We're seeing an interesting problem here. 2.4.22 appears to want to rely on ACPI tables for information on hyperthreading. Diffs seem to show that the linux code that works directly with the hardware to work out hyperthreading is gone.
What's happened now is that a dell box, without acpi bios support, will no longer enable hyperthreading. Turn the ACPI support on, and, on the occasions when 2.4.22 doesn't panic when coming up, it does find hyperthreaded CPUs. To panic 2.4.22 while coming up, just hit random keyboard characters while it is engaged in working out hyperthreading. It dies or, sometimes, locks up.
This seems like a counterproductive change to me, as it moves Linux more in the direction of depending on factory BIOSes which may, at times, be broken.
Has anyone else looked at 2.4.22, and can they confirm or disprove what we think we're seeing?
thanks
ron
Funny you should ask. I was just preparing 2.4.22 again. It does detect 4 processors, though it gives them rather strange numbers. This might be normal?
Attached is my serial console log for a Tyan Tiger i7501 and also attached is my 2.4.22 config file.
- Adam Agnew
On Tue, 9 Sep 2003, ron minnich wrote:
We're seeing an interesting problem here. 2.4.22 appears to want to rely on ACPI tables for information on hyperthreading. Diffs seem to show that the linux code that works directly with the hardware to work out hyperthreading is gone.
What's happened now is that a dell box, without acpi bios support, will no longer enable hyperthreading. Turn the ACPI support on, and, on the occasions when 2.4.22 doesn't panic when coming up, it does find hyperthreaded CPUs. To panic 2.4.22 while coming up, just hit random keyboard characters while it is engaged in working out hyperthreading. It dies or, sometimes, locks up.
This seems like a counterproductive change to me, as it moves Linux more in the direction of depending on factory BIOSes which may, at times, be broken.
Has anyone else looked at 2.4.22, and can they confirm or disprove what we think we're seeing?
thanks
ron
Linuxbios mailing list Linuxbios@clustermatic.org http://www.clustermatic.org/mailman/listinfo/linuxbios
Greetings,
The CPU numbers are normal enough. Or at least that's what I get on my boards.
G'day, sjames
On Tue, 9 Sep 2003, Adam Agnew wrote:
Funny you should ask. I was just preparing 2.4.22 again. It does detect 4 processors, though it gives them rather strange numbers. This might be normal?
Attached is my serial console log for a Tyan Tiger i7501 and also attached is my 2.4.22 config file.
- Adam Agnew
On Tue, 9 Sep 2003, ron minnich wrote:
We're seeing an interesting problem here. 2.4.22 appears to want to rely on ACPI tables for information on hyperthreading. Diffs seem to show that the linux code that works directly with the hardware to work out hyperthreading is gone.
What's happened now is that a dell box, without acpi bios support, will no longer enable hyperthreading. Turn the ACPI support on, and, on the occasions when 2.4.22 doesn't panic when coming up, it does find hyperthreaded CPUs. To panic 2.4.22 while coming up, just hit random keyboard characters while it is engaged in working out hyperthreading. It dies or, sometimes, locks up.
This seems like a counterproductive change to me, as it moves Linux more in the direction of depending on factory BIOSes which may, at times, be broken.
Has anyone else looked at 2.4.22, and can they confirm or disprove what we think we're seeing?
thanks
ron
Linuxbios mailing list Linuxbios@clustermatic.org http://www.clustermatic.org/mailman/listinfo/linuxbios
On Tue, 9 Sep 2003, Adam Agnew wrote:
Funny you should ask. I was just preparing 2.4.22 again. It does detect 4 processors, though it gives them rather strange numbers. This might be normal?
the numbers look ok.
Now I'm more confused than ever. Why would linuxbios work, and the Dell with ACPI turned off not work?
Could it be that the dell bios won't turn on hyperthreading unless ACPI is on?
This is 2.4.22 right?
ron
On Tue, 9 Sep 2003, ron minnich wrote:
Now I'm more confused than ever. Why would linuxbios work, and the Dell with ACPI turned off not work?
Could it be that the dell bios won't turn on hyperthreading unless ACPI is on?
Your guess is better than mine :)
This is 2.4.22 right?
Yes, sure is.
- Adam Agnew
-
Adam Agnew -
ron minnich -
steven james