2006-03-15 03:05:15

by John Richard Moser

[permalink] [raw]
Subject: ORMAP

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

And looking through the recent discussions I see in one thread...

Willy Tarreau wrote:

> It depends a lot on what people do with it in fact. For instance, it
> works better in memory-constrained systems, probably thanks to rmap.
> I have one 2.6 running reliably on my web server (hppa) where 2.4
> regularly oopsed because of low memory.


This reminds me, what the hell ever happened to ORMAP? That object
based rmap thingy I tried out in one of wli's patches made my system
boot like 3 times faster. There were other cool things going on that I
never got to try too, never saw that all out to fruition.

Status on some of the elements in the old 2.6-wli series from around
there would be nice. I'm curious as to what has gone in.

- --
All content of all messages exchanged herein are left in the
Public Domain, unless otherwise explicitly stated.

Creative brains are a valuable, limited resource. They shouldn't be
wasted on re-inventing the wheel when there are so many fascinating
new problems waiting out there.
-- Eric Steven Raymond

We will enslave their women, eat their children and rape their
cattle!
-- Evil alien overlord from Blasto
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iQIVAwUBRBeEKAs1xW0HCTEFAQJoiQ/+OPwYFP74F9Ysmbb9aLgUOxCo8gEfbK35
iHHVjNYDqGpoPm/yVy9tf/ilbtrbhk8NCP6q19AN7Z+8gpV8l5ITJBgJFdrtuRYv
aIqEwsqySk/mNkq7u91OnOpXFJVVlUsbALUPDistPUAGCkzcWp4M8qRmpdfbUBEO
TiZp/oWjI97nMD7bOSJf+YnzmVU9XBYQZ7mMSpX9JeP9skUjq1/kXj6hFg5w+vF/
73xZtjkntxQIvTW7mb7gQB3eNFUmY9A31ngY1Q3Maleu52FgSTNNOpb5rHUivmnl
C9ols4z9b7i+utUM8+4V527DO2JejfFEygQ8VvXLjx7luIM+dmgHYzWQaDrFkPB7
JWiCaizulRE35BEb7YVog3sfFDl0wG3/NnXGjfHdJAgALhL6JPOsLTZHvgfua7Ws
7hDDo9BvAYN5nU0kQ+EmLAAJWAQ/UL1TSKPAdsXfRGQHBrpEbyoVCfDw8D5T8r5X
ytgl+jo51WAd/dvnq+xUjnWYPauRxaRimy5xndRgEEIRCd4sxCEdmKMMxGIuRyhS
l+oPhGL+RN0wpgFHdStQ4llDF+7mu3Y3sEFrvXlIY95dmRvU5V4Y6kGTsCoChixT
5VYXWtmL2m95ywYHdyRLPJ35Z1oqVIs2s5f4YtsDLd6kpRK0B0s0EpJGQcwVvyFl
Mmnf+j31oTE=
=O+f2
-----END PGP SIGNATURE-----


2006-03-15 12:28:35

by Nick Piggin

[permalink] [raw]
Subject: Re: ORMAP

John Richard Moser wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> And looking through the recent discussions I see in one thread...
>
> Willy Tarreau wrote:
>
>
>>It depends a lot on what people do with it in fact. For instance, it
>>works better in memory-constrained systems, probably thanks to rmap.
>>I have one 2.6 running reliably on my web server (hppa) where 2.4
>>regularly oopsed because of low memory.
>
>
>
> This reminds me, what the hell ever happened to ORMAP? That object
> based rmap thingy I tried out in one of wli's patches made my system
> boot like 3 times faster. There were other cool things going on that I
> never got to try too, never saw that all out to fruition.
>
> Status on some of the elements in the old 2.6-wli series from around
> there would be nice. I'm curious as to what has gone in.
>

2.6 has an object based rmap system working nicely for quite
a while now (though it was probably not exactly what you saw
in the -wli tree, but a derivative).

