All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] staging: comedi: fix bug destroying subdevice files after parent
@ 2014-01-07 12:38 Ian Abbott
  0 siblings, 0 replies; only message in thread
From: Ian Abbott @ 2014-01-07 12:38 UTC (permalink / raw)
  To: driverdev-devel; +Cc: Greg Kroah-Hartman, Ian Abbott, Bernd Porr

`comedi_free_board_dev()` is called (via `comedi_auto_unconfig()` -->
`comedi_release_hardware_device()`) when an auto-configured comedi
device is removed.  This destroys the main sysfs class device and then
calls `comedi_device_cleanup()` to clean up the comedi device.  For
comedi devices that have comedi subdevices that asynchronous commands,
the clean up involves destroying the sysfs class devices associated with
those subdevices.

There is a bug in the above sequence because the sysfs class devices
associated with the comedi subdevices are children of the sysfs class
device associated with the main comedi device.  Therefore they will have
been automatically destroyed when the main sysfs class device is
destroyed.  When they are destroyed again as part of the clean-up, they
will not be found, leading to a warning and a stack trace similar to
this:

------------[ cut here ]------------
WARNING: CPU: 1 PID: 1213 at fs/sysfs/group.c:214
sysfs_remove_group+0x4e/0xa7()
sysfs group ffffffff817504c0 not found for kobject 'comedi4_subd0'
Modules linked in: nfsd auth_rpcgss oid_registry exportfs nfs_acl lockd
bridge stp llc sunrpc fuse binfmt_misc cpufreq_userspace sr_mod
snd_hda_codec_analog cdrom powernow_k8 kvm_amd kvm amplc_pci230(C)
8255(C) comedi(C) pcmcia xhci_hcd ehci_pci pcmcia_core ohci_pci ohci_hcd
ehci_hcd usbcore snd_hda_intel snd_hda_codec snd_pcm k8temp
snd_page_alloc 8139too snd_timer snd soundcore mii usb_common forcedeth
pata_amd
CPU: 1 PID: 1213 Comm: kworker/u4:6 Tainted: G         C
3.13.0-rc5-ija1+ #20
Hardware name: System manufacturer System Product Name/M2N-E, BIOS ASUS
M2N-E ACPI BIOS Revision 5001 03/23/2010
Workqueue: sysfsd sysfs_schedule_callback_work
 0000000000000000 ffff8800bf17fb38 ffffffff814672ce ffff8800bf17fb80
 ffff8800bf17fb70 ffffffff8103470b ffffffff8114f780 0000000000000000
 ffffffff817504c0 ffff8800bf39f410 ffff880139b68670 ffff8800bf17fbd0
Call Trace:
 [<ffffffff814672ce>] dump_stack+0x45/0x56
 [<ffffffff8103470b>] warn_slowpath_common+0x7a/0x93
 [<ffffffff8114f780>] ? sysfs_remove_group+0x4e/0xa7
 [<ffffffff8103476b>] warn_slowpath_fmt+0x47/0x49
 [<ffffffff8114e92d>] ? sysfs_get_dirent_ns+0x5e/0x66
 [<ffffffff8114f780>] sysfs_remove_group+0x4e/0xa7
 [<ffffffff8132aac0>] dpm_sysfs_remove+0x37/0x3b
 [<ffffffff81323781>] device_del+0x3e/0x173
 [<ffffffff813238c3>] device_unregister+0xd/0x18
 [<ffffffff8132392e>] device_destroy+0x33/0x37
 [<ffffffffa0212086>] comedi_free_subdevice_minor+0x80/0x92 [comedi]
 [<ffffffffa02128bb>] comedi_device_detach+0x79/0x152 [comedi]
 [<ffffffffa020f223>] comedi_device_cleanup+0x36/0x57 [comedi]
 [<ffffffffa020f275>] comedi_free_board_dev+0x31/0x3c [comedi]
 [<ffffffffa0211f2a>] comedi_release_hardware_device+0x5a/0x73 [comedi]
 [<ffffffffa0212547>] comedi_auto_unconfig+0xe/0x10 [comedi]
 [<ffffffffa021357c>] comedi_pci_auto_unconfig+0x10/0x12 [comedi]
 [<ffffffff811d2335>] pci_device_remove+0x40/0x8a
 [<ffffffff813261d0>] __device_release_driver+0x84/0xda
 [<ffffffff81326244>] device_release_driver+0x1e/0x2b
 [<ffffffff811cdcb5>] pci_stop_bus_device+0x44/0x87
 [<ffffffff811cdde2>] pci_stop_and_remove_bus_device+0xd/0x18
 [<ffffffff811d3f3d>] remove_callback+0x20/0x2f
 [<ffffffff8114d1f7>] sysfs_schedule_callback_work+0xf/0x70
 [<ffffffff81049498>] process_one_work+0x1d6/0x34c
 [<ffffffff81049a5f>] worker_thread+0x1cf/0x2b5
 [<ffffffff81049890>] ? rescuer_thread+0x258/0x258
 [<ffffffff8104e0e6>] kthread+0xd6/0xde
 [<ffffffff8104e010>] ? kthread_create_on_node+0x160/0x160
 [<ffffffff81472cbc>] ret_from_fork+0x7c/0xb0
 [<ffffffff8104e010>] ? kthread_create_on_node+0x160/0x160
---[ end trace 94722aa2936a7adf ]---

To correct the bug, rearrange `comedi_free_board_dev()` to destroy the
main sysfs class device *after* the clean-up operation.

Thanks to Bernd Porr for finding the bug and his initial attempt to fix
it.

Reported-by: Bernd Porr <mail@berndporr.me.uk>
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Cc: Bernd Porr <mail@berndporr.me.uk>
---
 drivers/staging/comedi/comedi_fops.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/staging/comedi/comedi_fops.c b/drivers/staging/comedi/comedi_fops.c
index d7f63c4..c22c617 100644
--- a/drivers/staging/comedi/comedi_fops.c
+++ b/drivers/staging/comedi/comedi_fops.c
@@ -153,11 +153,11 @@ static struct comedi_device *comedi_clear_board_minor(unsigned minor)
 static void comedi_free_board_dev(struct comedi_device *dev)
 {
 	if (dev) {
+		comedi_device_cleanup(dev);
 		if (dev->class_dev) {
 			device_destroy(comedi_class,
 				       MKDEV(COMEDI_MAJOR, dev->minor));
 		}
-		comedi_device_cleanup(dev);
 		comedi_dev_put(dev);
 	}
 }
-- 
1.8.5.2

_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

^ permalink raw reply related	[flat|nested] only message in thread

only message in thread, other threads:[~2014-01-07 12:38 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-01-07 12:38 [PATCH] staging: comedi: fix bug destroying subdevice files after parent Ian Abbott

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.