On Tue, 11 Mar 2003, Zwane Mwaikambo wrote:
> Have you had an opportunity to test this code yet? If you can't get that
> large a 32bit SMP box, you could fire off some 'extra' cpus on i386 then
> fail them during boot around cpu_up, making sure to put the failed cpus
> somewhere in the middle of your cpu map so that not only do you get a
> sparse map but some on the other side of 32.
Something crackpot ala;
The new boot process would look something like this (pseudo code as i
don't know the types you use yet), just watch out for num_booting_cpus
users.
/* the fake cpus we're going to fire up and fail */
cpumask_t fake_cpu_map = (1 << 13) | (1 << 14) | (1 << 15);
arch/i386/kernel/smpboot.c
smp_boot_cpus() {
...
Dprintk("CPU present map: %lx\n", phys_cpu_present_map);
for (bit = 0; bit < NR_CPUS; bit++) {
if ((1 << bit) & fake_cpu_map))
set_bit(bit, cpu_callin_map);
....
}
}
do_boot_cpu() {
cpu = ++cpucount;
if ((1 << cpu) & fake_cpu_map) {
/* you could probably stop here and set the boot error
* without decrementing the cpucount */
goto skip_hardware_wakeup_bits;
}
skip_hardware_wakeup_bits:
return boot_error;
}
Anyway coffee is wearing off and i must sleep...
Zwane
-- function.linuxpower.ca - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Mar 15 2003 - 22:00:24 EST