**** BEGIN LOGGING AT Thu Dec 10 02:59:56 2020 Dec 10 03:23:31 * GenTooMan listens to hands miss each other while trying to applaud. Dec 10 03:50:39 10 minutes! Dec 10 03:51:01 I am going to type up a PWM lib. Wish me luck? Dec 10 03:51:30 Now, if I could only figure out wrappers for Python - hint, hint - that is next! Dec 10 03:52:17 I enlisted on github.com under some heading I found for BBIO! Dec 10 03:52:24 They need help. Dec 10 03:52:37 I am just that top hatter to fix an issue... Dec 10 03:53:52 Anyway, I think I can. So, I got the rtos, some .h files, and includes, some pin muxing via the tool SYSCONFIG and I am going to try. Someone already reached climax but I am going to give it the ole heav-hoe! Dec 10 03:54:58 I have been reading ch. 15 on the TRM for the am335x. I am in good standing w/ notes thus far and some errors to conclude on. Dec 10 03:55:56 GenTooMan: I am slow, real slow. You will not like what i say or how I say it but, but, I can doo-eee! Dec 10 03:56:48 I just wanted to do output. Now, I think I have to do all the entire set up, syncin and syncout and so forth. eCap and eQep. and and and. Dec 10 03:56:52 Aw! Dec 10 03:58:24 Really, I think this will be good for me, loss or win, lose or fail, or whatever. Dec 10 03:59:06 10:00! Dec 10 03:59:15 Peace be w/ you and Godspeed! Dec 10 09:47:53 so this https://github.com/torvalds/linux/blob/master/drivers/power/supply/tps65217_charger.c#L146 is supposed to be what generated the udev event for the charger right? Dec 10 09:48:25 I don't see any udev event with 'udevadm monitor -pk' when I plug in or unplug the USB Dec 10 09:49:29 that charger driver seems to be running, it has directories in /sys/firmware/devicetree and /sys/devices/platform/ocp/ Dec 10 09:49:47 its not listed explicitly in the /proc/interrupts though Dec 10 09:50:39 the device's battery charges Dec 10 12:36:07 Hi I have a BB-AI. It's meant to have 1GB RAM. If i run free -m from terminal session its shows I have 577M RAM. Where is the other 423M? Dec 10 12:36:29 it's running the stock image with updates. Dec 10 15:50:24 Hmm, is it possible to add a remote endpoint to : https://github.com/beagleboard/linux/blob/5.4/Documentation/devicetree/bindings/usb/am33xx-usb.txt Dec 10 15:50:38 as described in:https://github.com/beagleboard/linux/blob/5.4/Documentation/devicetree/bindings/connector/usb-connector.txt Dec 10 15:55:43 as a clarification, I am getting a OF: graph: no port node found in /ocp/interconnect@48000000/segment@0/target-module@2a000/i2c@0/typec-portc@54 error when trying to istantiate a fusb302 driver Dec 10 15:56:54 I think it's due to me not having an endpoint to the connector described as here: https://github.com/beagleboard/linux/blob/5.4/Documentation/devicetree/bindings/usb/fcs%2Cfusb302.txt Dec 10 16:27:27 dom22: Anyone got a BeagleBone AI? Can you do a "free -m" and tell me how much memory it shows? Dec 10 19:21:27 Anyone has any experience with the wifi on beagleboneblack, I tried with connman by following this : http://beagleboard.org/static/librobotcontrol/networking_wifi.html Dec 10 19:21:33 it did get connected once and twice Dec 10 19:21:42 but not getting connected now Dec 10 19:21:52 everytime it boots I used to follow this Dec 10 19:21:57 is there a permanent way Dec 10 19:22:20 my device : Bus 001 Device 003: ID 0846:9021 NetGear, Inc. WNA3100M(v1) Wireless-N 300 [Realtek RTL8192CU] Dec 10 20:38:56 looking at the TRM it seems that only CLKSPIREF max freq is 48MHz, shouldn't this always be 48MHz? Dec 10 21:20:09 mm302: what do you mean? Dec 10 21:23:25 PER_CLKOUTM2 is normally always 192 MHz yes (hence PER_CLKOUTM2 / 4 is then 48 MHz) Dec 10 21:25:31 of course that doesn't mean you can't configure it to a lower rate, it just means that there might be things that break if you do Dec 10 21:28:15 yeah, I don't want to change and break things, I have no idea if this is configurable, just noticed table 24.2.2 "McSPI Clock and Reset Management" only mentions Max Freq=48 MHz, I'll assume this is the default Dec 10 21:32:23 basically I'm running my adc at 1.45MHz divising this by the maximum possible divisor Dec 10 21:32:30 apparently PER_CLKOUTM2 is lowered to 96 kHz in OPP50 (hence then the mcspi gets 24 MHz instead of 48 MHz) .. but you probably have no reason to care about OPP50 hence you can ignore this Dec 10 21:33:31 