svn commit: r309658 - head/sys/vm

[Bruce Evans]

On Tue, 6 Dec 2016, Mark Johnston wrote:

> Log:
>  Provide dummy sysctls for v_cache_count and v_tcached.
>
>  Some utilities (notably top(1)) exit if any of their input sysctls don't
>  exist, and the removal of the above-mentioned PG_CACHE-related sysctls
>  makes it difficult to run such utilities on different versions of the
>  kernel without recompiling.
>
>  Requested by:	bde

Thanks.

> Modified: head/sys/vm/vm_meter.c
> ==============================================================================
> --- head/sys/vm/vm_meter.c	Tue Dec  6 22:48:28 2016	(r309657)
> +++ head/sys/vm/vm_meter.c	Tue Dec  6 22:52:45 2016	(r309658)
> @@ -314,3 +314,14 @@ VM_STATS_VM(v_forkpages, "VM pages affec
> VM_STATS_VM(v_vforkpages, "VM pages affected by vfork()");
> VM_STATS_VM(v_rforkpages, "VM pages affected by rfork()");
> VM_STATS_VM(v_kthreadpages, "VM pages affected by fork() by kernel");
> +
> +#ifndef BURN_BRIDGES
> +/*
> + * Provide compatibility sysctls for the benefit of old utilities which exit
> + * with an error if they cannot be found.
> + */
> +SYSCTL_UINT(_vm_stats_vm, OID_AUTO, v_cache_count, CTLFLAG_RD,
> +    (u_int *)NULL, 0, "Dummy for compatibility");
> +SYSCTL_UINT(_vm_stats_vm, OID_AUTO, v_tcached, CTLFLAG_RD,
> +    (u_int *)NULL, 0, "Dummy for compatibility");
> +#endif

I don't like casting NULL, but SYSCTL_* enforces type safety for pointers
to variables in a way that breaks normal use of NULL.

You are supposed to obfuscate this by spelling the value as SYSCTL_UINT_NULL.
This is actually documented in sysctl(9).

Bruc