**** BEGIN LOGGING AT Sat Jun 25 02:59:58 2016 Jun 25 05:15:49 hey ZeekHuge Jun 25 05:16:09 Hi Visaoni :( struggling with PRU INTC Jun 25 05:16:31 ;( Jun 25 05:17:10 heh, was about to ask how you'd decided to move data from the PRUs to userspace Jun 25 05:25:45 will be using RPmsg, the standard way. And will later see what to do with RPMsg overheads. But as far as I have understood, it wont make much difference. Using RPMsg wont create much overhead. its internals are quite simple. And then I believe we can use the compiler optimization flags to make code much faster. which would result in much much less overhead. And Rpmsg is really simple inside. Jun 25 05:32:02 I'll point myself in that direction then. by the looks of your ADC I think you're probably pushing more data through than I am Jun 25 05:37:11 atleast 8 bits per sample and then, 15 M samples in one second/ Jun 25 05:52:46 ZeekHuge: you want to show me what you've got that's not working? Jun 25 05:53:37 https://github.com/ZeekHuge/BeagleScope/blob/master/docs/current_remoteproc_drivers.notes#L207 Jun 25 05:53:52 just a secx Jun 25 05:53:55 *sec Jun 25 05:53:57 so Jun 25 05:55:11 pru code 1 : http://paste.debian.net/754819/ Jun 25 05:55:35 pru code 0 :http://paste.debian.net/754820/ Jun 25 05:55:51 so 1 is generating system event 18 Jun 25 05:56:42 just a second theres some thing wrong there Jun 25 05:57:19 you're welcome ;) Jun 25 06:00:09 haha ! Jun 25 06:00:13 Thank you Wormo ! Jun 25 06:00:17 Got it working ! Jun 25 06:00:19 haha! Jun 25 06:00:22 niiice Jun 25 06:00:31 she's magic Jun 25 06:00:38 sometimes it does help just to start explaining it Jun 25 06:00:41 yes she definitely is ! Jun 25 06:02:12 So pru1 is interrupting pru1 in every 1 second and pru0 prints "Hi" when its getting interrupted. Jun 25 06:02:29 commit time! Jun 25 06:02:43 new example Jun 25 06:02:59 yeah . Jun 25 06:04:35 should I make a new branch for examples ? Jun 25 06:05:04 but then I will have to port the previous examples to the new RPMsg Jun 25 06:05:37 Wormo: ^ Jun 25 06:06:04 make a new branch for the new RPmsg Jun 25 06:06:17 Visaoni: did you look at the uart and ethernet examples? Jun 25 06:06:21 okay. Jun 25 06:06:53 sounded like it will not take long for you to get the existing examples tweaked Jun 25 06:07:13 nerdboy: he did another commit with a struct in local memory Jun 25 06:07:20 nerdboy: I looked at the uart one. looked like globals default to the PRU-specific memory. I ran the numbers and decided that's probably not a problem Jun 25 06:07:41 if it is I can do something more complicated with the shared memory too later Jun 25 06:07:47 yes true Jun 25 06:08:21 * nerdboy finally figured out this one is ethernet => http://git.ti.com/pru-software-support-package/pru-software-support-package/trees/master/examples/am335x/PRU_IEP Jun 25 06:08:34 industrial ethernet I think Jun 25 06:08:40 yup Jun 25 06:08:49 fun with TI acronyms Jun 25 06:09:13 AKA TIA Jun 25 06:09:17 the biggest thing the userspace hack people got right Jun 25 06:09:28 so, new RPMsg? Jun 25 06:10:32 Visaoni: yes, if you're on same kernel as ZeekHuge (and I recommend staying together on version if possible) Jun 25 06:13:07 Wormo: new with branches, so some dumb questions. Since the new branch will be for examples and for anyone who just want examples. so, I should make this branch pointing to the first commit right ? Jun 25 06:13:21 and then add the examples folder Jun 25 06:13:29 and port examples to it. Jun 25 06:13:45 I would go from current master, Jun 25 06:13:59 and just patch up the current examples Jun 25 06:14:17 it's not really "branch only for examples" necessarily Jun 25 06:14:27 you can Make a Tag Here if that helps... Jun 25 06:15:01 okay Jun 25 06:15:06 is there a place I should be looking for updated RPMsg stuff? Jun 25 06:15:17 but rather a branch where you are updating the examples including RPmsg Jun 25 06:15:53 you mean the interrupt stuff in v5? Jun 25 06:17:03 the debian iot thing on my green just updated to 4.4.12-r32 Jun 25 