Re: [PATCH 1/3] lockdep: Move lock events under lockdep recursion protection

[Jens Axboe <jens.axboe@oracle.com>]

On Wed, Mar 10 2010, Frederic Weisbecker wrote:
> On Tue, Mar 09, 2010 at 09:35:37AM +0100, Jens Axboe wrote:
> > Which, btw, throws a new lockdep warning fest:
> > 
> > [   42.247718] scsi7 : ioc0: LSISAS1068E B3, FwRev=011b0300h, Ports=1, MaxQ=483, IRQ=26
> > [   42.281125] BUG: key ffff880c7cb75250 not in .data!
> > [   42.288346] ------------[ cut here ]------------
> > [   42.294490] WARNING: at kernel/lockdep.c:2706 lockdep_init_map+0x545/0x5f0()
> > [   42.304013] Hardware name: QSSC-S4R
> > [   42.309674] Modules linked in: hid_apple usbhid ehci_hcd uhci_hcd usbcore nls_base mptsas(+) mptscsih mptbase scsi_transport_sas igb sg sr_mod cdrom
> > [   42.332072] Pid: 5197, comm: modprobe Not tainted 2.6.34-rc1 #176
> > [   42.340597] Call Trace:
> > [   42.345335]  [<ffffffff8107fb8d>] ? is_module_address+0x2d/0x60
> > [   42.353670]  [<ffffffff81074aa5>] ? lockdep_init_map+0x545/0x5f0
> > [   42.362154]  [<ffffffff81044648>] warn_slowpath_common+0x78/0xd0
> > [   42.371913]  [<ffffffff810446af>] warn_slowpath_null+0xf/0x20
> > [   42.380109]  [<ffffffff81074aa5>] lockdep_init_map+0x545/0x5f0
> 
> 
> 
> This doesn't look related to my patch, it also happen in -rc1 right?

It's not related, it does happen in -rc1 plain as well.

> > [   42.388402]  [<ffffffff81156636>] ? sysfs_new_dirent+0x76/0x120
> 
> 
> This calls:
> 
> #define sysfs_dirent_init_lockdep(sd)				\
> do {								\
> 	struct attribute *attr = sd->s_attr.attr;		\
> 	struct lock_class_key *key = attr->key;			\
> 	if (!key)						\
> 		key = &attr->skey;				\
> 								\
> 	lockdep_init_map(&sd->dep_map, "s_active", key, 0);	\
> } while(0)
> 
> 
> Looks like the struct lock_class_key is dynamically allocated
> and lockdep doesn't like that.

I saw someone else report the same warning, but from a different path
than mine. I'll try and bisect it.

-- 
Jens Axboe