It would be surprising if that made your system boot 3 times
faster though (unless it was on the edge of a swap storm or
something)

--
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com

2006-03-15 16:20:09

by John Richard Moser

[permalink] [raw]
Subject: Re: ORMAP

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1



Nick Piggin wrote:
> John Richard Moser wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> And looking through the recent discussions I see in one thread...
>>
>> Willy Tarreau wrote:
>>
>>
>>> It depends a lot on what people do with it in fact. For instance, it
>>> works better in memory-constrained systems, probably thanks to rmap.
>>> I have one 2.6 running reliably on my web server (hppa) where 2.4
>>> regularly oopsed because of low memory.
>>
>>
>>
>> This reminds me, what the hell ever happened to ORMAP? That object
>> based rmap thingy I tried out in one of wli's patches made my system
>> boot like 3 times faster. There were other cool things going on that I
>> never got to try too, never saw that all out to fruition.
>>
>> Status on some of the elements in the old 2.6-wli series from around
>> there would be nice. I'm curious as to what has gone in.
>>
>
> 2.6 has an object based rmap system working nicely for quite
> a while now (though it was probably not exactly what you saw
> in the -wli tree, but a derivative).
>
> It would be surprising if that made your system boot 3 times
> faster though (unless it was on the edge of a swap storm or
> something)

Dramatization. It was probably around 30 seconds faster on a 2-3 minute
boot sequence (I had a lot in rc.d), but it was noticeable :P

I was wondering about that stuff. There used to be a few cute things
out there but I can't remember any of it now. Page clustering etc etc.

>

- --
All content of all messages exchanged herein are left in the
Public Domain, unless otherwise explicitly stated.

Creative brains are a valuable, limited resource. They shouldn't be
wasted on re-inventing the wheel when there are so many fascinating
new problems waiting out there.
-- Eric Steven Raymond

We will enslave their women, eat their children and rape their
cattle!
-- Evil alien overlord from Blasto
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
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=k6aR
-----END PGP SIGNATURE-----

2006-03-15 23:31:58

by Nick Piggin

[permalink] [raw]
Subject: Re: ORMAP

John Richard Moser wrote:

>Nick Piggin wrote:
>
>>2.6 has an object based rmap system working nicely for quite
>>a while now (though it was probably not exactly what you saw
>>in the -wli tree, but a derivative).
>>
>>It would be surprising if that made your system boot 3 times
>>faster though (unless it was on the edge of a swap storm or
>>something)
>>
>
>Dramatization. It was probably around 30 seconds faster on a 2-3 minute
>boot sequence (I had a lot in rc.d), but it was noticeable :P
>
>I was wondering about that stuff. There used to be a few cute things
>out there but I can't remember any of it now. Page clustering etc etc.
>
>

Well I don't think any of that stuff was simply forgotten. Page
clustering for
i386, for example became less important because of objrmap, reductions
in size
of struct page, and the demise of insane highmem machines (due to x86-64).

Nick
--

Send instant messages to your online friends http://au.messenger.yahoo.com

2006-03-15 23:48:18

by John Richard Moser

[permalink] [raw]
Subject: Re: ORMAP

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1



Nick Piggin wrote:
> John Richard Moser wrote:
>
>> Nick Piggin wrote:
>>
>>> 2.6 has an object based rmap system working nicely for quite
>>> a while now (though it was probably not exactly what you saw
>>> in the -wli tree, but a derivative).
>>>
>>> It would be surprising if that made your system boot 3 times
>>> faster though (unless it was on the edge of a swap storm or
>>> something)
>>>
>>
>> Dramatization. It was probably around 30 seconds faster on a 2-3 minute
>> boot sequence (I had a lot in rc.d), but it was noticeable :P
>>
>> I was wondering about that stuff. There used to be a few cute things
>> out there but I can't remember any of it now. Page clustering etc etc.
>>
>>
>
> Well I don't think any of that stuff was simply forgotten. Page
> clustering for
> i386, for example became less important because of objrmap, reductions
> in size
> of struct page, and the demise of insane highmem machines (due to x86-64).
>

Cool cool. What juicy things are out there these days that will
probably make future kernels faster and more memory-light? New
schedulers? Ingo's SLAB rewrite and priority inheritance?

> Nick
> --
>
> Send instant messages to your online friends http://au.messenger.yahoo.com

- --
All content of all messages exchanged herein are left in the
Public Domain, unless otherwise explicitly stated.

Creative brains are a valuable, limited resource. They shouldn't be
wasted on re-inventing the wheel when there are so many fascinating
new problems waiting out there.
-- Eric Steven Raymond

We will enslave their women, eat their children and rape their
cattle!
-- Evil alien overlord from Blasto
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
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=jO4u
-----END PGP SIGNATURE-----