On Thu, Dec 4, 2008 at 07:28, Jar <jar@xxxxxxx> wrote: >> On Wed, 2008-12-03 at 15:10 -0800, Greg KH wrote: >>> On Wed, Dec 03, 2008 at 04:11:29PM +0100, Kay Sievers wrote: >>> > On Wed, 2008-12-03 at 13:12 +0100, Kay Sievers wrote: >>> > > On Wed, Dec 3, 2008 at 10:27, Kay Sievers <kay.sievers@xxxxxxxx> wrote: >>> > > > On Wed, Dec 3, 2008 at 07:45, Greg KH <greg@xxxxxxxxx> wrote: >>> > > >> On Sun, Nov 30, 2008 at 06:21:11AM +0100, Kay Sievers wrote: >>> > > >>> On Sat, Nov 29, 2008 at 18:37, Greg KH <greg@xxxxxxxxx> wrote: >>> > > >>> > On Sat, Nov 29, 2008 at 10:00:59AM +0200, Jar wrote: >>> > > >>> >> Greg KH wrote: >>> > > >>> >>> HAL contains a list of these modems and a mapping of each port to what >>> > > >>> >>> it does based on the specific device. >>> > > >>> >>> Try using that instead of udev specific rules for your accessed to the >>> > > >>> >>> modem, it will work much better. >>> > > >>> >> >>> > > >>> >> Thanks for your reply! Does HAL require that I have desktop installed, >>> this >>> > > >>> >> is my home "server" machine without X? Do you have any starting point >>> or >>> > > >>> >> where to learn more how to do this with HAL. >>> > > >>> > >>> > > >>> > I do not think HAL requires X. Try asking on your distro mailing list >>> > > >>> > for how they have incorporated HAL into the releases you are using. >>> > > >>> >>> > > >>> It might get pretty complicated to use D-Bus/HAL for simple setups, >>> > > >>> and rather static configurations like this. NetworkManager handles >>> > > >>> that, and we will even get modem-probing soon, but in cases like this >>> > > >>> it might be easier to have a simple - not very flexible - but working >>> > > >>> solution. >>> > > >>> >>> > > >>> Maybe we can have an "index" sysfs file at the serial device, which >>> > > >>> carries the instance number, per usb device? This would allow us to >>> > > >>> create persistent links which do not depend on the kernel device >>> > > >>> enumeration across multiple device. >>> > > >>> >>> > > >>> We did the same for v4l devices recently: >>> > > >>> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=539a7555b31e65e66fb84c881d07d2bf18c974d0 >>> > > >>> >>> > > >>> Greg, what do you think? >>> > > >> >>> > > >> Don't we already have something like this today (the "ports" on an >>> > > >> individual usb-serial device)? Or do we just need to export that a bit >>> > > >> better? >>> > > > >>> > > > Yeah, maybe that works already. I don't have such a device. >>> > > >>> > > Ok, I found one. You mean "port_number", right? But unfortunately it >>> > > looks pretty useless in this case. :) >>> > > >>> > > $ grep . /sys/bus/usb-serial/devices/*/port_number >>> > > /sys/bus/usb-serial/devices/ttyUSB1/port_number:0 >>> > > /sys/bus/usb-serial/devices/ttyUSB2/port_number:0 >>> > > >>> > > They would only be not "0" if the serial lines would be on the same >>> > > usb interface? >>> > > >>> > > We would need a per-device enumeration, but that's nothing the >>> > > usb-serial stuff knows/cares about if we have different usb >>> > > interfaces, right? >>> > >>> > Greg, do you have a usb multi-port serial card? Can you possibly give >>> > this a try, and show us "tree /dev/serial"? >>> > >>> > I have this here now for my "simple" devices: >>> > /dev/serial >>> > `-- by-id >>> > |-- usb-067b_2303-serial-if00-port0 -> ../../ttyUSB0 >>> > |-- usb-HUAWEI_Technology_HUAWEI_Mobile-serial-if00-port0 -> ../../ttyUSB1 >>> > `-- usb-HUAWEI_Technology_HUAWEI_Mobile-serial-if01-port0 -> ../../ttyUSB2 >>> > >>> > If we can make that work as expected, I'll look into by-path/, so we can >>> > have identical devices properly identified, if needed. Then we can check >>> > how we should finally name