Re: bcache: fix oops bug in cache_set_flush

Hi Liequan,

> 2024年11月13日 14:25，liequan che <[email protected]> 写道：
> 
> Signed-off-by: cheliequan <[email protected]>
> 
>   If the bcache cache disk contains damaged btree data,
> when the bcache cache disk partition is directly operated,
> the system-udevd service is triggered to call the bcache-register
> program to register the bcache device,resulting in kernel oops.
> 

What is the kernel version ? 

Interesting that why the btree node checking code during registration doesn’t cache the meta data error.



> crash> bt
> PID: 7773     TASK: ffff49cc44d69340  CPU: 57   COMMAND: "kworker/57:2"
> #0 [ffff800046373800] machine_kexec at ffffbe5039eb54a8
> #1 [ffff8000463739b0] __crash_kexec at ffffbe503a052824
> #2 [ffff8000463739e0] crash_kexec at ffffbe503a0529cc
> #3 [ffff800046373a60] die at ffffbe5039e9445c
> #4 [ffff800046373ac0] die_kernel_fault at ffffbe5039ec698c
> #5 [ffff800046373af0] __do_kernel_fault at ffffbe5039ec6a38
> #6 [ffff800046373b20] do_page_fault at ffffbe503ac76ba4
> #7 [ffff800046373b70] do_translation_fault at ffffbe503ac76ebc
> #8 [ffff800046373b90] do_mem_abort at ffffbe5039ec68ac
> #9 [ffff800046373bc0] el1_abort at ffffbe503ac669bc
> #10 [ffff800046373bf0] el1_sync_handler at ffffbe503ac671d4
> #11 [ffff800046373d30] el1_sync at ffffbe5039e82230
> #12 [ffff800046373d50] cache_set_flush at ffffbe50121fa4c4 [bcache]
> #13 [ffff800046373da0] process_one_work at ffffbe5039f5af68
> #14 [ffff800046373e00] worker_thread at ffffbe5039f5b3c4
> #15 [ffff800046373e50] kthread at ffffbe5039f634b8
> crash> dis cache_set_flush+0x94
> 0xffffbe50121fa4c8 <cache_set_flush+148>:       str     x23, [x20, #512]
> 
> ---
> drivers/md/bcache/super.c | 16 ++++++++++------
> 1 file changed, 10 insertions(+), 6 deletions(-)
> diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
> index fd97730479d8..8a41dfcf9fb6 100644
> --- a/drivers/md/bcache/super.c
> +++ b/drivers/md/bcache/super.c
> @@ -1741,8 +1741,10 @@ static void cache_set_flush(struct closure *cl)
>       if (!IS_ERR_OR_NULL(c->gc_thread))
>               kthread_stop(c->gc_thread);
> 
> -       if (!IS_ERR(c->root))
> -               list_add(&c->root->list, &c->btree_cache);
> +       if (!IS_ERR_OR_NULL(c->root)) {
> +               if (!list_empty(&c->root->list))
> +                       list_add(&c->root->list, &c->btree_cache);
> +       }
> 
>       /*
>        * Avoid flushing cached nodes if cache set is retiring
> @@ -1750,10 +1752,12 @@ static void cache_set_flush(struct closure *cl)
>        */
>       if (!test_bit(CACHE_SET_IO_DISABLE, &c->flags))
>               list_for_each_entry(b, &c->btree_cache, list) {
> -                       mutex_lock(&b->write_lock);
> -                       if (btree_node_dirty(b))
> -                               __bch_btree_node_write(b, NULL);
> -                       mutex_unlock(&b->write_lock);
> +                       if (!IS_ERR_OR_NULL(b)) {

The above check is not safe. 



> +                               mutex_lock(&b->write_lock);
> +                               if (btree_node_dirty(b))
> +                                       __bch_btree_node_write(b, NULL);
> +                               mutex_unlock(&b->write_lock);
> +                       }
>               }
> 
>       if (ca->alloc_thread)
> --
> 2.33.0


Thanks.

Coly Li