Re: [PATCH -v2] use per cpu data for single cpu ipi calls

[Peter Zijlstra <peterz@infradead.org>]

On Thu, 2009-01-29 at 10:08 -0800, Linus Torvalds wrote:

> And are you really sure it cannot be called from within interrupts? I'm 
> finding a lot of callers of smp_call_function_single(), and while I 
> couldn't find any that look like interrupts, I also couldn't find any 
> indication that it never happens.

Calls with wait=1 are deadlockable when done from within irqs or irq
disabled sections because two cpus could cross ipi each other and stay
waiting.

That leaves us with the !wait case, which wasn't safe because that
kmalloc could fail, and the fallback was wait, which yields the same
deadlock.

That leaves us with 6 cases,

root@laptop:/usr/src/linux-2.6# git grep "smp_call_function_single(.*0)"
arch/blackfin/kernel/kgdb.c:    smp_call_function_single(cpu, kgdb_passive_cpu_callback, NULL, 0);
arch/ia64/kernel/smpboot.c:     if (smp_call_function_single(master, sync_master, NULL, 0) < 0) {
arch/ia64/kvm/kvm-ia64.c:               smp_call_function_single(ipi_pcpu, vcpu_kick_intr, vcpu, 0);
arch/x86/kvm/x86.c:             smp_call_function_single(ipi_pcpu, vcpu_kick_intr, vcpu, 0);
arch/x86/oprofile/nmi_int.c:            smp_call_function_single(cpu, nmi_cpu_start, NULL, 0);
arch/x86/pci/amd_bus.c:         smp_call_function_single(cpu, enable_pci_io_ecs, NULL, 0);

The kgdb one looks to be buggy, as it calls smp_call_function_single()
from under irqs disabled.

Such users should use __smp_call_single_function() with a pre-allocated
csd.

Didn't check the others.