06:17:21 did you get iot not to kill serial then? Jun 25 06:17:38 and I mean generally some docs about RPMsg that's up to date Jun 25 06:17:42 there's accompanying firmware changes, but ZeekHuge just said they were minor Jun 25 06:17:43 i would think some of those examples in ti git v5 need interrupts to work Jun 25 06:18:24 i didn't have any serial trouble but i'm not using usb serial Jun 25 06:18:33 yes up-to-date docs are a good thing... Jun 25 06:18:37 On the kernel side, we really dont need to worry about anything with the newer rpmsg Jun 25 06:18:38 mostly just ssh and usb power Jun 25 06:18:58 oh, I thought you said the iot image killed it. nvm then Jun 25 06:19:02 everything is managed internally there on the kernel side Jun 25 06:19:21 ZeekHuge: got a link for the examples you used for figuring out the minor firmware change? Jun 25 06:19:22 and on the PRU firmware , there are some minor changes Jun 25 06:20:43 nope Jun 25 06:22:01 so is the PRU Training Hands on Lab thing up to date? mostly up to date? not at all? Jun 25 06:22:07 for RPMsg Jun 25 06:22:51 I think "mostly" but unlike ZeekHuge I didn't track down the exact firmware changes needed Jun 25 06:22:52 Wormo: yes these are those changes : http://git.ti.com/pru-software-support-package/pru-software-support-package/commit/69805828df0f262fb60363c2db189d1b8d0b693c/diffs Jun 25 06:23:12 Visaoni: yes its up to dare Jun 25 06:23:14 *date Jun 25 06:23:20 thanks ZeekHuge Jun 25 06:23:41 perfect Jun 25 06:23:50 awesome, thanks Jun 25 06:24:43 jasonreeder is managing both, rpmsg and the package, so, though he is making changes on both side, but keeping them synced too. Jun 25 06:28:18 Looking at this nice PRU Training lab, I see the source tgz is still using mailbox event Jun 25 06:28:22 http://processors.wiki.ti.com/index.php/PRU_Training:_PRU_PID_Motor_Demo Jun 25 06:28:47 so it could use another source tarball for newer kernels Jun 25 06:29:18 void rpmsg_interrupt(volatile struct pid_data* pid, struct pru_rpmsg_transport *transport, uint8_t *payload, Jun 25 06:29:21 uint16_t src, uint16_t dst, uint16_t len) { Jun 25 06:29:23 /* Check bit 31 of register R30 to see if the mailbox interrupt has occurred */ Jun 25 06:29:26 if(__R31 & HOST_INT){ Jun 25 06:29:57 Other than that, it looks like a really nice practical example Jun 25 06:30:32 includes both communication between PRU and with the ARM Jun 25 06:34:01 BTW somebody should do a wiki post showing the difference in API Jun 25 06:44:21 somebody like a student you mean? Jun 25 06:44:36 somebody exactly like a student Jun 25 06:45:26 one of those students who happen to be using RPMsg Jun 25 09:06:46 jic23b: there? Jun 25 09:20:21 jic23: there? Jun 25 09:21:10 jic23: one of the patch which I got from yours is not working :( Jun 25 12:30:21 Hi!! Jun 25 12:30:48 I'm facing some issue with the deb-pkg stuff.. I am not able to build the kernel with fakeroot.. can anyone help me out? Jun 25 12:31:05 ERROR: ld.so: object 'libfakeroot-sysv.so' from LD_PRELOAD cannot be preloaded (wrong ELF class: ELFCLASS32): ignored. Jun 25 13:30:03 Hi, anybody here? Jun 25 13:34:22 me Jun 25 13:38:54 I have a problem with with my alsa driver. everytime when i want to play something there are many kernel error of remote proc and dsp firmware Jun 25 13:39:04 the kernel freezes after some seconds. Jun 25 13:39:32 but my kernel module is in no connection with remoteproc driver Jun 25 13:41:32 remoteproc3: crash detected in 41000000.dsp: type mmufault Jun 25 13:41:39 omap-iommu 41502000.mmu: 41502000.mmu: errs:0x00000002 da:0xda66da90 pgd:0xeb09b698 *pgd:px00000000 Jun 25 13:41:45 remoteproc3: handling crash #1 in 41000000.dsp Jun 25 13:41:51 remoteproc3: recovering 41000000.dsp Jun 25 13:41:59 omap-iommu 40d02000.mmu: iommu fault: da 0xbff00000 flags 0x0 Jun 25 13:42:09 any idea? Jun 25 13:42:52 seems to be something with mmu and dsp Jun 25 13:59:23 jic23: did you see my last message? Jun 25 17:06:02 kiran4399: multilib problem? try pointing LD to 64bit libs? Jun 25 17:08:53 nerdboy: