Boot of smp system hangs in wakeup_secondary_cpu

[Bernhard Heibler <bernhard@heibler.de>]

I try to resolve the following issue. My SMP System hangs during boot. 

Software Environment: 2.6.8.1 but also happens with older 2.6 kernels
Hardware Environment: 2 x Pentium 4 Xeon 2.8 GHZ Hyper Threading, Tyan 
Tiger i7505

The hangup doesn't happen always it happens in maybe 33% of boot tries. 
It seams that the problem doesn't happen if Hyper threading is disabled 
in the Bios.

The boot process stops at this point:

....
Booting processor 2/6 eip 3000
Initializing CPU#2
masked ExtINT on CPU#2
ESR value before enabling vector: 00000000
ESR value after enabling vector: 00000000
Calibrating delay loop... 4767.74 BogoMIPS
CPU: Trace cache: 12K uops, L1 D cache: 8K
CPU: L2 cache: 512K
CPU: Physical Processor ID: 6
Intel machine check architecture supported.
Intel machine check reporting enabled on CPU#2.
CPU#2: Intel P4/Xeon Extended MCE MSRs (12) available
CPU2: Intel(R) Xeon(TM) CPU 2.40GHz stepping 07
Booting processor 3/7 eip 3000


The hangup doesn't happen all the time at the same CPU. Sometimes it 
happens at the second sometimes at the third one.
See my bugreport for a older kernel version for the full boot log file:  
http://bugme.osdl.org/show_bug.cgi?id=1659

I have add more printouts to the kernel and found out that the boot 
process gets stuck in the function  wakeup_secondary_cpu in 
arch/i386/kernel/smpboot.c  In my kernel the WAKE_SECONDARY_VIA_INIT 
version of wakeup_secondary  is used:

....
apic_write_around(APIC_ICR, APIC_INT_LEVELTRIG | APIC_INT_ASSERT  | 
APIC_DM_INIT);
Dprintk("Waiting for send to finish...\n");
 timeout = 0;
        do {
                Dprintk("+");
                udelay(100);
                send_status = apic_read(APIC_ICR) & APIC_ICR_BUSY;
        } while (send_status && (timeout++ < 1000));
Dprintk("Before wait\n");
######### STUCK HERE ###############
mdelay(10);
Dprintk("Deasserting INIT.\n");

If the problem happens the last thing printed out is my debug output 
before wait. The strange thing is that the problem doesn't seam to 
happen so often since I have added the debug printouts. Could this be a 
timing problem ? Could someone give me some hints how to debug this 
further ? Any hints are welcome.

Thanks
Bernhard.