**** BEGIN LOGGING AT Wed Sep 04 03:00:01 2019 Sep 04 04:04:40 Sorry for harassing this page earlier. My connection was terrible on the FET, BBB, and LED. I fixed it. Sep 04 05:42:26 uh, this isn't current anymore, is it? https://elinux.org/Building_BBB_Kernel Sep 04 05:43:20 i need to build a few modules, looking for how to get the source for 4.14.71-ti-r80 Sep 04 09:36:13 kenrestivo: https://pastebin.com/eLhrp1Hg Sep 04 09:37:03 some nots describing how to (re)build the standard kernels as debian packages and how to customize them Sep 04 09:39:15 *notes Sep 04 11:17:29 m Sep 04 14:27:53 Hi, I've upgraded to the latest LTS (4.19.69) on our AM3703 custom board similar to BeagleBoard. I see "omap-mcbsp 49022000.mcbsp: TX Buffer Overflow!" messages when playing short ding-dong sound. I was trying to change some parameters in sysfs and changing of 'dma_op_mode' helped. Sep 04 14:29:42 BTW I changed 'element' to 'threshold'. It is just a different way of DMA transfer start? May there be some side effect when playin ~2 sec ding-dong? Sep 04 14:29:55 *playing Sep 04 15:05:30 * zmatt takes a peek at the mcbsp driver Sep 04 15:09:38 but nothing is lost, the short sound seems ok to me. Although I must admit that our speaker is not HiFi at all... Sep 04 15:11:49 And just for the record. I found another "solution" on Logic PD forum - when CONFIG_VIDEO_OMAP3 is enabled, the message is also gone. I was really surprised that it helped. Sep 04 15:17:38 zmatt: thanks for the explanation Sep 04 15:32:50 tnovotny: maybe something to do with power management QoS ? I'm just making wild guesses here Sep 04 15:33:14 the difference in DMA mode is not immediately clear to me either Sep 04 15:34:14 there must be a reason that "element" is the default mode (unconditionally, no way to override via platform data or device tree) Sep 04 15:34:15 zmatt: maybe the power management is somehow related. I also found one mail stating that disabling of cpuidle should help. But it didn't in my case. That's why I was so surprised with the camera... Sep 04 15:34:42 but I'm not really familiar with McBSP or its driver Sep 04 15:35:15 (I don't have any omap3 device, only am335x and omap5) Sep 04 15:37:57 I have plenty of OMAP3 only :-)... So do you think that dma mode change available in sysfs only means something? Not just "nobody did it yet"? Sep 04 15:39:40 well apparently the setting has been around since kernel 2.6 Sep 04 15:41:47 so nobody cared yet? Sep 04 15:42:37 ? Sep 04 15:44:04 I mean that it is in the driver since 2.6, so nobody just make it configurable from pdata or dt? Sep 04 15:44:22 well it is not clear to me why the setting was introduced Sep 04 15:44:29 ok Sep 04 15:44:31 or based on what its value should be selected Sep 04 15:44:36 or who ought to be responsible for doing so Sep 04 15:44:53 it seems it used to have a third option ("frame") Sep 04 15:46:30 my guess would be that THRESHOLD makes the dma "burstier" Sep 04 15:47:01 i.e. instead of continuously keeping the fifo filled it will do so in bursts Sep 04 15:47:10 I don't see why that would be a desirable thing Sep 04 15:48:04 wait Sep 04 15:48:09 tx buffer *overflow* ? not underflow? Sep 04 15:48:13 what Sep 04 15:48:19 yep, overflow Sep 04 15:48:21 that implies a serious bug Sep 04 15:49:11 But as I wrote, I didn't notice a problem in my use case Sep 04 15:49:55 BTW that bug might be there for a longer time (at least on our platform with our config), because handling of overflow IRQ was added in ~4.4 Sep 04 15:50:38 I see reports of that error for kernel 4.9 Sep 04 15:51:45 the irq is enabled here: 