: Linux-2.5.1-pre5
Hi Linus,
I just got the following oops when running 2.5.1-pre5. pre5 has been running otherwise stable for a few days. The only oddity I'm seeing is that ever since I upgraded to -pre5 from -pre1 I'm seeing
keyboard: Timeout - AT keyboard not present?(f4)
keyboard: Timeout - AT keyboard not present?(f4)
whenever I (re)start X. This never happened before.
Regards,
Udo.
ksymoops 2.4.1 on i686 2.5.1-pre5. Options used
-V (default)
-k /proc/ksyms (default)
-l /proc/modules (default)
-o /lib/modules/2.5.1-pre5/ (default)
-m /boot/System.map-2.5.1-pre5 (specified)
Unable to handle kernel NULL pointer dereference at virtual address 00000028 c01d0538 *pde = 00000000 Oops: 0000 CPU: 0 EIP: 0010:[<c01d0538>] Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010286
eax: 00000000 ebx: c1be11c0 ecx: 00000030 edx: 00000200
esi: 00000001 edi: c1be14c0 ebp: c10688ac esp: cac1bdb0
ds: 0018 es: 0018 ss: 0018
Process kdeinit (pid: 223, stackpage=cac1b000)
Stack: 00000030 00000001 c1be11c0 00000000 c0135af7 00000001 c1be11c0 c010820c
0000000b c1101ad4 000001f0 00000020 00000a0e c010a238 c1101ad4 c1be14c0
c10688ac c1be14c0 c0135b5c c1be14c0 000001f0 00000018 00000018 ffffff0b
Call Trace: [<c0135af7>] [<c010820c>] [<c010a238>] [<c0135b5c>] [<c01346bc>]
[<c012c328>] [<c012c5e0>] [<c012c64c>] [<c012cf51>] [<c012d19c>] [<c012d230>]
[<c013f8d3>] [<c02638d5>] [<c022177f>] [<c013fb03>] [<c013fed2>] [<c01c9b5f>]
[<c0106d1b>]
Code: 8b 48 28 66 c1 ea 09 0f b7 d2 0f af 53 04 89 10 0f b7 53 0cQuoted 1 line
Quoted 1 line
EIP;
c01d0538<submit_bh+48/d0> <=====
Trace; c0135af7 <sync_page_buffers+97/b0>
Trace; c010820c <do_IRQ+7c/b0>
Trace; c010a238 <call_do_IRQ+5/d>
Trace; c0135b5c <try_to_free_buffers+4c/100>
Trace; c01346bc <try_to_release_page+1c/50>
Trace; c012c328 <shrink_cache+188/310>
Trace; c012c5e0 <shrink_caches+50/90>
Trace; c012c64c <try_to_free_pages+2c/50>
Trace; c012cf51 <balance_classzone+51/190>
Trace; c012d19c <__alloc_pages+10c/190>
Trace; c012d230 <__get_free_pages+10/20>
Trace; c013f8d3 <__pollwait+33/90>
Trace; c02638d5 <unix_poll+25/a0>
Trace; c022177f <sock_poll+1f/30>
Trace; c013fb03 <do_select+f3/200>
Trace; c013fed2 <sys_select+292/4a0>
Trace; c01c9b5f <keyboard_interrupt+f/20>
Trace; c0106d1b <system_call+33/38>
Code; c01d0538 <submit_bh+48/d0>
0000000000000000 <_EIP>:
Code; c01d0538 <submit_bh+48/d0> <=====
0: 8b 48 28 mov 0x28(%eax),%ecx <=====
Code; c01d053b <submit_bh+4b/d0>
3: 66 c1 ea 09 shr $0x9,%dx
Code; c01d053f <submit_bh+4f/d0>
7: 0f b7 d2 movzwl %dx,%edx
Code; c01d0542 <submit_bh+52/d0>
a: 0f af 53 04 imul 0x4(%ebx),%edx
Code; c01d0546 <submit_bh+56/d0>
e: 89 10 mov %edx,(%eax)
Code; c01d0548 <submit_bh+58/d0>
10: 0f b7 53 0c movzwl 0xc(%ebx),%edx
Quoted 2 lines · Udo A. Steinberg
keyboard: Timeout - AT keyboard not present?(f4)
keyboard: Timeout - AT keyboard not present?(f4)
I see those too. (the way it's written reminds me of a joke about silly MSDOS startup messages.. keyboard not detected, press F1 to continue:)
-Mike
Quoted 7 lines · Mike Galbraith
Quoted 2 lines · Udo A. Steinberg
keyboard: Timeout - AT keyboard not present?(f4)
keyboard: Timeout - AT keyboard not present?(f4)I see those too. (the way it's written reminds me of a joke about silly MSDOS startup messages.. keyboard not detected, press F1 to continue:)
Revert changes in drivers/char/pc_keyb.c (or, better yet, entire BKL removal from ->release() patch).
