Re: Strange interrupt behaviour

Andrea Arcangeli (arcangeli@mbox.queen.it)
Wed, 15 Jul 1998 03:04:25 +0200 (CEST)

On Mon, 13 Jul 1998, Linus Torvalds wrote:

>Oh, having separate interrupt stacks _works_, and on some architectures it

I hacked irq.c and now seems to work on Linux too ;-).

===================================================================
RCS file: linux/arch/i386/kernel/irq.c,v
retrieving revision 1.1
diff -u -r1.1 linux/arch/i386/kernel/irq.c
--- linux/arch/i386/kernel/irq.c 1998/07/14 18:52:27 1.1
+++ linux/arch/i386/kernel/irq.c 1998/07/15 00:51:23
@@ -46,6 +46,12 @@
atomic_t nmi_counter;

/*
+ * Per interrupt stack area.
+ */
+unsigned char irq_stacks[NR_IRQS*PAGE_SIZE] __attribute__ ((aligned (32)));
+unsigned char *old_stacks[NR_IRQS] __attribute__ ((aligned (32)));
+
+/*
* About the IO-APIC, the architecture is 'merged' into our
* current irq architecture, seemlessly. (i hope). It is only
* visible through 8 more hardware interrupt lines, but otherwise
@@ -725,6 +731,24 @@
spin_unlock_irqrestore(&irq_controller_lock, flags);
}

+static __inline__ void split_stack(register unsigned int irq)
+{
+ register unsigned char *stack = &irq_stacks[(irq+1)*PAGE_SIZE];
+ __asm__ __volatile__ ("movl %%esp, %0\n\t"
+ "movl %1, %%esp\n\t"
+ "pushl %2" :
+ "=m" (old_stacks[irq]) :
+ "r" (stack), "r" (irq));
+}
+
+static __inline__ void restore_stack(void)
+{
+ register unsigned int irq;
+ __asm__ __volatile__ ("popl %0" : "=r" (irq));
+ __asm__ __volatile__ ("movl %0, %%esp\n\t" : :
+ "m" (old_stacks[irq]));
+}
+
/*
* do_IRQ handles all normal device IRQ's (the special
* SMP cross-CPU interrupts have their own specific
@@ -755,7 +779,10 @@
int cpu = smp_processor_id();

kstat.irqs[cpu][irq]++;
+
+ split_stack(irq);
irq_desc[irq].handler->handle(irq, cpu, &regs);
+ restore_stack();

/*
* This should be conditional: we should really get

I am writing this with the patch applyed ;-).

I am not sure if this is really what I want (I am just very happy that the
kernel doesn' t crash though ;-). I want that every irq handler have a
page of ram to use. I want to avoid that 16 nested interrupt could run
all on the same 8kbyte kernel stack.

Andrea[s] Arcangeli

PS. Really time to sleep... (goodnight).

-
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.altern.org/andrebalsa/doc/lkml-faq.html