4e85e7776eba5f6d5a3b8b43f34001a73c5d08a1 Sep 04 15:52:24 okay, so maybe the problem is older but only recently exposed Sep 04 15:53:13 you might want to send an email to the linux-omap and/or alsa-devel list Sep 04 15:53:32 this definitely sounds like a kernel bug that should be investigated Sep 04 15:54:11 since if dma is configured correctly, a tx fifo overrun should never be possible Sep 04 15:55:52 ok, so I will report on both mailing lists later this week. Thanks again for your time. Sep 04 17:06:24 zmatt: thanks Sep 04 17:36:13 umm, let's say, the reason i'm building this kernel is to build a module that isn't included in the kernel config. how do i break free of the build system to create a new .config, i.e. how can i get access to make oldconfig or make menuconfig? Sep 04 17:36:47 the script will run make menuconfig to give you an opportunity to tweak the config Sep 04 17:36:52 I mentioned that in my notes Sep 04 17:38:10 thanks Sep 04 17:41:41 i see it https://github.com/RobertCNelson/ti-linux-kernel-dev/blob/ti-linux-4.14.y/build_deb.sh#L57 Sep 04 17:42:39 yeah, you could also modify patches/defconfig before running the script Sep 04 17:48:32 i kinda wanted to uncomment some stuff and make oldconfig, but instead i'll try to fish it out of menuconfig Sep 04 17:48:54 both options are available Sep 04 17:49:13 i see this AUTO_BUILD flag Sep 04 17:55:21 oh. i see what the problem is. the module isn't built, because the driver seems to have been removed from the kernel tree :/ Sep 04 17:57:27 what driver? Sep 04 17:59:20 KERNEL/drivers/iio/accel/adxl345_i2c.c and ignore/linux-src/drivers/iio/accel/adxl345_i2c.c appear to be there, but it's not in menuconfig Sep 04 18:01:41 huh, it's listed under INPUT drivers, that's bizarre Sep 04 18:01:42 it's enabled by default: https://github.com/RobertCNelson/ti-linux-kernel-dev/blob/ti-linux-4.14.y/patches/defconfig#L2683-L2685 Sep 04 18:01:46 CONFIG_INPUT_ADXL34X_I2C=m Sep 04 18:01:48 ohh Sep 04 18:01:51 that's a different one Sep 04 18:01:57 yeah that one conflicts with the IIO one Sep 04 18:02:07 so you need to disable the INPUT one if you want to use the IIO one Sep 04 18:02:09 aha! so having that one enable makes the iio one disappear Sep 04 18:02:12 magick Sep 04 18:02:22 i'll try the input one first, maybe that'll do Sep 04 18:33:24 crap, i'll kill the input one. ;) Sep 04 19:59:26 zmatt do you have Twitter? Sep 04 20:00:13 technically yes, but I don't use it Sep 04 21:45:18 hmm, ctypes.sh has some bugs on armhf (or 32-bit platforms in general I guess) Sep 05 00:24:59 mawk! Sep 05 00:25:15 Do you know if falstad has a push button? Sep 05 00:25:30 Or...should I just use the switch mechanism? Sep 05 00:25:51 dang it. Brb. Sep 05 00:30:25 Back in bbblack! Sep 05 00:33:36 Forget it. I found it! Sep 05 01:03:18 http://tinyurl.com/y4oxb6m2 is the circuit to the BeagleBone Black using a OptoCoupler, a resistor, and some power source on the opposite side of the BBB! Sep 05 01:04:29 Oh and it has a switch too. So, the switch is normally open. Is there a way to place the BBB in that diagram on falstad? Sep 05 01:05:38 Where the open ended section is on the right side is where the BBB should be placed. Sep 05 01:31:56 hi set_ Sep 05 01:32:01 yes there is a push button Sep 05 01:32:03 set it to temporary Sep 05 01:32:23 you can't place the BBB, simulate the BBB GPIO with a switch connected to either ground or 3.3V set_ Sep 05 01:32:31 it says "you don't have permission" this is odd Sep 05 01:32:55 can you just copy paste the link set_ ? tinyurl is terrible it seems Sep 05 01:38:13 Okay. Sep 05 01:38:21 Sure. Please