2.5.41-mm1 oops on boot (EIP at kmem_cache_alloc)

2.5.41-mm1 oops on boot (EIP at kmem_cache_alloc)

[Steven Cole]

Greetings,

I got an oops when booting 2.5.41-mm1 on my dual p3.
More configuration information available if needed.

I had to copy this down by hand, so only a minimal amount of information
was saved.  If more data is needed, I can repeat the boot.

EIP is at kmem_cache_alloc+0x18/0x50

Call Trace:
call_console_drivers+0xeb/0x100
kmem_cache_create+0x6f/0x560
release_console_sem+0x62/0xe0
init+0x51/0x1d0
init+0x0/0x1d0
kernel_thread_helper+0x5/0x10

Code: 8b 02 85 c0 74 12 c7 42 0c 00 00 00 48 89 02 8b 44 02 10

I was able to boot and run 2.5.41-bk2 on this same machine.

Steven





--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: 2.5.41-mm1 oops on boot (EIP at kmem_cache_alloc)

[Andrew Morton]

Steven Cole wrote:
> 
> Greetings,
> 
> I got an oops when booting 2.5.41-mm1 on my dual p3.

Manfred sent through an update - don't know if it will
fix this though:


--- 2.5.41/mm/slab.c~slab-split-10-list_for_each_fix	Tue Oct  8 15:40:52 2002
+++ 2.5.41-akpm/mm/slab.c	Tue Oct  8 15:40:52 2002
@@ -461,7 +461,7 @@ static kmem_cache_t cache_cache = {
 static struct semaphore	cache_chain_sem;
 static rwlock_t cache_chain_lock = RW_LOCK_UNLOCKED;
 
-#define cache_chain (cache_cache.next)
+struct list_head cache_chain;
 
 /*
  * chicken and egg problem: delay the per-cpu array allocation
@@ -617,6 +617,7 @@ void __init kmem_cache_init(void)
 
 	init_MUTEX(&cache_chain_sem);
 	INIT_LIST_HEAD(&cache_chain);
+	list_add(&cache_cache.next, &cache_chain);
 
 	cache_estimate(0, cache_cache.objsize, 0,
 			&left_over, &cache_cache.num);
@@ -2093,10 +2094,10 @@ static void *s_start(struct seq_file *m,
 	down(&cache_chain_sem);
 	if (!n)
 		return (void *)1;
-	p = &cache_cache.next;
+	p = cache_chain.next;
 	while (--n) {
 		p = p->next;
-		if (p == &cache_cache.next)
+		if (p == &cache_chain)
 			return NULL;
 	}
 	return list_entry(p, kmem_cache_t, next);
@@ -2107,9 +2108,9 @@ static void *s_next(struct seq_file *m, 
 	kmem_cache_t *cachep = p;
 	++*pos;
 	if (p == (void *)1)
-		return &cache_cache;
-	cachep = list_entry(cachep->next.next, kmem_cache_t, next);
-	return cachep == &cache_cache ? NULL : cachep;
+		return list_entry(cache_chain.next, kmem_cache_t, next);
+	return cachep->next.next == &cache_chain ? NULL
+		: list_entry(cachep->next.next, kmem_cache_t, next);
 }
 
 static void s_stop(struct seq_file *m, void *p)

.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: 2.5.41-mm1 oops on boot (EIP at kmem_cache_alloc)

[Steven Cole]

On Wed, 2002-10-09 at 10:07, Andrew Morton wrote:
> Steven Cole wrote:
> > 
> > Greetings,
> > 
> > I got an oops when booting 2.5.41-mm1 on my dual p3.
> 
> Manfred sent through an update - don't know if it will
> fix this though:
> 
> 
[patch snipped]

Yep, that worked.  Thanks.  I also got the same debug message
for mm/slab.c:1374 which I reported for 2.5.41-bk2, plus the usual
set of "bad: scheduling while atomic!" messages on boot up.

Steven

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/