From: "Lorenz (xha)" <me@xha.li>
Subject: Re: sysctl(2): make sysctl_int{,_lower}() mp-safe and unlock KERN_HOSTID
To: Vitaliy Makkoveev <mvs@openbsd.org>
Cc: tech@openbsd.org
Date: Wed, 14 Aug 2024 06:58:37 +0200

hi,

On Tue, Aug 13, 2024 at 02:24:34PM +0300, Vitaliy Makkoveev wrote:
> Make sysctl_int() mp-safe in the sysctl_int_bounded() way and unlock
> KERN_HOSTID. Except bootstrap called clockattach() of sparc64, `hostid'
> never used outside kern_sysctl().
> 
> mp-safe sysctl_int() is meaningless for sysctl_int_lower(), so rework it
> too. This time all affected paths are kernel locked, but this doesn't
> make sysctl_int_lower() worse.
> 
> ok?
> 
> Index: sys/kern/kern_sysctl.c
> ===================================================================
> RCS file: /cvs/src/sys/kern/kern_sysctl.c,v
> diff -u -p -r1.437 kern_sysctl.c
> --- sys/kern/kern_sysctl.c	11 Aug 2024 15:10:53 -0000	1.437
> +++ sys/kern/kern_sysctl.c	13 Aug 2024 11:03:24 -0000
> @@ -507,6 +507,10 @@ kern_sysctl(int *name, u_int namelen, vo
>  		return (sysctl_rdstring(oldp, oldlenp, newp, version));
>  	case KERN_NUMVNODES:  /* XXX numvnodes is a long */
>  		return (sysctl_rdint(oldp, oldlenp, newp, numvnodes));
> +	case KERN_HOSTID:
> +		/* XXX assumes sizeof long >= sizeof int */

the C standard specifies that this is the case, for example in C99:

6.2.5 Types: "For any two integer types with the same signedness
and different integer conversion rank (see 6.3.2.1), the range of
values of the type with smaller integer conversion rank is a subrange
of the values of the other type."