Date: Mon, 23 Apr 2001 11:36:40 -0700 From: Marcel Moolenaar <marcel@cup.hp.com> To: Robert Watson <rwatson@FreeBSD.org> Cc: emulation@FreeBSD.org Subject: Re: Linux-specific jail code in linuxulator Message-ID: <20010423113640.C42858@gauss.cup.hp.com> In-Reply-To: <Pine.NEB.3.96L.1010422164814.63356E-100000@fledge.watson.org>; from rwatson@FreeBSD.org on Sun, Apr 22, 2001 at 04:54:25PM -0400 References: <Pine.NEB.3.96L.1010422164814.63356E-100000@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Apr 22, 2001 at 04:54:25PM -0400, Robert Watson wrote: > > CVS annotate on linux_mib.c indicates that these features were present in > 1.1 of the file when it was originally added, so I was wondering if (a) > you were the author of the code and Yes, I was. > (b) what you knew about its use. I discussed this with phk at the time. The reason for this IIRC is that you could run a Linux binary in a jail, have these parameters tuned without affecting other Linux binaries. For example, some people objected I changed the name of the OS as returned by uname(2) from FreeBSD to Linux. Those were Netscape users, BTW. It is now possible to have netscape run in a jail and have the OS set to FreeBSD for those caring about web statistics and still have Linux returned in the normal cases. > Since > I'm rewriting largely from scratch (although keeping fairly close to the > original implementation when it comes to most features), now appears to be > the opportunity to determine if these features are used, if so whether > they are useful. From a security perspective they might have some value, but mostly to have the kld tuned for different Linux binaries without affecting each other. > Apparently they weren't part of Poul-Henning's orginal > implementation, so I assume they were added later. Correct. > If they are used, I > should make sure to include them in the revised version, and possibly > clean up interactions between optional components (such as sysvipc, > linuxlator, etc) and jail. If they're not used, removing them makes sense > because they do introduce complexity (especially in light of fine-grained > threading/protection in the kernel). I see. It's hard for me to say what the best action is here. I don't use the feature myself and I don't have any indication of people who do depend on this. So if people could speak up here. Maybe this is something for -arch as well. I think we should keep it if possible, if only to maintain current behaviour. But if that's too complicated, as you say, then I wouldn't have a problem with it being changed. -- Marcel Moolenaar mail: marcel@cup.hp.com / marcel@FreeBSD.org tel: (408) 447-4222 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-emulation" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010423113640.C42858>