Hello, I just noticed that /dev/ttyUSBx devices are not destroyed and released on USB device disconnect if some process keeps the file open, and reading from the file does not return an error - it successfully reads 0 bytes instead. This affects my gpsd usage, so I believe it must be a recent regression or I would have noticed before. I need help tracking down the cause of this. I guess it could be related to my usb_wwan changes, but I do not understand how... Disconnecting the device: Aug 20 11:10:36 nemi kernel: [116888.253611] PM: Removing info for usb:2-4:1.2 Aug 20 11:10:36 nemi kernel: [116888.253636] bus: 'usb': remove device 2-4:1.2 Aug 20 11:10:36 nemi kernel: [116888.253727] /usr/local/src/git/linux/drivers/usb/serial/usb_wwan.c: usb_wwan_indat_callback: nonzero status: -108 on endpoint 02. Aug 20 11:10:36 nemi kernel: [116888.253731] /usr/local/src/git/linux/drivers/usb/serial/usb_wwan.c: usb_wwan_indat_callback: nonzero status: -108 on endpoint 02. Aug 20 11:10:36 nemi kernel: [116888.253734] /usr/local/src/git/linux/drivers/usb/serial/usb_wwan.c: usb_wwan_indat_callback: nonzero status: -108 on endpoint 02. Aug 20 11:10:36 nemi kernel: [116888.253738] /usr/local/src/git/linux/drivers/usb/serial/usb_wwan.c: usb_wwan_indat_callback: nonzero status: -108 on endpoint 02. Aug 20 11:10:36 nemi kernel: [116888.253769] /usr/local/src/git/linux/drivers/usb/serial/usb-serial.c: usb_serial_disconnect Aug 20 11:10:36 nemi kernel: [116888.253802] /usr/local/src/git/linux/drivers/usb/serial/usb-serial.c: serial_hangup - port 2 Aug 20 11:10:36 nemi kernel: [116888.253810] PM: Removing info for usb-serial:ttyUSB2 Aug 20 11:10:36 nemi kernel: [116888.253830] bus: 'usb-serial': remove device ttyUSB2 Aug 20 11:10:36 nemi kernel: [116888.253851] device: 'ttyUSB2': device_unregister Aug 20 11:10:36 nemi kernel: [116888.253854] PM: Removing info for No Bus:ttyUSB2 Aug 20 11:10:36 nemi kernel: [116888.253909] qcserial ttyUSB2: Qualcomm USB modem converter now disconnected from ttyUSB2 Aug 20 11:10:36 nemi kernel: [116888.253926] qcserial 2-4:1.2: device disconnected Notice the "serial_hangup", which AFIACS should cause the TTY to start reporting errors. But it does not. Reading from the file handle continues to return 0 bytes and no error. Only after the userspace application closes the handle do we see this: Aug 20 11:11:06 nemi kernel: [116918.265281] /usr/local/src/git/linux/drivers/usb/serial/usb-serial.c: serial_close - port 2 Aug 20 11:11:06 nemi kernel: [116918.266455] /usr/local/src/git/linux/drivers/usb/serial/usb-serial.c: serial_cleanup - port 2 Aug 20 11:11:06 nemi kernel: [116918.266462] /usr/local/src/git/linux/drivers/usb/serial/usb-serial.c: destroy_serial - Qualcomm USB modem Aug 20 11:11:06 nemi kernel: [116918.266466] /usr/local/src/git/linux/drivers/usb/serial/usb-serial.c: return_serial Aug 20 11:11:06 nemi kernel: [116918.266474] /usr/local/src/git/linux/drivers/usb/serial/usb-serial.c: port_release - ttyUSB2 Aug 20 11:11:06 nemi kernel: [116918.266505] device: 'ttyUSB2': device_create_release The effect for me is that the serial port names change on system suspend/resume, because some application hangs on to the old port. That's annoying and not like I believe this should work. I'd appreciate any hints on where to start looking for the problem. I do not understand much of the TTY layer interaction, but I did expect serial_hangup to be enough here. Bjørn -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
