SMP Madness

[dogbert at clue4all.net (Brian J. Conway)]

This is a rather in-depth SMP problem, but if anyone has a good amount
of experience with Intel SMP and wishes to take a look, I'd be very
grateful.  I recently upgraded from a single PIII-450 cpu on my Tyan
Tiger 100 board to two PIII-550s, and am having some very strange
problems with SMP.  First off, here are the related specs:

Tyan Tiger S1832DL 100 dual motherboard with the latest bios revision
two PIII-550 cpus of the same stepping
MPS revision 1.4 set in bios (same results with 1.1)
kernel 2.2.15 with smp and rtc enabled, apm disabled (same results with
2.4.0test1 and 2.2.16pre)

The kinda behavior I'm getting is randomly having a CPU run at 100% for
a period of time while doing something heavily threaded (like xmms) or
just completely randomly, then settle down and have both CPUs running at
50% while just playing an mp3.  This is, of course, ridiculous, because
when I use a kernel with just one CPU (as well as in Windows, sorry)
everything is just fine, and mp3s take a whopping 3% CPU to play.  Also,
while playing Q3 or Unreal Tournament in an SMP kernel, both processors
are running and supposedly using half the load each, even without having
SMP enabled in the program (as neither support it, and only one
processor should be doing the work), and my mouse freezes periodically
for a second every 5 seconds or so when both processors max out while
playing. I've tried removing various game configuration files, but the
same thing still happens when I start them up again.  I really have no
idea why both processors start running crazy in SMP (kernel loads fine
and detects everything).  The system runs beautiful as uniprocessor, and
I've verified all the hardware.  I've included the dmesg output for sake
of hopefully finding something.  Of the people that use the Tiger 100,
any advice on the BIOS revision number that you have working or any
important settings?  I've heard many commments about Tyan having
beautiful hardware and shitty bioses.  This is both strange and
agravating.  Any help would be greatly appreciated (and save a lot of
money on a processor that's currently unused).  Please help my runaway
processors.

Brian J. Conway
dogbert at clue4all.net
-------------- next part --------------
Jun  2 18:59:30 ladyluck kernel: Linux version 2.2.15 (root at ladyluck.clue4all.net) (gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)) #1 SMP Fri Jun 2 18:51:53 EDT 2000
Jun  2 18:59:30 ladyluck kernel: Intel MultiProcessor Specification v1.4
Jun  2 18:59:30 ladyluck kernel:     Virtual Wire compatibility mode.
Jun  2 18:59:30 ladyluck kernel: OEM ID: INTEL    Product ID: 440GX        APIC at: 0xFEE00000
Jun  2 18:59:30 ladyluck kernel: Processor #0 Pentium(tm) Pro APIC version 17
Jun  2 18:59:30 ladyluck kernel: Processor #1 Pentium(tm) Pro APIC version 17
Jun  2 18:59:30 ladyluck kernel: I/O APIC #2 Version 17 at 0xFEC00000.
Jun  2 18:59:30 ladyluck kernel: Processors: 2
Jun  2 18:59:30 ladyluck kernel: mapped APIC to ffffe000 (fee00000)
Jun  2 18:59:30 ladyluck kernel: mapped IOAPIC to ffffd000 (fec00000)
Jun  2 18:59:30 ladyluck kernel: Detected 551261242 Hz processor.
Jun  2 18:59:30 ladyluck kernel: Console: colour dummy device 80x25
Jun  2 18:59:30 ladyluck kernel: Calibrating delay loop... 1101.00 BogoMIPS
Jun  2 18:59:30 ladyluck kernel: Memory: 322728k/327680k available (948k kernel code, 424k reserved, 3512k data, 68k init)
Jun  2 18:59:30 ladyluck kernel: Dentry hash table entries: 65536 (order 7, 512k)
Jun  2 18:59:30 ladyluck kernel: Buffer cache hash table entries: 524288 (order 9, 2048k)
Jun  2 18:59:30 ladyluck kernel: Page cache hash table entries: 131072 (order 7, 512k)
Jun  2 18:59:30 ladyluck kernel: Checking 386/387 coupling... OK, FPU using exception 16 error reporting.
Jun  2 18:59:30 ladyluck kernel: Checking 'hlt' instruction... OK.
Jun  2 18:59:30 ladyluck kernel: POSIX conformance testing by UNIFIX
Jun  2 18:59:30 ladyluck kernel: mtrr: v1.35a (19990819) Richard Gooch (rgooch at atnf.csiro.au)
Jun  2 18:59:30 ladyluck kernel: per-CPU timeslice cutoff: 100.04 usecs.
Jun  2 18:59:30 ladyluck kernel: CPU0: Intel Pentium III (Katmai) stepping 03
Jun  2 18:59:30 ladyluck kernel: calibrating APIC timer ...
Jun  2 18:59:30 ladyluck kernel: ..... CPU clock speed is 551.2800 MHz.
Jun  2 18:59:30 ladyluck kernel: ..... system bus clock speed is 100.2326 MHz.
Jun  2 18:59:30 ladyluck kernel: Booting processor 1 eip 2000
Jun  2 18:59:30 ladyluck kernel: Calibrating delay loop... 1101.00 BogoMIPS
Jun  2 18:59:30 ladyluck kernel: OK.
Jun  2 18:59:30 ladyluck kernel: CPU1: Intel Pentium III (Katmai) stepping 03
Jun  2 18:59:30 ladyluck kernel: Total of 2 processors activated (2202.01 BogoMIPS).
Jun  2 18:59:30 ladyluck kernel: enabling symmetric IO mode... ...done.
Jun  2 18:59:30 ladyluck kernel: ENABLING IO-APIC IRQs
Jun  2 18:59:30 ladyluck kernel: init IO_APIC IRQs
Jun  2 18:59:30 ladyluck kernel:  IO-APIC (apicid-pin) 2-0, 2-5, 2-9, 2-10, 2-11, 2-20, 2-21, 2-22, 2-23 not connected.
Jun  2 18:59:30 ladyluck kernel: number of MP IRQ sources: 17.
Jun  2 18:59:30 ladyluck kernel: number of IO-APIC #2 registers: 24.
Jun  2 18:59:30 ladyluck kernel: testing the IO APIC.......................
Jun  2 18:59:30 ladyluck kernel:
Jun  2 18:59:30 ladyluck kernel: IO APIC #2......
Jun  2 18:59:30 ladyluck kernel: .... register #00: 02000000
Jun  2 18:59:30 ladyluck kernel: .......    : physical APIC id: 02
Jun  2 18:59:30 ladyluck kernel: .... register #01: 00170011
Jun  2 18:59:30 ladyluck kernel: .......     : max redirection entries: 0017
Jun  2 18:59:30 ladyluck kernel: .......     : IO APIC version: 0011
Jun  2 18:59:30 ladyluck kernel: .... register #02: 00000000
Jun  2 18:59:30 ladyluck kernel: .......     : arbitration: 00
Jun  2 18:59:30 ladyluck kernel: .... IRQ redirection table:
Jun  2 18:59:30 ladyluck kernel:  NR Log Phy Mask Trig IRR Pol Stat Dest Deli Vect:
Jun  2 18:59:30 ladyluck kernel:  00 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel:  01 000 00  0    0    0   0   0    1    1    59
Jun  2 18:59:30 ladyluck kernel:  02 0FF 0F  0    0    0   0   0    1    1    51
Jun  2 18:59:30 ladyluck kernel:  03 000 00  0    0    0   0   0    1    1    61
Jun  2 18:59:30 ladyluck kernel:  04 000 00  0    0    0   0   0    1    1    69
Jun  2 18:59:30 ladyluck kernel:  05 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel:  06 000 00  0    0    0   0   0    1    1    71
Jun  2 18:59:30 ladyluck kernel:  07 000 00  0    0    0   0   0    1    1    79
Jun  2 18:59:30 ladyluck kernel:  08 000 00  0    0    0   0   0    1    1    81
Jun  2 18:59:30 ladyluck kernel:  09 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel:  0a 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel:  0b 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel:  0c 000 00  0    0    0   0   0    1    1    89
Jun  2 18:59:30 ladyluck kernel:  0d 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel:  0e 000 00  0    0    0   0   0    1    1    91
Jun  2 18:59:30 ladyluck kernel:  0f 000 00  0    0    0   0   0    1    1    99
Jun  2 18:59:30 ladyluck kernel:  10 0FF 0F  1    1    0   1   0    1    1    A1
Jun  2 18:59:30 ladyluck kernel:  11 0FF 0F  1    1    0   1   0    1    1    A9
Jun  2 18:59:30 ladyluck kernel:  12 0FF 0F  1    1    0   1   0    1    1    B1
Jun  2 18:59:30 ladyluck kernel:  13 0FF 0F  1    1    0   1   0    1    1    B9
Jun  2 18:59:30 ladyluck kernel:  14 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel:  15 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel:  16 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel:  17 000 00  1    0    0   0   0    0    0    00
Jun  2 18:59:30 ladyluck kernel: IRQ to pin mappings:
Jun  2 18:59:30 ladyluck kernel: IRQ0 -> 2
Jun  2 18:59:30 ladyluck kernel: IRQ1 -> 1
Jun  2 18:59:30 ladyluck kernel: IRQ3 -> 3
Jun  2 18:59:30 ladyluck kernel: IRQ4 -> 4
Jun  2 18:59:30 ladyluck kernel: IRQ6 -> 6
Jun  2 18:59:30 ladyluck kernel: IRQ7 -> 7
Jun  2 18:59:30 ladyluck kernel: IRQ8 -> 8
Jun  2 18:59:30 ladyluck kernel: IRQ12 -> 12
Jun  2 18:59:30 ladyluck kernel: IRQ13 -> 13
Jun  2 18:59:30 ladyluck kernel: IRQ14 -> 14
Jun  2 18:59:30 ladyluck kernel: IRQ15 -> 15
Jun  2 18:59:30 ladyluck kernel: IRQ16 -> 16
Jun  2 18:59:30 ladyluck kernel: IRQ17 -> 17
Jun  2 18:59:30 ladyluck kernel: IRQ18 -> 18
Jun  2 18:59:30 ladyluck kernel: IRQ19 -> 19
Jun  2 18:59:30 ladyluck kernel: .................................... done.
