* i686 requirements
@ 2005-01-07 12:34 Antonio Vargas
2005-01-07 14:36 ` Mark Williamson
` (2 more replies)
0 siblings, 3 replies; 4+ messages in thread
From: Antonio Vargas @ 2005-01-07 12:34 UTC (permalink / raw)
To: xen-devel
Regarding XEN requirements to run, I'd like to know if there is any
important reason to require i686-class cpus (ie: ppro, p2, etc.. and
k7...). In my limited understanding of 386 arch, ppro added
conditional moves and PAE, perhaps even huge-pages (have to check this
one). Is XEN explicitely (as in, used directly on assembler code)
using any of these?
My motivation is getting XEN to run on my (aging) k6 server machine.
Greets anyways :)
--
Greetz, Antonio Vargas aka winden of network
http://wind.codepixel.com/
Las cosas no son lo que parecen, excepto cuando parecen lo que si son.
-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: i686 requirements
2005-01-07 12:34 i686 requirements Antonio Vargas
@ 2005-01-07 14:36 ` Mark Williamson
2005-01-07 14:42 ` Ian Pratt
2005-01-07 18:03 ` Adam Sulmicki
2 siblings, 0 replies; 4+ messages in thread
From: Mark Williamson @ 2005-01-07 14:36 UTC (permalink / raw)
To: xen-devel, Antonio Vargas
Off the top of my head, 3 reasons that I know of:
1) Xen requires a local APIC for its timer functionality. There is some
compatibility stuff for non local-APIC chips to enable running VMWare but it
isn't often used AFAIK.
2) I'm not sure Xen includes all of the skanky fixups for the implementation
bugs in various older chips anymore.
3) Xen needs an rdtsc instruction and not all older chips have it (although I
believe there was an implementation in Intel's Pentium not all manufacturers
included it immediately)
I don't know how much work would be involved in running modern versions of Xen
on these older machines. I was going to try this a while back but never got
round to it and things have moved on since then...
HTH,
Mark
On Friday 07 January 2005 12:34, Antonio Vargas wrote:
> Regarding XEN requirements to run, I'd like to know if there is any
> important reason to require i686-class cpus (ie: ppro, p2, etc.. and
> k7...). In my limited understanding of 386 arch, ppro added
> conditional moves and PAE, perhaps even huge-pages (have to check this
> one). Is XEN explicitely (as in, used directly on assembler code)
> using any of these?
>
> My motivation is getting XEN to run on my (aging) k6 server machine.
>
> Greets anyways :)
-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: i686 requirements
2005-01-07 12:34 i686 requirements Antonio Vargas
2005-01-07 14:36 ` Mark Williamson
@ 2005-01-07 14:42 ` Ian Pratt
2005-01-07 18:03 ` Adam Sulmicki
2 siblings, 0 replies; 4+ messages in thread
From: Ian Pratt @ 2005-01-07 14:42 UTC (permalink / raw)
To: Antonio Vargas; +Cc: xen-devel, Ian.Pratt
> Regarding XEN requirements to run, I'd like to know if there is any
> important reason to require i686-class cpus (ie: ppro, p2, etc.. and
> k7...). In my limited understanding of 386 arch, ppro added
> conditional moves and PAE, perhaps even huge-pages (have to check this
> one). Is XEN explicitely (as in, used directly on assembler code)
> using any of these?
The current Xen code needs 4MB pages (PSE) support.
There was a patch posted a while back to avoid this constraint,
but I don't think it was fully working. We'd be very happy to
incorporate a patch that adds support for CPUs without PSE,
providing it didn't impact the normal code too much (i.e. didn't
make it less readable).
> My motivation is getting XEN to run on my (aging) k6 server machine.
Chopping a tiny pie into quarters doesn't make it a bigger pie
:-)
Ian
-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: i686 requirements
2005-01-07 12:34 i686 requirements Antonio Vargas
2005-01-07 14:36 ` Mark Williamson
2005-01-07 14:42 ` Ian Pratt
@ 2005-01-07 18:03 ` Adam Sulmicki
2 siblings, 0 replies; 4+ messages in thread
From: Adam Sulmicki @ 2005-01-07 18:03 UTC (permalink / raw)
To: Antonio Vargas; +Cc: xen-devel
> Regarding XEN requirements to run, I'd like to know if there is any
> important reason to require i686-class cpus (ie: ppro, p2, etc.. and
> k7...). In my limited understanding of 386 arch, ppro added
> conditional moves and PAE, perhaps even huge-pages (have to check this
> one). Is XEN explicitely (as in, used directly on assembler code)
> using any of these?
XEN is smart enough not to attempt to use PAE if it not supported.
As far as CMOVE and PSE goes, This my patch posted a while ago will fix
it. You can find it on web in varous xen archives, for example
http://article.gmane.org/gmane.comp.emulators.xen.devel/4062
http://sourceforge.net/mailarchive/message.php?msg_id=10258941
it is important to note that it is against xen 2.0.1 so you must use
bk clone -rRELEASE-2.0.1 bk://xen.bkbits.net/xen-2.0.bk xen-2.0.1.bk
instead of as
bk clone bk://xen.bkbits.net/xen-2.0.bk
as mentioned in the above said email.
It has worked for me and Rupert Schlick, but other than that it had fairly
limited testing. So the more ppl test it out the better.
I hope to port it to xen 2.0.2 and xen 2.0-testing eventually.
> My motivation is getting XEN to run on my (aging) k6 server machine.
I'm not that familar with specs of K6 so there migt be other gotchas, but
I suppose it is worth of a try.
-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2005-01-07 18:03 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2005-01-07 12:34 i686 requirements Antonio Vargas
2005-01-07 14:36 ` Mark Williamson
2005-01-07 14:42 ` Ian Pratt
2005-01-07 18:03 ` Adam Sulmicki
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.