2002-07-02 21:29:29

by Rob Landley

[permalink] [raw]
Subject: O(1) scheduler vs 2.4.19-rc1 (question).

I'm finally getting around to playing with the O(1) scheduler (well I found a
way to break something that this might help), and I'm a bit confused as to
what to apply to get the newest version running on 2.4.19-rc1:

In http://people.redhat.com/mingo/O(1)-scheduler/ there live a number of
files, the most interesting of which for my purposes appear to be:

http://people.redhat.com/mingo/O(1)-scheduler/sched-O1-2.4.18-pre8-K3.patch
http://people.redhat.com/mingo/O(1)-scheduler/sched-2.4.19-pre10-ac2-A4
http://people.redhat.com/mingo/O(1)-scheduler/sched-2.4.19-pre10-ac2-B3

Questions:

1) The 2.4.18-pre8 patch is from February 7th. Is that really the latest one
for straight 2.4? If nobody's found even a typo in the thing for almost five
months, can we expect it in 2.4.20-pre1?

2) Do the -ac patches bring the 2.4-mt O(1) up to the level that's in the -ac
tree, or are they against the -ac tree itself? I'd happily run the -ac tree
except it doesn't HAVE stable releases, it has "it compiled" releases which
do tend to be fairly stable but don't have nice clustering points where
enough people are running that particular variant that they can tell you what
the inevitable bugs actually are...

3) Is any of the stuff in ingo's directory actually the latest version? I
know he wrote it, but I've watched about five other people patch it (Robert
Love, etc.), and I didn't keep close track at the time but I'm fairly certain
it was more recent than February.

4) What's with the version numbers? I don't THINK the "B3" patch backlevels
K3 in a more recent -ac version, especially since "B3" is dated july and "K3"
is dated february... I seem to have missed a curve somewhere...

5) Huh?

Thanks,

Rob


2002-07-02 22:02:51

by Robert Love

[permalink] [raw]
Subject: Re: O(1) scheduler vs 2.4.19-rc1 (question).

On Tue, 2002-07-02 at 08:33, Rob Landley wrote:

> 1) The 2.4.18-pre8 patch is from February 7th. Is that really the latest one
> for straight 2.4? If nobody's found even a typo in the thing for almost five
> months, can we expect it in 2.4.20-pre1?

That is the latest patch Ingo has. I have been generating 2.4 patches
now and the latest (for 2.4.19-rc1) is at:

ftp://ftp.kernel.org/pub/linux/kernel/people/rml/sched/ingo-O1/sched-O1-rml-2.4.19-rc1-1.patch

People have found bugs and there is still active development. The last
patch being against 2.4.18-pre8 should not mean otherwise. Neither Ingo
nor I plan to pursue the O(1) scheduler in 2.4, especially not starting
with 2.4.20-pre.

Enjoy it in 2.5 or get 2.4-ac or the above patch.

> 2) Do the -ac patches bring the 2.4-mt O(1) up to the level that's in the -ac
> tree, or are they against the -ac tree itself? I'd happily run the -ac tree
> except it doesn't HAVE stable releases, it has "it compiled" releases which
> do tend to be fairly stable but don't have nice clustering points where
> enough people are running that particular variant that they can tell you what
> the inevitable bugs actually are...

I do not know if I agree with your view of 2.4-ac, but nonetheless I
have kept the scheduler in there fairly up-to-date. The patches above
of Ingo's will bring 2.4.19-pre10-ac2 up-to-date with the very latest
code.

> 3) Is any of the stuff in ingo's directory actually the latest version? I
> know he wrote it, but I've watched about five other people patch it (Robert
> Love, etc.), and I didn't keep close track at the time but I'm fairly certain
> it was more recent than February.

The 2.4-ac stuff is recent, obviously the 2.4.18-pre8 thing is a bit out
of date.

Ingo is still active in development, he just does not have the latest
2.4 stock patch. See mine above.

> 4) What's with the version numbers? I don't THINK the "B3" patch backlevels
> K3 in a more recent -ac version, especially since "B3" is dated july and "K3"
> is dated february... I seem to have missed a curve somewhere...
>
> 5) Huh?

Eh?

Robert Love

2002-07-02 23:11:51

by crimsun

[permalink] [raw]
Subject: Re: O(1) scheduler vs 2.4.19-rc1 (question).

On Tue, Jul 02, 2002 at 11:33:36AM -0400, Rob Landley wrote:
> I'm finally getting around to playing with the O(1) scheduler (well I found a
> way to break something that this might help), and I'm a bit confused as to
> what to apply to get the newest version running on 2.4.19-rc1:

The last one you list below is the latest.

> http://people.redhat.com/mingo/O(1)-scheduler/sched-O1-2.4.18-pre8-K3.patch
> http://people.redhat.com/mingo/O(1)-scheduler/sched-2.4.19-pre10-ac2-A4
> http://people.redhat.com/mingo/O(1)-scheduler/sched-2.4.19-pre10-ac2-B3
^^^^^^^^^^^^^^^^^^^^^^^^^

> 1) The 2.4.18-pre8 patch is from February 7th. Is that really the latest one
> for straight 2.4? If nobody's found even a typo in the thing for almost five
> months, can we expect it in 2.4.20-pre1?

While it's certainly nice, I think Ingo and Robert both stated best that
it's best not to change this in vanilla 2.4 where accountable behavior
should remain for a stable-branch kernel.

> 2) Do the -ac patches bring the 2.4-mt O(1) up to the level that's in the -ac
> tree, or are they against the -ac tree itself? I'd happily run the -ac tree
> except it doesn't HAVE stable releases, it has "it compiled" releases which
> do tend to be fairly stable but don't have nice clustering points where
> enough people are running that particular variant that they can tell you what
> the inevitable bugs actually are...

Against the -ac tree itself as stated in the patch name. For the record,
2.4.19-pre10-ac2 has been solid here in various incarnations for nearly
a month.

> 3) Is any of the stuff in ingo's directory actually the latest version? I
> know he wrote it, but I've watched about five other people patch it (Robert
> Love, etc.), and I didn't keep close track at the time but I'm fairly certain
> it was more recent than February.

See above (and below). :)

> 4) What's with the version numbers? I don't THINK the "B3" patch backlevels
> K3 in a more recent -ac version, especially since "B3" is dated july and "K3"
> is dated february... I seem to have missed a curve somewhere...

Ingo's mail dated 01 July 11:49:39 +0200 (CEST) has
sched-2.4.19-pre10-ac2-B3 as the latest.

--
Dan Chen [email protected]
GPG key: http://www.sh.nu/~crimsun/pubkey.gpg.asc