[Techtalk] Hyper-threading: identical circumstances, different boot behavior

[Dominik Schramm]

Hi all,

I have two identical servers (Dell 1750, with 3-disk SCSI RAID-5)
booting the same customized monolithic 2.4.22 kernel. It contains the
usual stuff plus drivers for the Perc/4i ("megaraid") RAID controller,
the Broadcom Tigon Gbit NIC and ACPI support for "Enumerating logical
processors" (hyper-threading).

Hyper-threading is enabled in the BIOS.  When this kernel boots the
first (older) machine, it lists four processors (#0, #6, #1, and
#7). When it boots the other (newer) one, it lists only two (#0 
and #6).  Here's an extract from /var/log/boot.msg (the OS is SuSE
7.3).

Note that the physical processor with ID 0 is "checked" (or whatever)
twice (which does not happen on the working server); also, 
there are two warnings that "No sibling [could be] found for CPU".

I tried several kernel options with lilo, like "forceacpi=yes"
(or something similar, I don't remember; I read about them in 
newsgroup posts with similar topics), but none of them worked.
Also, the working setup on the older server runs without any such 
options.

Any ideas or suggestions what I could try or where I could look?
I'm on the brink of my sanity.

Thanks in advance.
Dominik


<4>Linux version 2.4.22 (root at deb003) (gcc version 2.95.4 20011002 (Debian
prerelease)) #7 SMP Mon Nov 10 14:51:58 CET 2003
<6>BIOS-provided physical RAM map:
<4> BIOS-e820: 0000000000000000 - 00000000000a0000 (usable)
<4> BIOS-e820: 0000000000100000 - 00000000effd7000 (usable)
<4> BIOS-e820: 00000000effd7000 - 00000000effefc00 (ACPI data)
<4> BIOS-e820: 00000000effefc00 - 00000000effff000 (reserved)
<4> BIOS-e820: 00000000fec00000 - 00000000fec10000 (reserved)
<4> BIOS-e820: 00000000fee00000 - 00000000fee10000 (reserved)
<4> BIOS-e820: 00000000fff80000 - 0000000100000000 (reserved)
<5>2943MB HIGHMEM available.
<5>896MB LOWMEM available.
<4>found SMP MP-table at 000fe710
<4>hm, page 000fe000 reserved twice.
<4>hm, page 000ff000 reserved twice.
<4>hm, page 000f0000 reserved twice.
<4>On node 0 totalpages: 982999
<4>zone(0): 4096 pages.
<4>zone(1): 225280 pages.
<4>zone(2): 753623 pages.
<4>Intel MultiProcessor Specification v1.4
<4>    Virtual Wire compatibility mode.
<4>OEM ID: DELL     Product ID: PE 014A      APIC at: 0xFEE00000
<4>Processor #0 Pentium 4(tm) XEON(tm) APIC version 20
<4>Processor #6 Pentium 4(tm) XEON(tm) APIC version 20
<4>I/O APIC #8 Version 17 at 0xFEC00000.
<4>I/O APIC #9 Version 17 at 0xFEC01000.
<4>I/O APIC #10 Version 17 at 0xFEC02000.
<4>Enabling APIC mode: Flat.    Using 3 I/O APICs
<4>Processors: 2
<4>Kernel command line: auto BOOT_IMAGE=suse ro root=805
<6>Initializing CPU#0
<4>Detected 2788.995 MHz processor.
<4>Console: colour VGA+ 80x25
<4>Calibrating delay loop... 5570.56 BogoMIPS
<6>Memory: 3882996k/3931996k available (1198k kernel code, 48616k reserved,
351k data, 272k init, 3014492k highmem)
<6>Dentry cache hash table entries: 262144 (order: 9, 2097152 bytes)
<6>Inode cache hash table entries: 262144 (order: 9, 2097152 bytes)
<6>Mount cache hash table entries: 512 (order: 0, 4096 bytes)
<6>Buffer cache hash table entries: 262144 (order: 8, 1048576 bytes)
<4>Page-cache hash table entries: 524288 (order: 9, 2097152 bytes)
<6>CPU: Trace cache: 12K uops, L1 D cache: 8K
<6>CPU: L2 cache: 512K
<6>CPU: Physical Processor ID: 0
<6>Intel machine check architecture supported.
<6>Intel machine check reporting enabled on CPU#0.
<7>CPU:     After generic, caps: bfebfbff 00000000 00000000 00000000
<7>CPU:             Common caps: bfebfbff 00000000 00000000 00000000
<6>Enabling fast FPU save and restore... done.
<6>Enabling unmasked SIMD FPU exception support... done.
<6>Checking 'hlt' instruction... OK.
<4>POSIX conformance testing by UNIFIX
<4>mtrr: v1.40 (20010327) Richard Gooch (rgooch at atnf.csiro.au)
<4>mtrr: detected mtrr type: Intel
<6>CPU: Trace cache: 12K uops, L1 D cache: 8K
<6>CPU: L2 cache: 512K
<6>CPU: Physical Processor ID: 0
<6>Intel machine check reporting enabled on CPU#0.
<7>CPU:     After generic, caps: bfebfbff 00000000 00000000 00000000
<7>CPU:             Common caps: bfebfbff 00000000 00000000 00000000
<4>CPU0: Intel(R) Xeon(TM) CPU 2.80GHz stepping 09
<4>per-CPU timeslice cutoff: 1463.04 usecs.
<4>enabled ExtINT on CPU#0
<4>ESR value before enabling vector: 00000040
<4>ESR value after enabling vector: 00000000
<4>Booting processor 1/6 eip 2000
<6>Initializing CPU#1
<4>masked ExtINT on CPU#1
<4>ESR value before enabling vector: 00000000
<4>ESR value after enabling vector: 00000000
<4>Calibrating delay loop... 5570.56 BogoMIPS
<6>CPU: Trace cache: 12K uops, L1 D cache: 8K
<6>CPU: L2 cache: 512K
<6>CPU: Physical Processor ID: 3
<6>Intel machine check reporting enabled on CPU#1.
<7>CPU:     After generic, caps: bfebfbff 00000000 00000000 00000000
<7>CPU:             Common caps: bfebfbff 00000000 00000000 00000000
<4>CPU1: Intel(R) Xeon(TM) CPU 2.80GHz stepping 09
<6>Total of 2 processors activated (11141.12 BogoMIPS).
<4>WARNING: No sibling found for CPU 0.
<4>WARNING: No sibling found for CPU 1.
<4>ENABLING IO-APIC IRQs
<4>Setting 8 in the phys_id_present_map