lkml.org 
[lkml]   [2002]   [Sep]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
SubjectRe: [PATCH] BUG(): sched.c: Line 944
From
Date
On Tue, 2002-09-17 at 04:12, Robert Love wrote:

> + if (unlikely(!kernel_locked()))
> + tsk->lock_depth = -2;

Eh, this should be likely()...

Robert Love

diff -urN linux-2.5.35/include/asm-i386/hardirq.h linux/include/asm-i386/hardirq.h
--- linux-2.5.35/include/asm-i386/hardirq.h Sun Sep 15 22:18:46 2002
+++ linux/include/asm-i386/hardirq.h Tue Sep 17 03:20:00 2002
@@ -77,7 +77,8 @@
#define irq_enter() (preempt_count() += HARDIRQ_OFFSET)

#if CONFIG_PREEMPT
-# define in_atomic() (preempt_count() != kernel_locked())
+# define in_atomic() \
+ ((preempt_count() & ~PREEMPT_ACTIVE) != (current->lock_depth != -1))
# define IRQ_EXIT_OFFSET (HARDIRQ_OFFSET-1)
#else
# define in_atomic() (preempt_count() != 0)
diff -urN linux-2.5.35/kernel/exit.c linux/kernel/exit.c
--- linux-2.5.35/kernel/exit.c Tue Sep 17 03:18:53 2002
+++ linux/kernel/exit.c Tue Sep 17 03:55:12 2002
@@ -637,6 +637,21 @@
preempt_disable();
if (current->exit_signal == -1)
release_task(current);
+
+ /*
+ * This little bit of genius comes from the twisted mind of Linus.
+ * We need exit() to be atomic but we also want a debugging check
+ * in schedule() to whine if we are atomic. The wickedness is in
+ * these rules:
+ * - task->lock_depth = -2 means "validly nonpreemptable"
+ * - task->lock_depth = -1 means "BKL not held"
+ * - task->lock_depth >= 0 means "BKL held"
+ * release_kernel_lock and kernel_locked() check >=0, and
+ * in_atomic() checks != -1... the "fake BKL" will "cancel out"
+ * the preempt_disable() above and the world is happy.
+ */
+ if (likely(!kernel_locked()))
+ tsk->lock_depth = -2;
schedule();
BUG();
/*
diff -urN linux-2.5.35/kernel/sched.c linux/kernel/sched.c
--- linux-2.5.35/kernel/sched.c Sun Sep 15 22:18:24 2002
+++ linux/kernel/sched.c Tue Sep 17 03:50:04 2002
@@ -940,8 +940,10 @@
struct list_head *queue;
int idx;

- if (unlikely(in_atomic()))
- BUG();
+ if (unlikely(in_atomic())) {
+ printk(KERN_ERR "error: scheduling while non-atomic!\n");
+ dump_stack();
+ }

#if CONFIG_DEBUG_HIGHMEM
check_highmem_ptes();
@@ -959,7 +961,7 @@
* if entering off of a kernel preemption go straight
* to picking the next task.
*/
- if (unlikely(preempt_count() & PREEMPT_ACTIVE))
+ if (unlikely(preempt_count() == PREEMPT_ACTIVE))
goto pick_next_task;

switch (prev->state) {
-
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/

\
 
 \ /
  Last update: 2005-03-22 13:28    [from the cache]
©2003-2011 Jasper Spaans. Advertise on this site