all that, and possibly add that stuff to the >>> > default rule set. >>> >>> >>> This looks good, here's the output using udev 133 and the rules and a >>> few usb-serial devices plugged into the system at the same time: >>> >>> $ tree /dev/serial/ >>> /dev/serial/ >>> `-- by-id >>> |-- usb-Inside_Out_Networks_Edgeport >>> | |-- 4_04-01-006467-serial-if00-port0 -> ../../../ttyUSB0 >>> | |-- 4_04-01-006467-serial-if00-port1 -> ../../../ttyUSB1 >>> | |-- 4_04-01-006467-serial-if00-port2 -> ../../../ttyUSB2 >>> | `-- 4_04-01-006467-serial-if00-port3 -> ../../../ttyUSB3 >>> |-- >>> usb-Keyspan__a_division_of_InnoSys_Inc._USB_4-port_Serial_Adapter-serial-if00-port0 >>> -> ../../ttyUSB8 >>> |-- >>> usb-Keyspan__a_division_of_InnoSys_Inc._USB_4-port_Serial_Adapter-serial-if00-port1 >>> -> ../../ttyUSB9 >>> |-- >>> usb-Keyspan__a_division_of_InnoSys_Inc._USB_4-port_Serial_Adapter-serial-if00-port2 >>> -> ../../ttyUSB10 >>> |-- >>> usb-Keyspan__a_division_of_InnoSys_Inc._USB_4-port_Serial_Adapter-serial-if00-port3 >>> -> ../../ttyUSB11 >>> |-- usb-Prolific_Technology_Inc._USB-Serial_Controller-serial-if00-port0 -> >>> ../../ttyUSB7 >>> |-- usb-Sierra_Wireless__Incorporated_AirCard-serial-if00-port0 -> >>> ../../ttyUSB4 >>> |-- usb-Sierra_Wireless__Incorporated_AirCard-serial-if00-port1 -> >>> ../../ttyUSB5 >>> `-- usb-Sierra_Wireless__Incorporated_AirCard-serial-if00-port2 -> >>> ../../ttyUSB6 >>> >>> I'm not so sure about the long names, but it is unique :) >> >> Yeah, it looks really strange with your devices. Why put people poetry >> in the device strings? :) >> >> But it seems to work, which is great. Here is a new version of the >> serial.rules and a /lib/udev/path_id which needs to be replaced. >> >> Can you try that again on your "collection"? :) >> >> Thanks, >> Kay >> # do not edit this file, it will be overwritten on update >> >> ACTION!="add|change", GOTO="persistent_serial_end" >> SUBSYSTEM!="tty", GOTO="persistent_serial_end" >> KERNEL!="ttyUSB[0-9]*", GOTO="persistent_serial_end" >> >> SUBSYSTEMS=="usb-serial", ENV{ID_PORT}="$attr{port_number}" >> ENV{ID_PORT}=="", GOTO="persistent_serial_end" >> >> IMPORT="path_id" >> ENV{ID_PATH}=="?*", SYMLINK+="serial/by-path/$env{ID_PATH}-port$env{ID_PORT}" >> >> IMPORT="usb_id --export" >> ENV{ID_SERIAL}=="", GOTO="persistent_serial_end" >> SUBSYSTEMS=="usb", ENV{ID_IFACE}="$attr{bInterfaceNumber}" >> ENV{ID_IFACE}=="", GOTO="persistent_serial_end" >> SYMLINK+="serial/by-id/$env{ID_BUS}-$env{ID_SERIAL}-if$env{ID_IFACE}-port$env{ID_PORT}" >> >> LABEL="persistent_serial_end" > > I can't test this right now, but doesn't quite understand what this will do. Does > this solve the problem where the ttyUSBx interface changes and make it always be the > same? > > If I have e.g. ttyUSB1,ttyUSB2 and ttyUSB3 devices, what will be the symlink to > ttyUSB1 (used for pppd), which only interests? If you have different devices, all links in /dev/serial/by-id/ will stay the same, regardless of the order, or the USB hub/port they are plugged in. If you have identical devices, without any serial number, only the controller/hub/port number can be used to distinguish them. The /dev/serial/by-id/ links will overwrite each other for identical devices, and /dev/serial/by-path/ must be used. > And how about the other thing, how to > trigger some wvdial connect script via udev at the same time the > ttyUSB1 thing comes up? You can match on properties of the device, or even the link name itself with SYMLINK=="serial/by-..." with udev rules, that come after the 60-*.rules. These rules can trigger anything you need. Kay -- To unsubscribe from this list: send the line "unsubscribe linux-hotplug" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
