CPUTIME_ID
> 
> diff below.
> 
> What are the bits that should be tested for this? That is to say,
> what would be a good example of a linux program that needs
> to run on NetBSD, using NetBSD's clocks as if on linux?
> 
>       Jan
> 
> 
> 
> 
> Index: sys/compat/linux/common/linux_time.c
> ===================================================================
> RCS file: /pub/NetBSD-CVS/src/sys/compat/linux/common/linux_time.c,v
> retrieving revision 1.42
> diff -u -p -r1.42 linux_time.c
> --- sys/compat/linux/common/linux_time.c      19 Sep 2021 23:51:37 -0000      
> 1.42
> +++ sys/compat/linux/common/linux_time.c      25 Jun 2026 13:40:59 -0000
> @@ -334,19 +334,13 @@ linux_to_native_timer_create_clockid(clo
>               /*
>                * We can't create a timer with every sort of clock ID
>                * that the system understands, so filter them out.
> -              *
> -              * Map CLOCK_PROCESS_CPUTIME_ID to CLOCK_VIRTUAL.
> -              * We can't handle CLOCK_THREAD_CPUTIME_ID.
>                */
>               switch (id) {
>               case CLOCK_REALTIME:
>               case CLOCK_MONOTONIC:
> -                     break;
> -
>               case CLOCK_PROCESS_CPUTIME_ID:
> -                     id = CLOCK_VIRTUAL;
> +             case CLOCK_THREAD_CPUTIME_ID:
>                       break;
> -
>               default:
>                       return ENOTSUP;
>               }
> 
> 

Reply via email to