Jun  2 18:59:30 ladyluck kernel: checking TSC synchronization across CPUs: passed.
Jun  2 18:59:30 ladyluck kernel: PCI: PCI BIOS revision 2.10 entry at 0xfdb71
Jun  2 18:59:30 ladyluck kernel: PCI: Using configuration type 1
Jun  2 18:59:30 ladyluck kernel: PCI: Probing PCI hardware
Jun  2 18:59:30 ladyluck kernel: PCI: Enabling I/O for device 00:3a
Jun  2 18:59:30 ladyluck kernel: PCI->APIC IRQ transform: (B0,I17,P0) -> 17
Jun  2 18:59:30 ladyluck kernel: PCI->APIC IRQ transform: (B0,I18,P0) -> 18
Jun  2 18:59:30 ladyluck kernel: PCI->APIC IRQ transform: (B0,I19,P0) -> 19
Jun  2 18:59:30 ladyluck kernel: PCI->APIC IRQ transform: (B1,I0,P0) -> 16
Jun  2 18:59:30 ladyluck kernel: Linux NET4.0 for Linux 2.2
Jun  2 18:59:30 ladyluck kernel: Based upon Swansea University Computer Society NET3.039
Jun  2 18:59:30 ladyluck kernel: NET4: Unix domain sockets 1.0 for Linux NET4.0.
Jun  2 18:59:30 ladyluck kernel: NET4: Linux TCP/IP 1.0 for NET4.0
Jun  2 18:59:30 ladyluck kernel: IP Protocols: ICMP, UDP, TCP
Jun  2 18:59:30 ladyluck kernel: TCP: Hash tables configured (ehash 524288 bhash 65536)
Jun  2 18:59:30 ladyluck kernel: Starting kswapd v 1.5
Jun  2 18:59:30 ladyluck kernel: parport0: PC-style at 0x378 (0x778) [SPP,ECP,ECPPS2]
Jun  2 18:59:30 ladyluck kernel: fb: Voodoo3 memory = 16384K
Jun  2 18:59:30 ladyluck kernel: fb: MTRR's  turned on
Jun  2 18:59:30 ladyluck kernel: tdfxfb: reserving 1024 bytes for the hwcursor at 0xd57ff000
Jun  2 18:59:30 ladyluck crond: crond startup succeeded
Jun  2 18:59:30 ladyluck kernel: Console: switching to colour frame buffer device 128x48
Jun  2 18:59:30 ladyluck kernel: fb0: 3Dfx Voodoo3 frame buffer device
Jun  2 18:59:30 ladyluck kernel: Detected PS/2 Mouse Port.
Jun  2 18:59:30 ladyluck kernel: Serial driver version 4.27 with no serial options enabled
Jun  2 18:59:30 ladyluck kernel: ttyS00 at 0x03f8 (irq = 4) is a 16550A
Jun  2 18:59:30 ladyluck kernel: ttyS01 at 0x02f8 (irq = 3) is a 16550A
Jun  2 18:59:30 ladyluck kernel: pty: 256 Unix98 ptys configured
Jun  2 18:59:30 ladyluck kernel: lp0: using parport0 (polling).
