Re: 2.6.0-test9 and sleeping function called from invalid context

[Andrew Morton]

Arkadiusz Miskiewicz <arekm@xxxxxxxxxxxxx> wrote:
>
> Debug: sleeping function called from invalid context at 
>  include/asm/semaphore.h:119
>  in_atomic():1, irqs_disabled():0
>  Call Trace:
>   [<c011dfbb>] __might_sleep+0xab/0xf0
>   [<c019ed91>] inode_doinit_with_dentry+0x41/0x5d0
>   [<cf9b91b0>] bm_fill_super+0x0/0x40 [binfmt_misc]
>   [<c0170904>] d_instantiate+0x64/0x80
>   [<cf9b8f30>] bm_register_write+0x150/0x1d0 [binfmt_misc]
>   [<cf9b8de0>] bm_register_write+0x0/0x1d0 [binfmt_misc]
>   [<c01586ea>] vfs_write+0xca/0x140
>   [<c01587ff>] sys_write+0x3f/0x60
>   [<c010b37f>] syscall_call+0x7/0xb

Not nice.

fs/binfmt_misc.c:bm_register_write() is doing d_instantiate()
under a write_lock(), and the selinux d_instantiate() hook is
performing sleeping functions.

Now, we can just move the d_instantiate() call to outside the lock:

--- 25/fs/binfmt_misc.c~bm_register_write-locking-fix	2003-10-25 18:39:45.000000000 -0700
+++ 25-akpm/fs/binfmt_misc.c	2003-10-25 18:40:29.000000000 -0700
@@ -529,8 +529,8 @@ static ssize_t bm_register_write(struct 
 	inode->u.generic_ip = e;
 	inode->i_fop = &bm_entry_operations;
 
-	write_lock(&entries_lock);
 	d_instantiate(dentry, inode);
+	write_lock(&entries_lock);
 	list_add(&e->list, &entries);
 	write_unlock(&entries_lock);
 

which I think is OK, but the wider question would be: is the SELinux
d_instantiate callout allowed to sleep?  A quick audit seems to indicate
that it's OK, but only by luck I think.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/