Re: buffer cache patches - anybody willing to summarize?

Michael H. Warfield (
Sun, 3 Aug 1997 15:59:09 -0400 (EDT)

Linus, etc al...

> As David Miller got fed up with the bickering on pre-2.0.31, I'm in the
> unenviable position of having to do it myself. I have all of the patches
> that David has gotten over time, but quite frankly I don't consider 2.0.x
> to be all that interesting, and for me 2.0.31 only makes sense as a
> stability version that fixes the worst problems of 2.0.x rather that
> trying to fix all the small nagging things..

I think that all of the jerks who had to give David a hard time
all own him a major appology. That, unfortunately, assumes they have the
intelligence to recognize what they did. They could not have screwed the
2.0.31 release up worse even if they were Micro$oft employees. Anyone
check those E-Mail addresses? According to Computer World, Micro$oft
is getting pretty belligerent lately with refusing to ship Java Classes
and such. Maybe they've decided to send in a few aggitants to disrupt
the Linux development process. Maybe I'm being paranoid, but then again,
if someone wanted to tear apart the Linux development effort, I can't think
of a better way to do it!

> In short, I would like to have all the final versions, and I'd like to
> have _only_ the buffer-cache related stuff. You can pretty much assume
> that I'm working off a clean 2.0.30 - ignore any pre-patches by David.

David was looking at newer drivers for the 3c900 serial ethernet
cards as well. Due to some dust-up over some recent drivers, it was a
debate as to whether the 0.30-all driver with some patches went into the
kernel OR the latest driver (currently 0.42 and reasonably stable) went
it. The 0.40 version of drivers/net/3c59x.c was radioactive. The 0.41 was
better, especially with a few patches. The 0.42 driver incorporates those
patches, but Donald Becker has reported that he is still working at getting
down to the root cause of some stray problems. In this religious war, I'm
very much on the 0.42 side of things. I've got it in numerous systems and
it's been very stable. The 0.30 driver does not provide the performance on
100MB links. As far as I can determine amongst the dozens of Linux systems
I've got at home and at the office, 0.42 seems to be as stable as 0.30-all
and is a better performer to boot.

> ALSO! If people made other, non-buffer-cache related patches and sent them
> to David, please consider them lost. I do have Davids patches but I want
> to look over the separate patches rather than the one humongous patch I
> have now. Again, please summarize exactly why you think the patch is
> needed: and you can pretty much assume that I'm unwilling to apply any
> patches that aren't obviously _really_ needed.

> NOTE NOTE NOTE! The above sounds like I want others to do as much work for
> 2.0.31 as possible, and that's exactly correct. I hope that nobody is
> offended by having to re-do work they already did, but think of what I
> have to go through.. Most of the patch-creators at least have to re-do
> only one patch (their own), while I will have to go through a lot of
> patches from many sources. That is why I want to try to get others to do
> as much as possible before I then distill the final 2.0.31.

Sounds like a manager to me (and a good one at that). :-)

> This is what people get for being difficult to David. I may be a lot more
> thick-skinned than David, but because of that I'm also a lot more bitchy.

WAY TO GO! That's exactly the way I have to be at the office with
about 4 dozen engineers in the department. That's the only way to survive
with your sanity intact. Senior Engineer is a job position with all the
responsibilities of a monitor at a kindegarden recess!

The people who bitched at David got exactly what they deserved,
longer delays in the release cycle. And since none of them have ponied
up to help out, I have no sympathy. Unfortunately, the rest of us got
caught up as collateral damage in their excerise in optical rectitus. I
still suspect that some of it was deliberate antagonism by subversives, but
I'm paranoid anyways. Remember what Heinlein said - "Even paranoids have

> List of patches currently applied to my tree:
> - axp math bits by Richard Henderson
> - init_tty patch by Bill Hawes

> That's it,

Plus the 3c59x.c patches... Pretty please... :-) I WILL help out
there anyway I can!

> Linus


 Michael H. Warfield    |  (770) 985-6132   |
  (The Mad Wizard)      |  (770) 925-8248   |
  NIC whois:  MHW9      |  An optimist believes we live in the best of all
 PGP Key: 0xDF1DD471    |  possible worlds.  A pessimist is sure of it!