Jun  2 18:59:30 ladyluck kernel: Real Time Clock Driver v1.09
Jun  2 18:59:30 ladyluck kernel: PIIX4: IDE controller on PCI bus 00 dev 39
Jun  2 18:59:30 ladyluck kernel: PIIX4: not 100% native mode: will probe irqs later
Jun  2 18:59:30 ladyluck kernel:     ide0: BM-DMA at 0xffa0-0xffa7, BIOS settings: hda:DMA, hdb:DMA
Jun  2 18:59:30 ladyluck kernel:     ide1: BM-DMA at 0xffa8-0xffaf, BIOS settings: hdc:DMA, hdd:DMA
Jun  2 18:59:30 ladyluck kernel: hda: SAMSUNG SV0844A, ATA DISK drive
Jun  2 18:59:30 ladyluck kernel: hdb: SAMSUNG VG34323A (4.32GB), ATA DISK drive
Jun  2 18:59:30 ladyluck kernel: hdc: CREATIVEDVD6630E, ATAPI CDROM drive
Jun  2 18:59:30 ladyluck kernel: hdd: CR-2801TE, ATAPI CDROM drive
Jun  2 18:59:30 ladyluck kernel: ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
Jun  2 18:59:30 ladyluck kernel: ide1 at 0x170-0x177,0x376 on irq 15
Jun  2 18:59:30 ladyluck kernel: hda: SAMSUNG SV0844A, 8063MB w/220kB Cache, CHS=1027/255/63, UDMA
Jun  2 18:59:30 ladyluck kernel: hdb: SAMSUNG VG34323A (4.32GB), 4124MB w/496kB Cache, CHS=525/255/63, UDMA
Jun  2 18:59:30 ladyluck kernel: Floppy drive(s): fd0 is 1.44M
Jun  2 18:59:30 ladyluck kernel: FDC 0 is a National Semiconductor PC87306
Jun  2 18:59:30 ladyluck kernel: scsi0 : SCSI host adapter emulation for IDE ATAPI devices
Jun  2 18:59:30 ladyluck kernel: scsi : 1 host.
Jun  2 18:59:30 ladyluck kernel:   Vendor: CREATIVE  Model: DVD6630E          Rev: b13A
Jun  2 18:59:30 ladyluck kernel:   Type:   CD-ROM                             ANSI SCSI revision: 02
Jun  2 18:59:30 ladyluck kernel: Detected scsi CD-ROM sr0 at scsi0, channel 0, id 0, lun 0
Jun  2 18:59:30 ladyluck kernel:   Vendor: MITSUMI   Model: CR-2801TE         Rev: 1.07
Jun  2 18:59:30 ladyluck kernel:   Type:   CD-ROM                             ANSI SCSI revision: 02
Jun  2 18:59:30 ladyluck kernel: Detected scsi CD-ROM sr1 at scsi0, channel 0, id 1, lun 0
Jun  2 18:59:30 ladyluck kernel: scsi : detected 2 SCSI cdroms total.
Jun  2 18:59:30 ladyluck kernel: sr0: scsi3-mmc drive: 38x/32x cd/rw xa/form2 cdda tray
Jun  2 18:59:30 ladyluck kernel: Uniform CDROM driver Revision: 2.56
Jun  2 18:59:30 ladyluck kernel: sr1: scsi3-mmc drive: 8x/8x writer xa/form2 cdda tray
Jun  2 18:59:30 ladyluck kernel: 3c59x.c:v0.99H 11/17/98 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/vortex.html
Jun  2 18:59:30 ladyluck kernel: eth0: 3Com 3c595 Vortex 100baseT4 at 0xef20,  00:a0:24:98:0e:c5, IRQ 18
Jun  2 18:59:30 ladyluck kernel:   64K word-wide RAM 3:1 Rx:Tx split, MII interface.
Jun  2 18:59:30 ladyluck kernel:   MII transceiver found at address 1, status 9801.
Jun  2 18:59:30 ladyluck kernel: eth0: Overriding PCI latency timer (CFLT) setting of 64, new value is 248.
Jun  2 18:59:30 ladyluck kernel: Partition check:
Jun  2 18:59:30 ladyluck kernel:  hda: hda1 hda2 hda3 hda4
Jun  2 18:59:30 ladyluck kernel:  hdb: hdb1
Jun  2 18:59:30 ladyluck kernel: VFS: Mounted root (ext2 filesystem) readonly.
Jun  2 18:59:30 ladyluck kernel: Freeing unused kernel memory: 68k freed
Jun  2 18:59:30 ladyluck kernel: Adding Swap: 40156k swap-space (priority -1)
Jun  2 18:59:30 ladyluck kernel: Creative EMU10K1 PCI Audio Driver, version 0.6, 18:57:35 Jun  2 2000
Jun  2 18:59:30 ladyluck kernel: emu10k1: EMU10K1 rev 8 model 0x8027 found, IO at 0xef40-0xef5f, IRQ 17