Bug 81445

Summary:	SMP kernel fails to see second physical CPU with hyperthreading disabled.
Product:	[Retired] Red Hat Linux	Reporter:	Larry Troan <ltroan>
Component:	kernel	Assignee:	Arjan van de Ven <arjanv>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Brian Brock <bbrock>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	9	CC:	ichute, tao
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2003-04-05 16:31:52 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	79579

Description Larry Troan 2003-01-09 16:16:57 UTC

SMP kernel fails to see second physical CPU with hyperthreading disabled 

Operating System: Red Hat 8.1 Beta 1

Details:
Server (and Generation): ML370-G3
Rom version: P28 (8/16/2002)  
Processor info (class, speed & number of processors): 2 - 2.2GHz/512KB Xeons
Memory (amount only): 512M
What cards are installed and slot locations (also include firmware versions):  
  Base configuration
  Broadcom NetXtreme boot rom v2.2.7

Drive locations: Internal bays 0 and 1 (9.1GB each)

Boot controller:  3.02.01 Compaq Ultra 3 (Adaptec)

Steps to reproduce:
1) Disable hyperthreading in RBSU
2) Install the OS (text-mode)
3) After installation, prompted to load either SMP or UP kernel by grub, choose SMP.
4) cat /proc/cpuinfo, reports only one CPU instead of 2.
5) init 6, enabled hyperthreading in RBSU and repeat step 3.
6) cat /proc/cpuinfo, reports four (4) CPUs now.

uname -a output: Linux ml370g3 2.4.19-0.pp.15smp #1 SMP Thu Nov 14 13:57:02 EST
2002                  i686 i686 i386 GNU/Linux

Is the issue reproducible? Yes, repeatable on this system, as well as the 380-G3.  

Workaround: None found.
----------
Action by: Bryan.Leopard
Issue Registered
----------
Action by: arjanv
please try "acpi=off noht" as kernel option.
If that fixes it your ACPI bios has a bug where it enumerates CPU's wrong.

Status set to: Waiting on Client

----------
Action by: Bryan.Leopard
That does work, in the sense that it allows the system to see
both physical processors, but HT is obviously still an issue.  

Status set to: Waiting on Tech

Comment 1 Larry Troan 2003-01-09 16:17:58 UTC

Issue Tracker 11856

Comment 2 Arjan van de Ven 2003-01-10 09:41:34 UTC

please retry this with the rawhide kernel; the HT cpu enumeration is
significantly modified there.

Comment 3 Arjan van de Ven 2003-04-05 16:31:52 UTC

closed due to inactivity in NEEDINFO