On Sun, 2004-09-12 at 22:25, Russell Hires wrote:
> Debian-sarge. Home built kernel.org generic tree 2.4.27 on powerpc.
Really? You sure XFS is in there?
BTW, SGI only supported XFS on kernel 2.4 on x86 and IA-64.
> Yeah, I figured it was okay after it finally got officially included in
> the later 2.4.x kernels. But I guess not.
XFS was _never_ included in 2.4.x kernels. It is _only_ included in
2.6. There is too much added kernel support required for 2.4, hence why
it was _not_ added until 2.6.
> That's one of those "I wish they would practice what they preach"
> situations. Somebody should have cared about it.
SGI very much did. That's why they _did_ release their own XFS
distributions that ran _atop_ of Red Hat Linux (starting with release
1.0 for Red Hat Linux 7.1 in 2001 April). Many of us _were_ running XFS
on kernel 2.4 -- but _only_ on Red Hat Linux.
Dude, understand that there is a _lot_ that was missing in kernel 2.4.
It took a _lot_ of VFS, VM and scheduler integration in the 2.5 series.
Luckily, SGI did a lot, and _all_ filesystems are benefiting from those
GPL donations. SGI basically donated the entire POSIX EA/ACL
implementation, and made the few changes Linus wanted (as of 2.5.3 when
Linus standardized the interfaces).
> And it seems to be so much detail that no one has a good guide for it.
> Or wants to take some time to explain it.
SGI _fully_ detailed the changes. They maintained their own fork of the
kernel, and eventually built a dual-patchset -- one of the kernel
changes, one for the VFS XFS support. Anyone was free to patch it in,
but the changes for just the former were _massive_. SGI and Linus
totally saw eye-to-eye -- the changes _were_ good for the Linux kernel,
but they were too big for the stock 2.4 kernel -- and were left to 2.6.
These subsystems had _nothing_ to do with XFS itself, but were required
for a lot of the XFS features. Unlike JFS, which came from OS/2 (not
AIX) and totally lacked a lot of UNIX interfaces, and Ext3 and ReiserFS
that were built on and for Linux, XFS came over _verbatim_ from Irix.
So SGI had to add all those features into the Linux kernel. They did
so. Linux has now benefited.
For those that absolutely felt they "must" run XFS, then SGI offered a
well supported release for Red Hat Linux. The changes were just too
massive. Mandrake's kernel releases were totally worthless, and they
didn't even include the correct off-line utilities on several occasions.
Never tried Debian. But SGI repeatedly said that they would only stand
by their release for 2.4 on Red Hat Linux. There are lots of reasons
for this, but understand that attempting to track all the patches on any
additional distro's kernel would have doubled the workload.
> I have other problems with 2.6, mainly the voodoo3 card. I think I may
> have found someone on the debian boot list that is willing to help me
> figure that one out.
X11 drivers shouldn't depend on the kernel, except for AGPGART support.
_No_ 3dfx chip _ever_ supported Direct-in-Memory Execution (DiME). But
for some unknown reason, the UtahGLX stuff requires it. That's a
Windows attitude IMHO.
But as far as 3dfx, are you just talking framebuffer? For 2D, you
shouldn't need anything from the kernel, unless you are doing
framebuffer.
-- Bryan J. Smith b.j.smith@ieee.org ------------------------------------------------------------------ "Communities don't have rights. Only individuals in the community have rights. ... That idea of community rights is firmly rooted in the 'Communist Manifesto.'" -- Michael Badnarik----------------------------------------------------------------------- This list is provided as an unmoderated internet service by Networked Knowledge Systems (NKS). Views and opinions expressed in messages posted are those of the author and do not necessarily reflect the official policy or position of NKS or any of its employees.
This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 17:32:03 EDT