Date: Mon, 28 Aug 1995 14:11:08 -0700 (MST) From: Terry Lambert <terry@Artisoft.COM> To: davidg@Root.COM Cc: julian@freefall.freebsd.org, hackers@freefall.freebsd.org, terry@freefall.freebsd.org Subject: Re: Terry's changes.. Message-ID: <199508282111.OAA21684@coyote.Artisoft.COM> In-Reply-To: <199508272337.QAA17571@corbin.Root.COM> from "David Greenman" at Aug 27, 95 04:37:46 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> I REALLY wish people would work on integrating the 4.4-lite2 changes before > going hog-wild. If these changes don't conflict with the lite-2 stuff, then by > all means proceed... The major FS conflicts with the Lite2 stuff are the existing initialization changes in vfs_conf.c and the changes already in place in the FreeBSD code but not in the Lite2 code for diskless support and for the new VM system. Not much else has changed, except in miscfs/* and ufs/lfs/*, both of which are pretty much untouched by my VFS patches. The unified diffs are pretty clean -- they are just some cosmetic stuff to make the code more readable in terms of what a compare againsts the symlinklength has to do with anything, plus some cleanup to make the code less dependent on strucuture alignment and element size. The SYSINIT stuff is about as conflicting as the PSEUDO_SET stuff for statically installed pseudo-devices. It trades the ability to linearly read down the init_main.c (which is unclear anyway because of the startinit return after fork returning to the calling assembly instead of calling an assembly routine at the ens of startinit) for the ability to linearly look down the header file kernel.h for gross initialization order. It also allows dropping in of binary modules, and because of the gapping, extension of the initialization tasks on an as-needed basis without header file or kernel mods. In fact, the use of the linker set code allows the modules to be included or omitted with no additional kernel changes. I think this is a fair trade for support of binary modules from third party vendors who want to do system initialization time tasks. It just means that the operation of the kernel startup should be documented -- this was a requirement before, and now it's just a slightly stronger requirement. #ifdef's all over the kernel should be removed anyway, they are warts from a lack of an adequate use of callback registration mechanisms. -- Terry
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199508282111.OAA21684>