2 msg`set autologout = (15 1)' aborts

Re: SMP in machine/params.h
\ Pietro Cerutti (24 Jan 2008)
. \ Ivan Voras (24 Jan 2008)

4 msg[Fwd: SMP in machine/params.h]
3 msgsyncache_timer: Response timeout and other msgs...
18 msgDTrace for FreeBSD - status
4 msgRe: Low NTFS read performance
19 msgXEN and ZFS Status?
5 msgfetch fails on 8-current
1 msgPerformance Tracker project update
4 msgRELENG_7_0: vm_thread_new: kstack allocation fa...
6 msgtcsh coredumps
7 msgbsnmpd daemon eating all cpu
13 msgIPSEC on 7.0-PRERELEASE
5 msgcvs in CURRENT broken?
4 msgCFT: re(4) WOL support
8 msgCan't delete any files on my filled up ZFS pool
2 msgFreeBSD Status Reports due: January 28th, 2008
4 msgLatencyTOP and FreeBSD
1 msgbsdlabel 26 partitions
5 msgRe: SCHED_ULE & niceness / rtprio
Subject:Re: SMP in machine/params.h
Group:Freebsd-current
From:Pietro Cerutti
Date:24 Jan 2008


 

Some more interesting findings in this field lead me to another related
question:

what is the relation between kern.smp.cpus, kern.smp.active and
kern.smp.maxcpus?

To my understanding, it should be:
If the machine has more than one CPU, and SMP and apic are enabled in
the config file, then

kern.smp.cpus = nof cpus in the system
kern.smp.active = 1
kern.smp.maxcpus = 16

else

kern.smp.cpus = 1
kern.smp.active = 0
kern.smp.maxcpus = 1

This is the expected result, and also the result I get from my single
CPU Pentium 3, running 6.3-RELEASE:

kern.smp.cpus: 1
kern.smp.active: 0
kern.smp.maxcpus: 1

From my single CPU sparc64, running 7.0-PRERELEASE, I get:

kern.smp.cpus: 1
kern.smp.active: 0
kern.smp.maxcpus: 16 <-- why not 1??

From my Core2Due, running 8.0-CURRENT, I get:

kern.smp.cpus: 2
kern.smp.active: 1
kern.smp.maxcpus: 16

which is OK, but nevertheless SMP is not defined in sys/param.h (see
previous posts).


Thanks for any insight,




--
Pietro Cerutti

PGP Public Key:
http://gahr.ch/pgp




© 2004-2008 readlist.com