**** BEGIN LOGGING AT Mon Sep 27 02:59:56 2021
Sep 27 08:00:47 <Guest2865> hi
Sep 27 08:01:49 <Guest2865> Can you help on selection of processor board which can support PCIe Gen2 X1 which can be configured as End point
Sep 27 08:10:14 <Guest2865> hi ikarso
Sep 27 16:49:07 <Guest83> hello
Sep 27 17:28:30 <Guest83> hey
Sep 27 17:28:48 <Guest83> i have some issue with beaglebone
Sep 27 18:39:22 <zmatt> noted.
Sep 27 23:19:50 <the_person48> hey guys, I am writing a dtsi boot overlay file for my beaglebone. I am doing two separate ones for different boot configurations. I got one of them working, and then was just modifying it in terms of which pins are inputs/outputs, and pulled up/down. But I am running into a weird compilation error:
Sep 27 23:20:04 <the_person48> https://pastebin.com/tdpNJBHa
Sep 27 23:20:07 <the_person48> here's my overlay
Sep 27 23:20:52 <the_person48> it's giving me an error on line 244: Error: power_board_boot_config.dtsi:244.7-8 syntax error
Sep 27 23:20:52 <the_person48> FATAL ERROR: Unable to parse input tree
Sep 27 23:21:02 <the_person48> for P9_41
Sep 27 23:21:30 <the_person48> I tried commenting that line, got the same error on the subsequent line. when I commented out line 245 as well, the error went away
Sep 27 23:21:38 <the_person48> the weird thing is I can't figure out what's different about those two
Sep 27 23:21:41 <the_person48> any ideas?
Sep 27 23:45:36 <the_person48> I'm also getting a warning (that I think is separate) about clock build issues:
Sep 27 23:45:37 <the_person48> make: Warning: File 'Makefile' has modification time 1794 s in the future
Sep 27 23:45:45 <the_person48> make: warning:  Clock skew detected.  Your build may be incomplete.
Sep 27 23:45:50 <the_person48> is that important?
Sep 27 23:48:23 <Guest15> The beaglebone ai gets incredibly hot just sitting idle.  Is there a way to tun off unused heaters?
Sep 27 23:52:24 <zmatt> Guest15: are you using TIDL ?  if not, try a non-TIDL image
Sep 27 23:53:28 <zmatt> the_person48: the warning is caused when your bbb's system time is wildly wrong because it didn't have internet access to sync its clock (it has no clock battery hence the bbb will lose system date/time when it loses power)
Sep 27 23:53:52 <the_person48> ah I see, so I'll just plug in ethernet then to fix
Sep 27 23:54:16 <zmatt> yeah, once you plug it in I think time should automatically sync
Sep 27 23:54:28 <zmatt> and time will keep across reboots, just not power-cycles
Sep 27 23:55:03 <the_person48> cool, makes sense
Sep 27 23:55:05 <the_person48> I'll try now
Sep 27 23:55:50 <zmatt> as for the error, yes dtc's reporting of syntax errors is absolutely terrible
Sep 27 23:55:57 <zmatt> and uninformative
Sep 27 23:56:43 <zmatt> but a common cause is a bogus identifier showing up in an expression, e.g. because you got the name of a macro wrong
Sep 27 23:56:59 <the_person48> I have internet now but it's still giving me the timestamp and clock skew warning
Sep 27 23:57:09 <the_person48> huh ok
Sep 27 23:57:14 <the_person48> I'll double check the macros
Sep 27 23:57:27 <zmatt> there's probably some files that now have a bogus timestamp because of files modified while the system date was wrong
Sep 27 23:57:38 <zmatt> ls -lt   will show files sorted by date
Sep 27 23:57:50 <zmatt> you can use "touch" to set the modification date of a file to current date/time
Sep 27 23:58:28 <zmatt> and the problem is P9_41 and P9_42, check the list of macros in include/bone/black.h to see why
Sep 27 23:58:45 <zmatt> your gpio number comment for P9_15b is also wrong
Sep 27 23:59:33 <the_person48> I'm not seeing a bone directory
Sep 27 23:59:35 <the_person48> and gotcha
Sep 28 00:00:04 <zmatt> well it is there, you're #including bone/black.h at the top of your file
Sep 28 00:00:22 <zmatt> https://github.com/mvduin/overlay-utils/blob/master/include/bone/black.h
Sep 28 00:00:42 <the_person48> oh I see
Sep 28 00:00:46 <the_person48> I wasn't looking in the include folder
Sep 28 00:00:55 <the_person48> also, I don't see what's wrong with the gpio number comment on P9_15b?