hold. Sep 05 01:39:22 Nope. I closed the page. I cannot retrieve it. Sep 05 01:42:32 mawk: Would you like me to restart the diagram? Sep 05 01:45:30 While I wait for your answer, I will just try to recreate the diagram circuit. Sep 05 01:45:50 no Sep 05 01:45:53 just give me the link Sep 05 01:45:57 in the menu click export as link Sep 05 01:46:00 and paste it Sep 05 01:46:09 instead of using pastebin Sep 05 01:46:45 poor set_ Sep 05 01:46:49 you're recreating it for nothing Sep 05 01:49:35 well it's not a completely wasted effort I suppose. Sep 05 01:50:18 I closed the page. Sep 05 01:50:32 I have to recreate it. I am almost done. Sep 05 01:50:45 I do not have that page open anymore. I am almost done. Sep 05 01:50:48 Please wait. Sep 05 01:51:08 www.falstad.com/circuit/circuitjs.html Sep 05 01:51:10 Like this? Sep 05 01:51:21 nooo Sep 05 01:51:22 in the menu Sep 05 01:51:29 you have an "export as link" button set_ Sep 05 01:51:31 use that Sep 05 01:51:38 it gives you a long link to copy/paste Sep 05 01:51:47 http://www.falstad.com/circuit/circuitjs.html?cct=$+1+0.000005+10.20027730826997+50+5+50%0Ar+752+288+1056+288+0+2200%0As+576+288+752+288+0+1+true%0AR+576+288+576+368+0+0+40+5+0+0+0.5%0Ag+576+352+832+352+0%0Aw+832+352+1056+352+0%0A407+1056+288+1088+288+1%0Aw+1056+352+1056+320+0%0Aw+1152+288+1456+288+0%0Aw+1456+288+1456+576+0%0Aw+1152+320+1312+320+0%0Aw+1312+320+1312+576+0%0A Sep 05 01:51:50 Sorry. Sep 05 01:51:50 thanks Sep 05 01:51:55 You are welcome. Sep 05 01:52:25 wow your diagram is od Sep 05 01:52:26 d Sep 05 01:52:30 Oh? Sep 05 01:52:38 hmm Sep 05 01:52:45 you're using the optocoupler ? this is intended ? Sep 05 01:52:50 Yes! Sep 05 01:52:52 ok good Sep 05 01:52:56 Optocoupler for a push button. Sep 05 01:52:56 a bit less odd, but still Sep 05 01:53:00 you have a broken connection Sep 05 01:53:00 Ha. Sep 05 01:53:03 I know. Sep 05 01:53:09 is there any reason why you want an optocoupler ? Sep 05 01:53:33 Yes. If I put in a 5v supply later, I do not want it near my BBB's GPIO pins. Sep 05 01:53:54 you don't strictly need a pullup here because the led won't flash when floating because it needs substantial current, but it's always good in practice to keep in mind Sep 05 01:54:05 Okay. Sep 05 01:54:11 I don't understand the supply thing Sep 05 01:54:18 you don't have to connect the button to 5V Sep 05 01:54:23 I tried 5v and GND. Sep 05 01:54:33 there is no 3.3V pin on the bbb ? Sep 05 01:54:36 I slightly doubt it Sep 05 01:54:43 There is. Sep 05 01:54:48 yeah, so you can use it Sep 05 01:54:48 GPIO is 3.3v. Sep 05 01:54:54 yeah Sep 05 01:54:56 but let's pretend you need an optocoupler Sep 05 01:54:58 it's a good exercise Sep 05 01:55:01 Okay. Great! Sep 05 01:55:04 Nice! Sep 05 01:55:29 ok so you have a (photo)transistor in the optocoupler, you need to manipulate it correctly Sep 05 01:55:39 Right! Sep 05 01:55:40 you want it in some state when the led is flashing, in some other state when the led isn't flashing Sep 05 01:55:48 Okay. Sep 05 01:56:13 Right. The phototransistor is either lit and showing connection or not and closed. Sep 05 01:56:19 I don't know what the canonical circuit is for the transistor, you could just use it as a grounding device Sep 05 01:56:25 let me show you Sep 05 01:56:29 Okay. Sep 05 01:56:58 I have a transistor. My optocoupler is six pin and this version I have found only shows four pin. Sep 05 01:59:14 