On Sun, 12 Jun 2005, Martin Kessler wrote: > Well here it is, 'control queue full' appeared around 22:48 I took a > snapshot from my logfile and have attached it. its 5MB so I compressed > it, hope thats ok ;-) If you need data prior to 21:00 as well, let me know.
This helps. The log clearly shows rapidly repeated control URBs being sent. They arise because some program is doing an ioctl to the hiddev driver (hiddev_ioctl() gets called), resulting in a report being submitted via a control URB (hid_submit_ctrl() gets called). Here's a sample of one occurrence; this gets repeated over and over in the log: Jun 11 22:00:46 devel kernel: [<c02c1f84>] usb_submit_urb+0x357/0x3ea Jun 11 22:00:46 devel kernel: [<c02cda20>] hid_submit_ctrl+0x13d/0x1dd Jun 11 22:00:46 devel kernel: [<c02ce5f4>] hid_wait_io+0xb0/0xbd Jun 11 22:00:46 devel kernel: [<c02ce29c>] hid_submit_report+0x211/0x4b9 Jun 11 22:00:46 devel kernel: [<c011aac8>] default_wake_function+0x0/0xc Jun 11 22:00:46 devel kernel: [<c02d00ca>] hiddev_ioctl+0x53a/0xa07 Jun 11 22:00:46 devel kernel: [<c018b702>] do_select+0x33b/0x440 Jun 11 22:00:46 devel kernel: [<c02043f2>] copy_to_user+0x4d/0x5f Jun 11 22:00:46 devel kernel: [<c01808ef>] cp_new_stat64+0xf0/0x102 Jun 11 22:00:46 devel kernel: [<c02043f2>] copy_to_user+0x4d/0x5f Jun 11 22:00:46 devel kernel: [<c02cfa06>] hiddev_read+0x109/0x23c Jun 11 22:00:46 devel kernel: [<c02cfb90>] hiddev_ioctl+0x0/0xa07 Jun 11 22:00:46 devel kernel: [<c018a9f9>] do_ioctl+0x39/0x52 Jun 11 22:00:46 devel kernel: [<c018aba7>] vfs_ioctl+0x57/0x195 Jun 11 22:00:46 devel kernel: [<c018ad44>] sys_ioctl+0x5f/0x6f Jun 11 22:00:46 devel kernel: [<c0103903>] syscall_call+0x7/0xb The messages started at 21:01:46 and recurred at one-minute intervals. Every five minutes beginning at 21:05:46 the interval contained lots of these messages, whereas the other intervals contained only a few. Interestingly, the messages stopped at 22:00:46 after 61 repetitions, and at 22:48:53 (almost an hour later!) the first "control queue full" message appeared. So it's hard to say whether there's any definite connection. But maybe this pattern will provide a clue. At this point you should take out the patch I sent, and add some debugging printks to hiddev_ioctl() and hid_submit_ctrl(). It might be particularly interesting to see the value of current->comm. It also would be interesting to see why these calls didn't show up in Adam's debugging. Alan Stern ------------------------------------------------------- This SF.Net email is sponsored by: NEC IT Guy Games. How far can you shotput a projector? How fast can you ride your desk chair down the office luge track? If you want to score the big prize, get to know the little guy. Play to win an NEC 61" plasma display: http://www.necitguy.com/?r=20 _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel