Re: Hyperthreading and debian kernels
Thanks. I've tried a 2.6.10 ( kernel.org version ) with SMP+SMT but it
refuse to boot, same kind of error. Is there a special option to
activate in the .config ( or during make menuconfig ) ?
It's really weird, I don't understand. If I disable HT in the bios,
it boots well.
Does the HT work really well ( I mean, more performance ) under
linux ? Better than under M$ ? Less than M$ systems ?
It's pretty cool to see cat /proc/cpuinfo returns 2 cpus^^
On Wed, 12 Jan 2005 13:13:50 +0100, maximilian attems
> On Wed, 12 Jan 2005, Rem wrote:
> > I've encoutered some problems this afternoon ( France guys, it's 9am
> > right now ;) ). I tried to install a debian on this hardware :
> > Intel i865 + dual DDR 400
> > Pentium 4 HT
> > ICH5 controller but I use one P-ATA ( IDE ) Drive.
> > I tried debian testing, with 2.6.3 kernel default, it works fine, but
> > Hyperthreading is not assumed, even if I enabled it in BIos. So I
> > recompiled the kernel, the debian 2.6.8. First, this fuck... sh.. refuse
> > to boot, unable to mount root fs. This is not my first kernel
> > recompilation, I have already compiled more than 100 kernels on several
> > computeur and servers and wrote article about it, it's my job ( System
> > engineer ). There is no log, due to the quick kernel panic, and my ide
> > controller is "built-in", such as my FS. My /boot/grub/menu.lst is good
> > too. So I tried a 2.6.7 from kernel.org,and of course it works great
> > with the same .config and even a 2.6.10 still from kernel.org and it
> > works really good too. Well, the 2.6.8 sucks...but I don't care, my
> > 2.6.10 is allright.
> 2.6.8 has known acpi problems for smp boards that prevent booting, yes.
> due too the big pile of fixes it is not possible to backport that without
> destabilising the debian sarge kernel.
> > Another thing interesteding, if I activate HT in the bios, and SMP is
> > set, the kernel won't boot, whatever the kernel version ( 2.6.8 2.6.7
> > and 2.6.10 ). It bugs, and shows messages about APIC and Null pointer
> > exception, it look likes the 2 virtuals processors don't run well
> > together. I don't understand why, is there something special to activate
> > ? In fact, it's just the SMP option who matters, SMT or not SMT.
> try the smp 2.6.10 kernel-image from unstable,
> that one should work.
> a++ maks
Remy HAREL - email@example.com
Linux Registered User #224740