On Wed, 26 Apr 2017 14:53:41 -0700 Lisa Nguyen lisa.nguyen@linaro.org wrote:
On 20 April 2017 at 02:57, Neil Williams codehelp@debian.org wrote:
On Tue, 18 Apr 2017 12:13:01 -0700 Lisa Nguyen lisa.nguyen@linaro.org wrote:
Hi Dave,
On 12 April 2017 at 08:09, Dave Pigott dave.pigott@linaro.org wrote:
Hi Lisa,
I need to fix the device dict based on some things that Neil did.
Still on my radar.
https://pmwg.validation.linaro.org/scheduler/job/1405#L385
This problem is now fixed. For reference, USB devices which show a serial number which includes forward slash / will have that slash replaced by an underscore when processed through pyudev. The device dictionary has been updated for pmwg panda-01. A documentation fix is in preparation.
(BTW 2017.4 includes NFS support for panda, so I used that in my test job, your ramdisk test job will work too.)
Thanks for looking into this, Neil.
I wrote a test definition to build, install, and run the arm-probe command-line tool in LAVA, but I'm running into the error where the probe path cannot be accessed without running into the 'exclusive' error: https://pmwg.validation.linaro.org/scheduler/job/1489#L3776
OK, I've identified what is wrong here. LAVA is only adding devices from the 'usb' subsystem to the LXC but the energy probe software needs the 'tty' element instead or as well.
I've tried to use mknod to create the device within the LXC but LXC itself seems to do extra work to allow programs like minicom to use the device. In local tests, I can't get minicom to be happy with ttyUSB0 created using mknod inside the LXC, even thought the device looks the same as it is outside the LXC. Using lxc-device add works with minicom. (I'm testing with a usb serial device connected to a BBB rather than an energy probe but I'm hoping the energy probe code is essentially doing the same operations as minicom.)
So this will need a code change (which I've already prepared) to find the tty device node and pass that to lxc-device add. We are looking to release 2017.5 relatively soon - once we have had time to do full testing on staging after the much needed reorganisation of staging.validation.linaro.org. (The re-org includes the provision of devices with energy probes attached, so we can include checks on this functionality in future releases.)
In a manual setting, I expect to see output similar to this:
# configuration: config-panda-lab # config_name: pandaboard # trigger: 0.400000V (hyst 0.200000V) 0.000000W (hyst 0.200000W) 400us Configuration: pandaboard # date: Wed, 26 Apr 2017 22:47:59 +0100 # host: pmwg-server-01.pmwglab #
- /dev/serial/by-id/usb-NXP_SEMICOND_ARM_Energy_Probe_S_NO44440001-if00
Starting... sending start to 0 # VDD_ALL VDD ROOT #ff0000 SoC # # time VDD(V) VDD(A) VDD(W) 0.000500 5.19 0.0994 0.51568 0.000600 5.19 0.0704 0.36538 0.000700 5.19 0.0994 0.51538 0.000800 5.19 0.0764 0.39608 0.000900 5.20 0.0654 0.33967 0.001000 5.19 0.0934 0.48444 0.001100 5.19 0.0754 0.39097 0.001200 5.19 0.0734 0.38052 ... ...
However, I was able to verify that the LXC detected the probe by writing an inline test definition to use the lsusb command: https://pmwg.validation.linaro.org/scheduler/job/1489#L2049
As we discussed during our sync, the device dictionary looks correct, yet I'm not sure if the probe is detected inside the LXC. I ran another job and it's the same result as the previous one I posted in this thread.
Also I don't know how USB devices are listed inside LXCs, but I would expect to see something like /dev/bus/usb/***, /dev/ttyUSB*, or /dev/ttyS*.
Can you check to see if I'm overlooking something in the log below? https://pmwg.validation.linaro.org/scheduler/job/1389#L1030
Dave
On 12 Apr 2017, at 16:08, Lisa Nguyen lisa.nguyen@linaro.org wrote:
On 19 March 2017 at 22:10, Senthil Kumaran S senthil.kumaran@linaro.org wrote:
Hi Lisa,
On Saturday 18 March 2017 12:49 AM, Lisa Nguyen wrote:
After re-running a sample test job[1] to test AEP support, I noticed that the usb path to the probe for the pandaboard was not detected in the container anymore:
https://pmwg.validation.linaro.org/scheduler/job/997#L1014
To verify that the /dev/usb/ directory didn't exist, I tweaked the test job a little to check the contents of /dev inside lxc and ran it again:
This has quite changed in 2017.2 as a result of https://projects.linaro.org/browse/LAVA-814
We do not use the symlink provided by udev rules anymore, instead we use the device_info parameter from device dictionary. See https://staging.validation.linaro.org/static/docs/v2/admin-lxc-deploy.html#a... for how to use device_info.
In order to use the above the panda device dictionary should be updated.
I see that the device dictionary got updated with device_info added.
I'm not sure if I see the path of the probe under /dev unless I'm missing something in the job definition? https://pmwg.validation.linaro.org/scheduler/job/1335#L1051
Thank You.
Senthil Kumaran http://www.stylesen.org/ http://www.sasenthilkumaran.com/
linaro-validation mailing list linaro-validation@lists.linaro.org https://lists.linaro.org/mailman/listinfo/linaro-validation
linaro-validation mailing list linaro-validation@lists.linaro.org https://lists.linaro.org/mailman/listinfo/linaro-validation
--
Neil Williams