Re: One more boobytrap needed for 2.2.15pre ?

From: Manfred Spraul (manfreds@colorfullife.com)
Date: Wed Feb 16 2000 - 03:20:00 EST

Alan Cox wrote:
>
> > Ok, but now the question is: which interrupt changes current->state? Perhaps
> > a special boobytrap in the interrupt handlers could detect that?
>
> Numerous interrupt handlers change task->state. Every interrupt handler that
> does a wake_up_interruptible() for example. It could easily be 'current'
>
I've written such a bobbytrap, it's attached. I'm using it right now,
but as I've expected, I receive no messages. Either a specific hardware
driver calls "__set_current_state()/ set_current_state()/
current->state==NULL", or my theory is wrong.

Andris, could you apply my patch, and check if it reports any messages? 

--
	Manfred

--- 2.2/kernel/sched.c Tue Jan 4 22:23:55 2000 +++ build-2.2/kernel/sched.c Wed Feb 16 08:37:23 2000 @@ -441,6 +441,7 @@ * "current->state = TASK_RUNNING" to mark yourself runnable * without the overhead of this. */ +extern unsigned char woken[NR_CPUS]; void wake_up_process(struct task_struct * p) { unsigned long flags; @@ -453,6 +454,8 @@ if (p->next_run) goto out; add_to_runqueue(p); + if(p==current) + woken[smp_processor_id()] = 1; spin_unlock_irqrestore(&runqueue_lock, flags); reschedule_idle(p); --- 2.2/arch/i386/kernel/irq.c Wed Oct 20 16:59:36 1999 +++ build-2.2/arch/i386/kernel/irq.c Wed Feb 16 08:48:49 2000 @@ -803,6 +803,7 @@ * SMP cross-CPU interrupts have their own specific * handlers). */ +unsigned char woken[NR_CPUS]; asmlinkage void do_IRQ(struct pt_regs regs) { /* @@ -817,11 +818,24 @@ */ int irq = regs.orig_eax & 0xff; /* subtle, see irq.h */ int cpu = smp_processor_id(); + unsigned long state; + int checking = 0; + woken[cpu]=0; + state = current->state; kstat.irqs[cpu][irq]++; irq_desc[irq].handler->handle(irq, &regs); - /* + if(woken[cpu]==0) { + if(current->state != state) { +static int limit=20; + if(limit>0) { + limit--; + printk("do_IRQ: bad state change in irq %d.\n",irq); + } + } + current->state = state; + } /* * This should be conditional: we should really get * a return code from the irq handler to tell us * whether the handler wants us to do software bottom @@ -830,6 +844,16 @@ if (1) { if (bh_active & bh_mask) do_bottom_half(); + } + if(woken[cpu]==0) { + if(current->state != state) { +static int limit=20; + if(limit>0) { + limit--; + printk("do_IRQ: bad state change in bh.\n"); + } + } + current->state = state; } }

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Wed Feb 23 2000 - 21:00:14 EST