Re: 5.0-RELEASE under VMWare 3 : slowdown and even hangup

[Garance A Drosihn]

At 10:51 AM +0100 2/13/03, [EMAIL PROTECTED] wrote:

I'm kinda in the same situation : XP Pro, WMWare 3.2, several
virtual machines running fine (linux,w2k), but the freebsd
5-RELEASE is quite a pain to use in this env.

The guest OS keeps on slowing down, to the point that it's
unusable.

Kris Kenna talked about a kernel option for fix this. The
problem would be related to a specific opcode emulation
done by VMWare.


If you look in the file:
   /usr/src/sys/i386/conf/NOTES

You will find an option called CPU_DISABLE_CMPXCHG

You need to compile a kernel which has this option when running
under VMWare.  You can not use this option with the SMP option,
but then you probably should not be using an SMP kernel when
running under VMWare!

I realize you then have the problem of how to compile the new
kernel when it takes so long to do anything with the standard
GENERIC kernel.  Perhaps it would go better if you booted up in
single-user mode, and then compiled and installed the new kernel.
However, that is just a guess on my part.  It would probably be
easier to get someone else to compile a 5.0-release kernel with
that CPU_DISABLE_CMPXCHG option.

--
Garance Alistair Drosehn=   [EMAIL PROTECTED]
Senior Systems Programmer   or  [EMAIL PROTECTED]
Rensselaer Polytechnic Instituteor  [EMAIL PROTECTED]

To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-questions in the body of the message

Re: 5.0-RELEASE under VMWare 3 : slowdown and even hangup

[[EMAIL PROTECTED]]

Hi there everyone

At 10:51 AM +0100 2/13/03, [EMAIL PROTECTED] wrote:
You will find an option called CPU_DISABLE_CMPXCHG

I read that one and didn't read (my newbie fault) that it was 
so closely related to VMWare.

I realize you then have the problem of how to compile the new
kernel when it takes so long to do anything with the standard
GENERIC kernel.  Perhaps it would go better if you booted up 
in
single-user mode, and then compiled and installed the new 
kernel.

That's a good idea, even if I'm not sure single user would 
make a difference. I imagine that the same machine code is 
being called when doing a make depend, make, cc  stuff.

My short term method has been (all day long) to reboot the 
guest os from time to time (when things get slow, but before 
things get *BLOCKED*) and relaunch the build procedure.

However, this is quite difficult and cumbersome to achieve.

I, for sure, would have preferred a sysctl option if I 
understand that it is far too low level for this.

In the end, compiling a new kernel on a non-VMWare host sounds 
interesting but might sound bizarre when using the VMWare-do-
not-need-new-hardware-to-try-new-OS approach.

My new kernel build is not over right now and I'll probably 
find it blocked tomorrow morning because of no reboot :(

Thanks for your help anyway, and sorry for the RTFM glitch ;-)

Yvon

Accédez au courrier électronique de La Poste : www.laposte.net ; 
3615 LAPOSTENET (0,13 €/mn) ; tél : 08 92 68 13 50 (0,34€/mn)




To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-questions in the body of the message