linuxbios, linux, hyperthreading, ACPI, and 2.4.22

ron minnich rminnich at lanl.gov
Tue Sep 9 18:13:01 CEST 2003


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




More information about the coreboot mailing list