this is how we build the deb file right? Jun 25 17:08:54 https://gist.github.com/alexanderhiam/29fcf814f731dd94f8850961f1a61ae3 Jun 25 17:47:54 hey ZeekHuge, noticed a cleanup commit which is good -- might want to add a .gitignore so that new gen/ dirs don't accidentally get checked in Jun 25 17:48:06 like https://github.com/Visaoni/beagle-sonic-anemometer/blob/master/.gitignore Jun 25 17:49:05 Wormo: oh, okay. Will do. Thank you. :) Jun 25 17:52:02 also put back the stuff you took out of the README in new branch, it was useful stuff Jun 25 17:56:31 kiran4399: take a look at the build_deb.sh/sgx_create_package.sh scripts in robert's ti-linux-kernel-dev ? Jun 25 17:56:40 *repo Jun 25 17:57:24 kiran4399: why are you using fakeroot? Jun 25 17:57:47 should work fine without it Jun 25 17:59:20 as long as you're doing this where you have permissions Jun 25 18:01:06 nerdboy.. if I have this: arm-debian-jessie.rootfs.tgz Jun 25 18:01:17 don't I need to use fakeroot? Jun 25 18:16:08 no for building/packaging a kernel in your $HOME... Jun 25 18:16:16 try it and see Jun 25 18:16:31 s/no/not/ Jun 25 18:25:33 ZeekHuge: Did you get the package yet? Jun 25 18:25:46 what's the current status? Jun 25 18:27:44 yes, I received the mail that it is for delivery today, but it didnt came . Its in the local office . and will be delivered probably on Monday now . Jun 25 18:31:36 Abhishek_: ^ Jun 25 18:31:57 Sounds good. So you had to fill up a form right? Jun 25 18:34:07 Abhishek_: yep. I have already submitted it after the last meeting itself. :) Jun 25 18:37:06 Abhishek_: Isnt it visible there ? Jun 25 18:37:09 I did it . Jun 25 18:37:13 let me check Jun 25 18:51:01 Visaoni: I hear you still need to fill out mid-term form for students, don't forget to do that today Jun 25 18:53:56 Not sure how it is today, but in the previous years failing to complete student evaluation leads to an automatic failure, even if mentors give you a clear evaluation. Jun 25 19:12:26 yep, same this time too. We got emails saying the same. Jun 25 22:46:32 Wormo: ah, I knew I was forgetting something last night. will get on it shortly Jun 25 22:47:20 Visaoni: I guess the github version pru-software-support-package looks pretty out of date, are you using the original at git.ti.com for reference? Jun 25 22:48:07 They keep this one updated: http://git.ti.com/pru-software-support-package/pru-software-support-package/ Jun 25 22:48:30 yep, that's wnat I ended up with Jun 25 22:48:36 Ok good Jun 25 22:49:02 (beware: PRU training page links to outdated unmaintained versions packaged by TI) Jun 25 22:50:07 sad but true, some of those examples were quite cool Jun 25 22:59:32 can either fix them or roll back to mailboxes i guess... Jun 25 23:00:03 probably better to stick with interrupts/latest Jun 25 23:00:20 the new stuff looks nicer, actually. I think the examples in the support package are probably enough Jun 25 23:03:25 the uart one uses interrupts to poll on data_received_blah (with a really tiny fifo) Jun 25 23:32:24 Visaoni: nothing new to push yet? we like to see wip and make snarky comments... Jun 25 23:36:29 * Wormo tries to refrain from snarky comments Jun 25 23:36:45 but WIP is good Jun 25 23:40:52 * nerdboy blames it on being raised by snarky wolves... Jun 25 23:42:37 nerdboy: not just yet. just getting back to things for today Jun 25 23:47:59 nerdboy, he's on dburr timezone not PDT Jun 26 00:12:55 none of that now Jun 26 01:14:12 hey nerdboy, Wormo_, am I right in thinking that I can more or less ignore the RPMsg src/dst stuff for now if I've just got one channel going? Jun 26 01:20:46 sounds good? Jun 26 01:21:45 firmware's running but not doing anything message-y yet? Jun 26 01:24:51 I don't actually know if it's running... but it's compiling? we all know how strict that C compiler is... Jun 26 01:25:56 msg-y stuff should be (theoretically) up shortly. I'll set it up sending some fake data back and then jump over to the userspace side and see about receiving it **** ENDING LOGGING AT Sun Jun 26 02:59:58 2016