Hi,

I am porting a USB-to-serial & parallel driver to 2.6.17.7

I can load the driver, but couldn't unload it.
(But It is working fine in 2.6.19 , 2.6.18 &FC4)

The following are the log messages:

make: *** [unload] Segmentation fault
kernel BUG at kernel/workqueue.c:110!
invalid opcode: 0000 [#1]

Modules linked in: mos7715 savage drm usbserial nfsd exportfs lockd nfs_acl irnet ppp_generic slhc irtty_sir sir_dev ircomm_tty ircomm autofs4 hidp rfcomm l2cap bluetooth sunrpc usbnet dm_mirror dm_mod video button battery ac ipv6 parport_pc parport floppy nvram uhci_hcd ehci_hcd snd_via82xx gameport snd_ac97_codec snd_ac97_bus snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_pcm_oss snd_mixer_oss snd_pcm i2c_prosavage i2c_algo_bit via_ircc snd_timer snd_page_alloc snd_mpu401_uart snd_rawmidi snd_seq_device 8139too irda crc_ccitt i2c_viapro i2c_core snd soundcore 8139cp mii sata_via libata sd_mod scsi_mod

Call Trace:
<cebdb765> usb_serial_disconnect+0x54/0x94 [usbserial] <c026e058> usb_unbind_interface+0x34/0x6a
<c0235f8c> __device_release_driver+0x60/0x78 <c023625e> driver_detach+0x97/0xc5
<c023578c> bus_remove_driver+0x5a/0x78 <c023636c> driver_unregister+0x8/0x13
<c026df4c> usb_deregister+0x7f/0x87 <cec4821a> moschip7715_exit+0x1e/0x40 [mos7715]
<c012f4b1> sys_delete_module+0x15b/0x182 <c02ef3e9> do_page_fault+0x221/0x584
<c01029ef> sysenter_past_esp+0x54/0x75
Code: 58 89 d8 5b 5e 5f e9 92 7d 1c 00 8b 48 14 89 c2 8b 01 eb ae 53 89 c3 0f ba 2a 00 19 c0 31 c9 85 c0 75 1c 8d 42 04 39 42 04 74 08 <0f> 0b 6e 00 6d d5 30 c0 8b 03 e8 88 ff ff ff b9 01 00 00 00 5b
EIP: [<c01268f5>] queue_work+0x17/0x2f SS:ESP 0068:c4185ee8

Any help on this is appriciated.


Regards,
-Mahipal

Have a look at the following articles:

http://lwn.net/Articles/211279/
http://lwn.net/Articles/213149/

I wish I could add more but I'm just learning about this myself at this time. Hope this helps shed some light on the problem.

It sounds like the workqueue was not yet empty when the module was unloaded.

As we all know from painful experience, "achieving a clean shutdown" of any application can be very tricky when asynchronous processing is involved. Since there are several ways that you could have gotten into this mess, there's no single way to get out of it.