Quoted 12 lines · Udo A. Steinberg
Unable to handle kernel NULL pointer dereference at virtual address 00000028 c01d0538 *pde = 00000000 Oops: 0000 CPU: 0 EIP: 0010:[<c01d0538>] Not taintedUsing defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010286
eax: 00000000 ebx:c1be11c0ecx: 00000030 edx: 00000200
esi: 00000001 edi:c1be14c0ebp:c10688acesp:cac1bdb0
ds: 0018 es: 0018 ss: 0018
Process kdeinit (pid: 223, stackpage=cac1b000)
Quoted 10 lines
Stack: 00000030 00000001 c1be11c0 00000000 c0135af7 00000001 c1be11c0 c010820c 0000000b c1101ad4 000001f0 00000020 00000a0e c010a238 c1101ad4 c1be14c0 c10688ac c1be14c0 c0135b5c c1be14c0 000001f0 00000018 00000018 ffffff0b Call Trace: [<c0135af7>] [<c010820c>] [<c010a238>] [<c0135b5c>] [<c01346bc>] [<c012c328>] [<c012c5e0>] [<c012c64c>] [<c012cf51>] [<c012d19c>] [<c012d230>] [<c013f8d3>] [<c02638d5>] [<c022177f>] [<c013fb03>] [<c013fed2>] [<c01c9b5f>] [<c0106d1b>] Code: 8b 48 28 66 c1 ea 09 0f b7 d2 0f af 53 04 89 10 0f b7 53 0cQuoted 1 line
Quoted 1 line
EIP;
c01d0538<submit_bh+48/d0> <=====
This should fix it.
Signature · 1 line
Jens Axboe
bio.c renamed from opt/kernel/linux-2.5.1-pre5/fs/bio.c +11 -11
@@ -35,7 +35,7 @@#include <asm/uaccess.h> kmem_cache_t *bio_cachep; -static spinlock_t __cacheline_aligned bio_lock = SPIN_LOCK_UNLOCKED; +static spinlock_t __cacheline_aligned_in_smp bio_lock = SPIN_LOCK_UNLOCKED;static struct bio *bio_pool;static DECLARE_WAIT_QUEUE_HEAD (bio_pool_wait);static DECLARE_WAIT_QUEUE_HEAD (biovec_pool_wait);
@@ -74,7 +74,7 @@struct bio *bio;if ((bio = bio_pool)) { -BUG_ON (bio_pool_free <=0 ); +BIO_BUG_ON (bio_pool_free <=0 ); bio_pool = bio->bi_next; bio->bi_next =NULL ; bio_pool_free--;
@@ -90,7 +90,7 @@spin_lock_irqsave (&bio_lock, flags); bio =__bio_pool_get (); -BUG_ON (!bio && bio_pool_free); +BIO_BUG_ON (!bio && bio_pool_free);spin_unlock_irqrestore (&bio_lock, flags);return bio;
@@ -121,8 +121,7 @@ } } -#define BIO_CAN_WAIT (gfp_mask) \ - (((gfp_mask) & (__GFP_WAIT | __GFP_IO)) == (__GFP_WAIT | __GFP_IO)) +#define BIO_CAN_WAIT (gfp_mask) ((gfp_mask) & __GFP_WAIT)static inline struct bio_vec *bvec_alloc (int gfp_mask,int nr,int *idx) {
@@ -198,13 +197,15 @@ {struct biovec_pool *bp = &bvec_list[bio->bi_max]; -BUG_ON (bio->bi_max >=BIOVEC_NR_POOLS ); +BIO_BUG_ON (bio->bi_max >=BIOVEC_NR_POOLS );/* * cloned bio doesn't own the veclist */ -if (!(bio->bi_flags & (1 <<BIO_CLONED ))) +if (!(bio->bi_flags & (1 <<BIO_CLONED ))) {kmem_cache_free (bp->bp_cachep, bio->bi_io_vec); +wake_up_nr (&bp->bp_wait,1 ); + }bio_pool_put (bio); }
@@ -212,13 +213,13 @@inline void bio_init (struct bio *bio) { bio->bi_next =NULL ; -atomic_set (&bio->bi_cnt,1 ); bio->bi_flags =0 ; bio->bi_rw =0 ; bio->bi_vcnt =0 ; bio->bi_idx =0 ; bio->bi_size =0 ; bio->bi_end_io =NULL ; +atomic_set (&bio->bi_cnt,1 ); }static inline struct bio *__bio_alloc (int gfp_mask,bio_destructor_t *dest)
@@ -314,14 +315,13 @@**/ void bio_put (struct bio *bio) { -BUG_ON (!atomic_read (&bio->bi_cnt)); +BIO_BUG_ON (!atomic_read (&bio->bi_cnt));/* * last put frees it */ if (atomic_dec_and_test (&bio->bi_cnt)) { -BUG_ON (bio->bi_next); - +BIO_BUG_ON (bio->bi_next);bio_free (bio); } }
Quoted 1 line · Jens Axboe
This should fix it.
So, what was the actual problem?
--
David Hansen
[email protected]
Quoted 5 lines · David C. Hansen
Quoted 1 line · Jens Axboe
This should fix it.
So, what was the actual problem?
bio_alloc() not waiting on the reserved pool for free entries, even though __GFP_WAIT was set. No need for __GFP_IO in that case too.
There were two hunks with actual code changes in there, did you read the patch at all? :-)
Signature · 1 line
Jens Axboe