http://www.falstad.com/circuit/circuitjs.html?cct=$+1+0.000005+10.20027730826997+50+5+50%0Ar+768+352+1072+352+0+2200%0As+592+352+768+352+0+1+true%0AR+592+352+592+304+0+0+40+5+0+0+0.5%0Ag+848+416+848+512+0%0Aw+848+416+1072+416+0%0A407+1072+352+1104+352+1%0Aw+1072+416+1072+384+0%0Ag+1168+400+1168+416+0%0AR+1168+288+1168+240+0+0+40+3.3+0+0+0.5%0Ar+1168+288+1168+352+0+1000%0Ap+1280+352+1280+400+1+0%0Aw+1168+352+1280+352+0%0Aw+1168+384+1168+400+0%0Aw+1168+400+1280+400+0 Sep 05 01:59:15 %0A Sep 05 01:59:20 slightly too long Sep 05 01:59:22 let me shorten it Sep 05 01:59:25 Okay. Sep 05 01:59:54 set_: http://bit.ly/2lyjcJG Sep 05 02:00:16 the logic is inverted for the right side but you shouldn't care Sep 05 02:00:29 a read of HIGH will mean the switch is open, a read of LOW will mean the switch is closed Sep 05 02:00:47 you could increase that 1K on the right side, it's just a sample value Sep 05 02:01:22 but anyway yeah you get your voltage reading now, as long as the LED is polarized strongly enough (eg if your real world optocoupler can work with the big 2.2k resistance you put) Sep 05 02:02:11 Okay. I got what you are doing here. The circuit I am making is from a book. This book just wants me to read the change in 0 or 1 in sysfs depending on if the push button has been pressed or released. Sep 05 02:02:12 ... Sep 05 02:02:23 Is there a way to perform this function in falstad? Sep 05 02:02:39 I just showed you Sep 05 02:02:44 Oh. Sep 05 02:02:45 Okay. Sep 05 02:02:45 just look at the voltmeter Sep 05 02:02:46 Nice! Sep 05 02:02:48 almost 3.3V = 1 Sep 05 02:02:50 almsot 0V = 0 Sep 05 02:02:56 I saw that. Sep 05 02:02:59 I can put a schmitt trigger if you want a more clear separation between 0 and 1 Sep 05 02:03:34 No issue. You do not need to do that or anything else. Sep 05 02:03:39 Thank you. Sep 05 02:04:23 I would have to read about the schmitt trigger anyway. Sep 05 02:04:47 set_: http://bit.ly/2k3DwSU Sep 05 02:04:58 I put a graph too Sep 05 02:05:03 showing you the reading Sep 05 02:05:18 the schmitt trigger is inside the BBB, you don't have to do it yourself Sep 05 02:05:20 just know it exists Sep 05 02:05:27 (I think ? on stm32 there are schmitt triggers) Sep 05 02:05:51 Oh. Sep 05 02:06:07 Okay. Sep 05 02:06:52 Hey. I think the diagram is listed differently. I have a photo. I tried imgur the other day but the people have a sign in now. Sep 05 02:07:06 I just want to post. Sheesh. Sep 05 02:12:44 try my site Sep 05 02:12:48 https://pix.watch Sep 05 02:17:02 https://docs.google.com/document/d/17mvBaXYBprDzcaz-soPx9RUxSed2tDw4BY0UIrAJmOg/edit?usp=sharing. There. Sep 05 02:17:19 That is from the Molloy book, "Exploring BeagleBone." Sep 05 02:17:36 I goofed up and got the incorrect optocoupler. Sep 05 02:18:39 any should kinda work Sep 05 02:18:42 don't need the exact reference Sep 05 02:19:10 I don't like this circuit set_ Sep 05 02:19:16 it's floating when the transistor isn't polarized Sep 05 02:19:19 Okay. I have a six pin while the photo referenced is four pin. Sep 05 02:19:21 Oh. Sep 05 02:19:22 would work with the internal pullup Sep 05 02:19:31 ah yeah they mention it, ok nvm Sep 05 02:19:31 Yep. Sep 05 02:19:54 So, if the internal pull-up was not there, I would just need to add another resistor? Sep 05 02:19:56 the GPIO output will be 0.6V less than the 3.3V rail, so 2.7V, and it's above 2.2V which is the CMOS voltage limit Sep 05 02:19:57 but it's close Sep 05 02:20:01 my circuit doesn't have this limitation Sep 05 02:20:12 yes resistor to ground on the emitter Sep 05 02:20:12 Oh. Sep 05 02:20:16 Okay. Sep 05 02:20:38 I have to go, you can leave questions and I'll answer later Sep 05 02:20:39 bye ! Sep 05 02:20:42 Hey. You say emitter, are you discussing the optocoupler? Sep 05 02:20:43 Later! **** ENDING LOGGING AT Thu Sep 05 03:00:58 2019