2002-12-31 10:26:16

by Aniruddha M Marathe

[permalink] [raw]
Subject: 5.53 mm2 oops during LMbench.

Oops came when Lmbench was calculating memory load latency.

Couldn't capture the entire call trace. Here is the last screen.

3ef4c>] shrink_list+0x3c/0x660
[<c0109cea>] apic_timer_interrupt+0x1a/0x20
[<c013dfca>] __pagevec_release+0x1a/0x30
[<c013f7a8>] shrink_cache+0x238/0x460
[<c01401b3>] shrink_zone+0x83/0xb0
[<c0140250>] shrink_caches+0x70/0xa0
[<c0140309>] try_to_free_pages+0x89/0xc0
[<c01382a7>] __alloc_pages+0x1f7/0x2c0
[<c013839a>] __get_free_pages+0x2a/0x70
[<c013b6e1>] cache_grow+0x141/0x380
[<c013ba51>] __cache_alloc_refill+0x131/0x440
[<c013bda4>] cache_alloc_refill+0x44/0x60
[<c013c316>] kmem_cache_alloc+0x56/0x100
[<c01485b5>] pgtable_add_rmap+0x55/0x80
[<c01416a1>] pte_alloc_map+0xe1/0x180
[<c0148f61>] pte_chain_alloc+0x41/0x60
[<c0141ca6>] copy_page_range+0x466/0x6c0
[<c011ce4d>] mm_init+0xdd/0x120
[<c011d37b>] copy_mm+0x3ab/0x4b0
[<c011df55>] copy_process+0x5c5/0xbd0
[<c011e5e7>] do_fork+0x87/0x140
[<c01076c9>] sys_fork+0x19/0x30
[<c0109323>] syscall_call+0x7/0xb


2002-12-31 10:51:31

by Andrew Morton

[permalink] [raw]
Subject: Re: 5.53 mm2 oops during LMbench.

Aniruddha M Marathe wrote:
>
> Oops came when Lmbench was calculating memory load latency.
>
> Couldn't capture the entire call trace. Here is the last screen.
>
> 3ef4c>] shrink_list+0x3c/0x660
> [<c0109cea>] apic_timer_interrupt+0x1a/0x20
> [<c013dfca>] __pagevec_release+0x1a/0x30
> [<c013f7a8>] shrink_cache+0x238/0x460
> [<c01401b3>] shrink_zone+0x83/0xb0
> [<c0140250>] shrink_caches+0x70/0xa0
> [<c0140309>] try_to_free_pages+0x89/0xc0
> [<c01382a7>] __alloc_pages+0x1f7/0x2c0
> [<c013839a>] __get_free_pages+0x2a/0x70
> [<c013b6e1>] cache_grow+0x141/0x380
> [<c013ba51>] __cache_alloc_refill+0x131/0x440
> [<c013bda4>] cache_alloc_refill+0x44/0x60
> [<c013c316>] kmem_cache_alloc+0x56/0x100
> [<c01485b5>] pgtable_add_rmap+0x55/0x80
> [<c01416a1>] pte_alloc_map+0xe1/0x180
> [<c0148f61>] pte_chain_alloc+0x41/0x60

That's probably a `scheduling while atomic' warning coming out
of pte_chain_alloc(). It has a bug.