Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 5 Oct 2003 14:44:50 +0100
From:      Bruce M Simpson <bms@spc.org>
To:        Don Lewis <truckman@FreeBSD.org>
Cc:        cvs-all@FreeBSD.org
Subject:   Re: cvs commit: src/sys/kern kern_sysctl.c
Message-ID:  <20031005134450.GC13164@saboteur.dek.spc.org>
In-Reply-To: <200310051226.h95CQJN1049247@gw.catspoiler.org>
References:  <200310050937.h959bldI091908@repoman.freebsd.org> <200310051226.h95CQJN1049247@gw.catspoiler.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Oct 05, 2003 at 05:26:19AM -0700, Don Lewis wrote:
> In the SMP case the data can change even without pre-emption.  There
> have been a number of discussions (arch@, smp@, arch-handbook, etc.)
> about adding a mutex parameter to the sysctl API.  Someone even
> submitted a PR with a patch a few months ago (kern/54439), which I had
> hoped to review but never found the time to.

My GENERIC kernel with vslock() et al. reintroduced, and the pre-emption
check in sysctl_handle_opaque(), appears to be OK.

I am confident the security issue has now been addressed in -CURRENT
(it was limited to sysctl_handle_opaque()), but we now have the larger
problem of how to deal with procedural sysctl() handlers in the wider kernel.

I can see Peter has encouraged me to open a huge can of worms. Let's
continue discussion about what to do on -arch.

This has been a learning experience...

BMS



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20031005134450.GC13164>