All of lore.kernel.org
 help / color / mirror / Atom feed
* 2.4.22 SMP kernel build for hyper threading P4
@ 2003-11-15 15:40 Job 317
  2003-11-15 16:40 ` Eduard Bloch
  0 siblings, 1 reply; 5+ messages in thread
From: Job 317 @ 2003-11-15 15:40 UTC (permalink / raw)
  To: linux-kernel

[-- Attachment #1: Type: text/plain, Size: 707 bytes --]

Alles,

I have a new P4 Hyperthreading PC that I have installed RedHat9 on. I am
trying to add some features to my kernel (e.g. NTFS read support, crypto
api, etc.) but when I build my new 2.4.22 kernel with CONFIG_SMP=y set
then reboot, I am seeing only one processor in /proc/cpuinfo. When I
boot with my stock RH9 2.4.20-20.9smp kernel I see two virtual
processors show up in /proc/cpuinfo.

I've even build a 2.4.22 kernel with the config-2.4.20-20.9smp
configuration that came with RH9.

I build with the straightforward 'make dep clean bzImage modules
modules_install' command. Is this correct?

Am I missing a step to build a smp kernel for hyper threading?

Your help would be most appreciated.

Job

^ permalink raw reply	[flat|nested] 5+ messages in thread
* Re: 2.4.22 SMP kernel build for hyper threading P4
@ 2003-11-15 18:36 Job 317
  0 siblings, 0 replies; 5+ messages in thread
From: Job 317 @ 2003-11-15 18:36 UTC (permalink / raw)
  To: Mark Hahn; +Cc: linux-kernel

[-- Attachment #1: Type: text/plain, Size: 2138 bytes --]

On 15-Nov-2003 17:30:11 +0100, you wrote:
> > I have a new P4 Hyperthreading PC that I have installed RedHat9 on.
I am
> 
> I presume you realize that anyone who understands HT has mostly scorn
for it.

I had no idea but it is still a huge improvement over my HP 755MHz
Celleron that I am replacing ;) Might I as why so and am I wasting my
time when a non-SMP kernel builds and works just fine?

> 
> > api, etc.) but when I build my new 2.4.22 kernel with CONFIG_SMP=y
set
> > then reboot, I am seeing only one processor in /proc/cpuinfo. When
I
> > boot with my stock RH9 2.4.20-20.9smp kernel I see two virtual
> > processors show up in /proc/cpuinfo.
> 
> I'm guessing you don't have acpi enabled.

Hmm... Let's have a look. I started by loading the
kernel-2.4.20-i686-smp.config file that came with the RH9 distro into
xconfig. I just turned on ntfs and a couple of other things and saved.
The comparision is strange. My new config file has:

# CONFIG_ACPI is not set

however the RH9 config file has:

# CONFIG_ACPI is not set
# CONFIG_ACPI_DEBUG is not set
# CONFIG_HOTPLUG_PCI_ACPI is not set
# CONFIG_ACPI_HT_ONLY is not s
CONFIG_ACPI_BUSMGR=m
CONFIG_ACPI_SYS=m
CONFIG_ACPI_CPU=m
CONFIG_ACPI_BUTTON=m
CONFIG_ACPI_AC=m
CONFIG_ACPI_EC=m
CONFIG_ACPI_CMBATT=m
CONFIG_ACPI_THERMAL=m
CONFIG_ACPI_RELAXED_AML=y

So should I have all this or no?

> 
> > I've even build a 2.4.22 kernel with the config-2.4.20-20.9smp
> > configuration that came with RH9.
> 
> RH doesn't ship kernel.org sources, of course.  hence the -20.

Rgr. I got this.

> 
> > I build with the straightforward 'make dep clean bzImage modules
> > modules_install' command. Is this correct?
> 
> sure.  obviously, you need to verify that you're booting the kernel
> that you just made, etc.

It does boot and everything else "seems" fine. Just that I still only
get one virtual CPU showing up in /proc/cpuinfo.

> 
> > Am I missing a step to build a smp kernel for hyper threading?
> 
> HT is a non-feature; it requires nothing from the kernel.  I'm
> guessing your problem is bios or acpi.
> 

I still don't understand. So am I wasting my time with HT?

Thanks,

Job

^ permalink raw reply	[flat|nested] 5+ messages in thread
* RE: 2.4.22 SMP kernel build for hyper threading P4
@ 2003-11-23 20:06 Brown, Len
  0 siblings, 0 replies; 5+ messages in thread
From: Brown, Len @ 2003-11-23 20:06 UTC (permalink / raw)
  To: Job 317, Mark Hahn; +Cc: linux-kernel

> Am I missing a step to build a smp kernel for hyper threading?

Yes -- below.

> Hmm... Let's have a look. I started by loading the
> kernel-2.4.20-i686-smp.config file that came with the RH9 distro into
> xconfig. I just turned on ntfs and a couple of other things and saved.
> The comparision is strange. My new config file has:
> 
> # CONFIG_ACPI is not set
> 
> however the RH9 config file has:
> 
> # CONFIG_ACPI is not set
> # CONFIG_ACPI_DEBUG is not set
> # CONFIG_HOTPLUG_PCI_ACPI is not set
> # CONFIG_ACPI_HT_ONLY is not s
> CONFIG_ACPI_BUSMGR=m
> CONFIG_ACPI_SYS=m
> CONFIG_ACPI_CPU=m
> CONFIG_ACPI_BUTTON=m
> CONFIG_ACPI_AC=m
> CONFIG_ACPI_EC=m
> CONFIG_ACPI_CMBATT=m
> CONFIG_ACPI_THERMAL=m
> CONFIG_ACPI_RELAXED_AML=y

Use "make oldconfig" to re-write your .config and delete stale junk and
options whose depencencies make no sense.  For example, the ACPI drivers
will go away if CONFIG_ACPI is not enabled.

For 2.4.22, try this:

CONFIG_ACPI=y

By itself for full ACPI -- which includes HT.

And this way to limit it to just HT:

CONFIG_ACPI=y
CONFIG_ACPI_HT_ONLY=y

Note that this config scheme proved too confusing and in 2.4.23 and
2.6.0 CONFIG_ACPI_HT_ONLY goes away, and you'll get enough HT to enable
ACPI even without CONFIG_ACPI, as long as you specify CONFIG_SMP in 2.4
or CONFIG_X86_HT=y in 2.6.

> So am I wasting my time with HT?

Measure the performance of something that matters to you before/after
and then you can decide based on objective information.

Cheers,
-Len

Ps. If you run ACPI on 2.4.22 -- we support a patch supplying the latest
ACPI code for that release here:
http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/
2.4.22/  This includes the 2.4.23 config changes above, among other
fixes.




^ permalink raw reply	[flat|nested] 5+ messages in thread
end of thread, other threads:[~2003-11-23 20:07 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-11-15 15:40 2.4.22 SMP kernel build for hyper threading P4 Job 317
2003-11-15 16:40 ` Eduard Bloch
2003-11-23 15:06   ` not fixed in 2.4.23-rc3 (was: Re: 2.4.22 SMP kernel build for hyper threading P4) Eduard Bloch
  -- strict thread matches above, loose matches on Subject: below --
2003-11-15 18:36 2.4.22 SMP kernel build for hyper threading P4 Job 317
2003-11-23 20:06 Brown, Len

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.