**** BEGIN LOGGING AT Mon Mar 19 03:00:02 2018 Mar 19 07:21:36 please no PM's unless there is a very good reason. Mar 19 07:21:43 they will be ignored. Mar 19 10:48:35 jkridner, abhishek: As there is no support for remote-proc on PyPRUSS, how is the BeagleBone controlled right now? Mar 19 10:49:17 As that is one of the main functions of remote-proc on a PRU. Mar 19 15:01:40 spiritus Check this out - https://groups.google.com/forum/m/#!msg/beagleboard/sFu-K8YPQ3s/UVAATO3xBgAJ Mar 19 15:09:58 abhishek_: jkridner: did you take a look at my draft proposal for PyPRUSS? Mar 19 15:10:37 muneeb17: Link? Mar 19 15:11:02 abhishek_: https://elinux.org/BeagleBoard/GSoC/PyPRUSS Mar 19 15:11:30 I also shared it through the gsoc dashboard Mar 19 15:13:13 * jkridner[m] is looking now Mar 19 15:13:48 muneeb17: to avoid spam, you might obfuscate your e-mail. Mar 19 15:14:05 jkridner: ok Mar 19 15:15:37 muneeb17: PRU control is something already handled by the remoteproc kernel interface Mar 19 15:16:28 You should go into detail into the pru-remoteproc driver in the kernel tree of v4.9 ( Find it out yourself ) Mar 19 15:17:14 muneeb17: proposal is pretty clean. Can you be specific when talking about functions and specify the function names? Mar 19 15:17:36 jkridner: Sure Mar 19 15:18:01 can you also list off the examples? Mar 19 15:18:14 it'll make it real easy to track progress that way. Mar 19 15:18:37 and looking them over should help you with the scope. Mar 19 15:18:40 jkridner: the existing examples in the PyPRUSS, you mean? Mar 19 15:19:23 at each milestone where you say things like "remoteproc examples". Mar 19 15:19:45 just enumerate what examples will be working by what phase.... Mar 19 15:19:53 Ok, will do Mar 19 15:20:02 we can iterate on some more specific feedback based on that. Mar 19 15:20:26 Sure. w Mar 19 15:20:48 Is resource table generation fine, or is it overkill? Mar 19 15:22:26 jkridner: ^ Mar 19 15:22:43 I don't think that's overkill. Mar 19 15:22:49 Also, I don't feel RPMsg should be such a stretch goal. Mar 19 15:23:32 Yeah, I was thinking of adjusting it in the timeline as well Mar 19 15:24:19 jkridner: We do have to use RPMsg for interrupts control, right? Mar 19 15:24:54 no. Mar 19 15:25:01 we can send interrupts without RPMsg. Mar 19 15:25:18 using remoteproc? Mar 19 15:25:40 But the interrupts are in the kernel driver, isn't it? Mar 19 15:26:14 yeah, I will check the rproc driver once as well Mar 19 15:27:33 Rpmsg takes those interrupts and provides a nice interface to the outside world in userspace. Mar 19 15:27:58 yeah, using kick and virtqueues, right? Mar 19 15:31:53 abhishek_: do they always need to contain valid messages if no message is expected? Mar 19 15:32:24 if there was a lower-effort way of sending notifications, it'd be nice to support. if it'd send the kernel in the weeds, I guess RPMsg might be needed. Mar 19 15:33:01 Get raw interrupts to userspace, and the remoteproc interface will start to look more like what UIO is today. Mar 19 15:33:27 hjhee_: does Sigrok have integrations for what you are trying as well? I'd hope there is an easy way for people to specify the triggers. Mar 19 15:34:52 When I brought forward BeagleLogic from 3.8.13 to 4.9, I initially did play with RPMsg but eventually decided to directly interact with the raw PRU interrupts and built a memory-mapped interface to the PRU for sending and receiving commands directly in the kernel module. Mar 19 15:35:45 If RPMsg is loaded, it directly handles the PRU interrupts. Mar 19 15:36:37 Iillustris , m_w: maybe I'm missing something about the terminal server as it sounds rather basic and done to me. Why is a GSoC project needed to help the community with this? Can you do something to enable more ports via PRU? Mar 19 15:37:42 abhishek_: so what do you advise to use now, for the PyPRUSS Mar 19 15:37:58 abhishek_: if there's a way to pass raw interrupts, it'd be nice to support. I think people get very intimidated by RPMsg, when direct memory poking and interrupts has been possible with UIO. Mar 19 15:39:32 muneeb17: Go through remoteproc and RPMsg drivers and try to understand it the best you can and then structure your proposal around it. Mar 19 15:40:36 abhishek_: Sure thing Mar 19 15:40:43 jkridner: I think people getting intimidated by RPMsg is a little sad, as when it works right, it's like you're talking to a Serial Port. Mar 19 15:42:29 abhishek_: that's a good reason to make sure PyPRUSS covers RPMsg, not just as a stretch goal. Mar 19 15:43:27 I agree. Mar 19 15:44:31 Roger that Mar 19 16:13:06 jkridner, abhishek: As there is no support for remote-proc on PyPRUSS, how is the BeagleBone controlled right now? Mar 19 16:13:21 As that is one of the main functions of remote-proc on a PRU. Mar 19 19:02:36 spiritus: remoteproc doesn't need Python to control the PRUs. It provides a sysfs interface. Mar 19 19:56:37 Hi, Would like to get suggestions about a project idea : developing a blocks editor on top of BoneScript and Cloud9, so that BB users without much scripting experience can develop projects easily. Mar 19 21:08:52 Hmmmm **** ENDING LOGGING AT Tue Mar 20 03:00:06 2018