Date: Sat, 12 Jan 2008 19:40:05 +0000 From: John Birrell <jb@what-creek.com> To: Robert Watson <rwatson@FreeBSD.org> Cc: Perforce Change Reviews <perforce@freebsd.org> Subject: Re: PERFORCE change 133090 for review Message-ID: <20080112194005.GA47058@what-creek.com> In-Reply-To: <20080112123333.B36731@fledge.watson.org> References: <200801120500.m0C50viG006775@repoman.freebsd.org> <20080112123333.B36731@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Jan 12, 2008 at 12:35:33PM +0000, Robert Watson wrote: > On Sat, 12 Jan 2008, John Birrell wrote: > > >http://perforce.freebsd.org/chv.cgi?CH=133090 > > > >Change 133090 by jb@jb_freebsd1 on 2008/01/12 05:00:46 > > > > Add a simple libproc for use by libdtrace. This is based on > > ptrace, kevent and kvm instead of procfs like the one in Solaris. > > Does it actually depend on kvm (/dev/kmem) or just on libkvm interfaces? > We've been trying to eliminate dependence on /dev/kmem for tools that > otherwise don't require privilege over the last few years, in the interests > of eliminating setgid kmem. Generally, we've done this by adding sysctls > that explicitly export the required information, and in some cases, those > sysctls have been wrapped by existing libkvm interfaces. I guess this is > most easily answered by knowing if things using libdtrace->libproc require > setgid to operate :-). I started using the process status via kvm, but found that it was better obtained as a kevent, so the code at the moment doesn't do anything via kvm. I can probably complete the functionality via the ptrace() call and remove the libkvm stuff altogether. -- John Birrell
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20080112194005.GA47058>