oh interesting trick, so many loopholes and caveats Dec 10 21:33:50 thanks Matt Dec 10 21:38:55 are RAM and ROM the only memory regions on the AM3358 that are executable? Dec 10 21:39:47 i figured that any program binary on flash/emmc was copied into RAM before being executed but i guess i could be wrong Dec 10 21:45:36 I belive instructions will always be loaded in RAM to be executed, PRU has IRAM memory for it for example Dec 10 21:50:19 thanks m302, i just assumed its only RAM and ROM Dec 10 21:50:50 and that the ROM bootloader copies an +X binary into some special kind of RAM (or maybe some small amount of SRAM? I need to look it up) before pointing htere Dec 10 21:55:27 there's no real difference between instruction fetch and data read, so anything in the address space that's readable is "executable" in the sense that you can attempt to execute from it (ignoring whether this is a good idea. in case of peripheral registers it probably isn't :P) Dec 10 21:57:29 typically you'd execute code only from on-chip SRAM, on-chip ROM (bootrom), external DDR ram (via EMIF), and external memory (e.g. NOR flash) connected via GPMC (other than NAND which doesn't support direct execution and requires copying to ram) Dec 10 21:59:16 bootrom will load the next stage (e.g. u-boot SPL) into on-chip SRAM Dec 10 22:05:17 (the MPU subsystem private SRAM (0x402f0000-0x402fffff) and the OCMC SRAM (0x40300000-0x4030ffff), which are 64K each but they're directly adjecent in the cortex-a8 physical address space so it just treats them as if it were a single continuous 128K memory. the first 1K of mpu subsystem sram is reserved and inaccessible hence the boot image is placed at 0x402f0400 (max 109K, the rest is used by ... Dec 10 22:05:23 ...bootrom while it's doing its thing) Dec 10 22:06:41 Can any block really be executed? My understanding is that even if there is legit code in some memory most blocks would cause a hardfault Dec 10 22:08:45 hardfault is a cortex-m concept Dec 10 22:08:53 ah Dec 10 22:09:06 and you probably meant memmanage fault Dec 10 22:09:11 or whatever the thing is called Dec 10 22:10:46 i mean i just remember certain chips had addressable external NOR that was or wasn't executable Dec 10 22:12:10 that or the cortex-A equivalent is the result of attempting to execute from a region of memory that is marked as "non-executable". on cortex-m which parts of memory are non-executable can either be fixed (if it's a cortex-m without memory protection unit) or almost entirely configurable by software (if it has a memory protection unit), except the PPB is always non-executable iirc Dec 10 22:12:47 on cortex-A which parts of memory are non-executable is entirely defined by the MMU translation tables set up by software Dec 10 22:13:52 So I just wrote my "What is a program counter", I did "what is a memory map" I'm going to do "What are interrupts", "How do peripheral registers work", "What is priviledged execution", "What is the difference between userspace and kernelspace" Dec 10 22:13:56 and THEN Dec 10 22:13:56 we're doing py-uio Dec 10 22:13:58 ("device memory" is implicitly non-executable, but of course nothing obligates you to configure the memory region corresponding to some device to be "device memory") Dec 10 22:15:43 results from "What is a memory map" post: http://i.imgur.com/3VLFhtk.png Dec 10 22:15:58 if i get one or two people that want to learn about py-uio at the end of this, i'd be super pumped Dec 10 22:40:04 so it's not true that using the local cpu ram like iram from the pru would be needed to run instructions? Dec 10 22:41:11 I'm talking about the cortex-a8. pru can only execute from its iram (which is tightly coupled to the cpu) because it requires extremely low access latency Dec 10 22:42:24 the cortex-a8 doesn't give a shit what exactly it's fetching instructions from and how long it takes Dec 10 22:43:17 oh, so it loads one instruction at the time instead of caching a block? Dec 10 22:43:25 no Dec 10 22:43:47 I didn't say anything about the chunk size in which it reads instructions Dec 10 22:46:07 if it fetches more then one instruction at the time it means it uses a local ram/cache, I'm not sure about cortex, but I vaguely remember x86 doing some local caching like that and using a pipeline Dec 10 22:47:37 normally you're always executing from cacheable memory (and have instruction cache enabled) since doing otherwise is performance suicide Dec 10 22:48:48 though even when executing from uncacheable memory its instructions fetches are up to 16 bytes Dec 10 22:49:02 up to 64 bytes sorry Dec 10 22:49:16 same as cache line size Dec 10 22:50:32 rings a bell, thanks Dec 10 22:56:15 my guess is that when executing from non-cacheable memory (typically only during early initialization before caches have been enabled) it just behaves very similar to cacheable memory (except for not actually caching) to minimize the amount of complexity spent on something that's almost never used Dec 10 23:01:24 (a fetch per instruction would make no senes for many reasons, including that the cortex-a8 can execute two instructions simultaneously) Dec 10 23:07:26 night, see you Dec 10 23:11:01 i'm totally gonna set up 2FA on ssh for fun Dec 10 23:13:31 Hey ayjay_t: Are you making tutorials for the am335x to rely info. from the TRM to the userspace land? Dec 10 23:13:55 i mean like i siad everythinggggg i'm doing is on FB Dec 10 23:14:09 I understand. Okay. Dec 10 23:14:16 but yeah we're going to be doing like py-uio stuff eventually Dec 10 23:14:22 probably breaking things. Dec 10 23:15:00 I am not interested in py-uio yet. i am still trying to configure TRM related registers and defines in C/C++ for use in source. Dec 10 23:15:22 There is still a lot I am missing. So, I am moving slow. Dec 10 23:16:56 probably should be interesting in py-uio Dec 10 23:16:58 but its up to you Dec 10 23:17:57 Okay. I just saw earlier that you were trying to make up some articles or videos on the am335x and "peripheral registers" and other videos. Is that correct? Dec 10 23:18:19 I will go and sign up. Dec 10 23:18:23 Facebook. Okay. Dec 10 23:20:01 uh we're doing some background Dec 10 23:20:04 but the explanations are really short Dec 10 23:20:13 just enough background to basically understand Dec 10 23:20:42 it should be fine Dec 10 23:21:08 Okay. No issue. Dec 10 23:21:17 I will sign up and sign in. Dec 10 23:21:42 oh god a new facebook user eh if you'd prefer not to, i can maybe posts gists or something eh Dec 10 23:23:25 Ha. Dec 10 23:23:39 I have used facebook before but they kicked me due to something. Dec 10 23:23:48 I am not completely sure yet why. Dec 10 23:23:56 They want to make sure i am me or something. Dec 10 23:24:07 THey wanted my id or driver license. Dec 10 23:24:08 Ha. Dec 10 23:24:42 I have not met anyone from facebook every. WHy would i give out my ID or driver license? Dec 10 23:24:45 Odd, heh? Dec 10 23:25:09 What is the link again? Dec 10 23:25:45 facebook.com/groups/BeagleBone? Dec 10 23:26:08 why would anyone every want to got that spyware site? Dec 10 23:26:51 Um, to listen to ayjay_t of course! Dec 10 23:27:09 Hello ds2: How are you? Dec 10 23:28:46 hi set_ Dec 10 23:30:04 wow what they want an ID or driver license thats crazy Dec 10 23:30:09 i mean i signed up a long time ago and they didn't ask for that wtf Dec 10 23:30:20 yeah i mean unfortunately lots of people use it Dec 10 23:30:58 that does not change that fact it is by definition spyware Dec 10 23:31:00 i'm /BeagleBoneBlack Dec 10 23:31:03 it's an unofficial channel Dec 10 23:31:59 so is /BeagleBone Dec 10 23:32:22 but we have 3.5K members to their 1.5 /shrug Dec 10 23:32:25 and we get more everyday Dec 10 23:44:46 I had signed up and then all of a sudden, they said I could not sign in and they needed my id. Dec 10 23:44:58 Looking for set_. Ha. Dec 10 23:45:07 Okay. Dec 10 23:45:36 I got lost in facebook again. Sorry. spyware. Dec 10 23:47:44 Oh yea. I remember now. They said I could not have the name de funct. Dec 11 02:11:27 Hi I have a few questions on beagleboard. Dec 11 02:54:59 Is there a beaglebone_driver_config.h file for driver configuration? Dec 11 02:56:20 For instance, outside of the libs. for Linux, there should be a specified X_X_config.h file for the different drivers? Right? Dec 11 02:57:13 Like this...one for gpio from gpio.h, another for pwm from pwm.h, and so on. Dec 11 02:58:37 or...is the overlay section and config-pin in place of this X_X_config.h file? Dec 11 02:59:32 I mean, I can see that the overlay is a smart routine b/c of the am335x being a vast chip w/ many peripherals and drivers associated w/ those peripherals. **** ENDING LOGGING AT Fri Dec 11 02:59:56 2020