Date: Mon, 28 Dec 1998 16:53:10 +0800 From: Peter Wemm <peter@netplex.com.au> To: mjacob@feral.com Cc: John Birrell <jb@cimlogic.com.au>, Mike Smith <mike@smith.net.au>, committers@FreeBSD.ORG Subject: Re: The recent fracas involving danes, war axes and wounded developers Message-ID: <199812280853.QAA63951@spinner.netplex.com.au> In-Reply-To: Your message of "Mon, 28 Dec 1998 00:12:27 PST." <Pine.LNX.4.04.9812280011020.23283-100000@feral-gw>
next in thread | previous in thread | raw e-mail | index | archive | help
Matthew Jacob wrote: > > > All we need to remind people is that a change that is likely > > to affect people needs a "HEADS UP" message. > > I agree- but with the proviso that a 'reasonable' amount of time so that > someone can say "but, wait....!"- and then discuss it offline as needed. Given the events of the past day, it's clear that this is essential. It seems to me (and quite a few others I might add) that if some advance warning was given, it would have become very clear quite quickly that Voxware and pcvt were going to be a major drama and people are NOT ready to let them go quietly. I've been told by a number of people (including within core) that these two at least should come back, especially since those two were explicitly NOT deleted with -core approval. However, re: LKM's (bear in mind that I have an interest here): 1: It's been pretty obvious for a while that LKM's are on the way out. However, I doubt that deleting the build code without warning is the best way of achieving it. :-/ 2: Deleting the source to vinum has already been acknowledged as being an unintended accident. This is already being fixed up as we speak. 3: KLD's have been installed and used in preference to LKM's for quite a while now. Many people are using KLD's instead of LKM's *already* without being aware of it - a.out kernels included. 4: The LKM loading mechanism has been within an inch of being broken beyond repair a number of times over the last few months and it's sheer luck that the mechanism has remained working. 5: Leaving the LKM loader active is not a good option if we're not building LKM's.. It's inviting disaster if somebody does a manual modload with a stale /lkm. There are interface changes in the pipeline (hopefully) for kernel-assisted thread support that will make all existing LKM's *incompatable* and require recompiling anyway. 6: Converting to KLD from LKM's is pretty easy in general, but there isn't any exact recipe. Basically it invoves deleting the LKM glue and installing a modevent driver instead, based on the examples floating around in the tree. 7: We're coming up to 3.0 branching time and we need to get the LKM/KLD issue sorted out preferably before then. I for one want to make the kernel LKM support an option that is not in GENERIC. (or even, to make the point better, make a KLD module that implements LKM loading - now that would have me chuckling for weeks. :-) ) As far as I'm aware, the only reason to keep LKM support is OSS - and I understand that the 3.0 version (is there one?) doesn't work too well on the 3.x kernel. If we get the rfork/sigshare/etc stuff through, the 3.0-RELEASE LKM won't work on 3.0-current or 3.0.1 anyway. Incidently, I believe the OSS folks could convert the 3.0 version to KLD quite easily, and it would be in there interests to do so since the one KLD works on both a.out and ELF kernels. Cheers, -Peter To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199812280853.QAA63951>