Sep 28 00:01:02 <zmatt> even though I explicitly said "include/bone/black.h" :P
Sep 28 00:01:16 <zmatt> "PIN_NOPULL( P9_15b, 7 ) //gpio 1.16"
Sep 28 00:01:26 <zmatt> it's 2.00
Sep 28 00:01:33 <zmatt> 1.16 is P9_15a
Sep 28 00:01:40 <the_person48> yeah my bad
Sep 28 00:01:51 <zmatt> (two separate cpu pins that both connect to P9_15)
Sep 28 00:02:10 <the_person48> ah, the problem is P9_41 and P9_42 have a and b's
Sep 28 00:02:36 <the_person48> so should I just set both like I did with P9_15a/b to be safe?
Sep 28 00:02:38 <zmatt> yeah so this is dtc's obnoxiously terrible way of saying: Unknown identifier "P9_41"
Sep 28 00:03:33 <the_person48> I see
Sep 28 00:03:35 <the_person48> will adjust
Sep 28 00:03:48 <the_person48> also, I tried touching the file but it's still giving the clock skew thing
Sep 28 00:03:53 <the_person48> will that cause problems or can I just ignore it?
Sep 28 00:03:55 <zmatt> when using P9_15, P9_41, or P9_42, I always recommend configuring pinmux for both pins of the pair even though you're probably only using one
Sep 28 00:04:32 <zmatt> there's no file with an obviously bogus date in ls -lt?  have you confirmed system time has indeed synchronized? (use "date")
Sep 28 00:05:09 <the_person48> ok will do
Sep 28 00:05:20 <the_person48> all the dates seem wrong actually
Sep 28 00:05:30 <the_person48> they all say november 3rd for some reason
Sep 28 00:05:37 <zmatt> is your system date set right?
Sep 28 00:05:39 <the_person48> and that fixed the compilation error!
Sep 28 00:05:45 <the_person48> no
Sep 28 00:05:50 <the_person48> it says novemeber 3rd haha
Sep 28 00:05:54 <zmatt> well there's your first problem
Sep 28 00:06:25 <the_person48> yeah
Sep 28 00:06:31 <the_person48> ok I'm googling how to fix the system time
Sep 28 00:06:55 <zmatt> if you have ethernet connection and are too lazy to reboot, try:   sudo systemctl try-restart systemd-timesyncd
Sep 28 00:07:14 <the_person48> I already did reboot
Sep 28 00:07:16 <the_person48> but I will try that
Sep 28 00:07:21 <zmatt> with ethernet plugged in?
Sep 28 00:07:34 <zmatt> if you did a reboot then just restarting systemd-timesyncd is pointless
Sep 28 00:07:55 <the_person48> yes, with ethernet in
Sep 28 00:08:06 <the_person48> I'll try again
Sep 28 00:10:08 <the_person48> date is still wrong...
Sep 28 00:10:17 <the_person48> does it matter, or should it work anyways?
Sep 28 00:10:21 <the_person48> maybe I just won't worry about that
Sep 28 00:10:27 <zmatt> you should
Sep 28 00:10:51 <zmatt> date being wrong means file modification dates will continue being wrong, make will continue being confused and not able to correctly do its job, etc
Sep 28 00:11:04 <the_person48> ok
Sep 28 00:11:11 <the_person48> is there any way to manually sync it?
Sep 28 00:11:16 <the_person48> not sure why it's not happening automatically
Sep 28 00:11:21 <zmatt> does the beaglebone not have internet access?
Sep 28 00:11:30 <the_person48> it does
Sep 28 00:11:35 <the_person48> I'm able to ping google.com right now
Sep 28 00:11:44 <zmatt> you must have a weird firewall that's blocking NTP or something
Sep 28 00:12:09 <the_person48> that's possible?
Sep 28 00:12:22 <the_person48> I'm not aware of any firewall on this though
Sep 28 00:12:34 <zmatt> "this" being your office network
Sep 28 00:12:40 <the_person48> no, home network
Sep 28 00:12:43 <zmatt> oh
Sep 28 00:12:43 <zmatt> huh
Sep 28 00:12:50 <zmatt> okay that's really odd
Sep 28 00:12:51 <the_person48> yeah otherwise I would highly suspect that
Sep 28 00:13:00 <the_person48> but yeah, I set up my home network so I can't think of what it would be
Sep 28 00:14:01 <zmatt> timedatectl timesync-status
Sep 28 00:14:53 <Guest15> @zmatt yeah I'm using TIDL Image.  Anyway to change, say the dtb, without reflashing.
Sep 28 00:15:38 <zmatt> Guest15: uhh there's lots of differences I think, definitely a different kernel, packages that are installed, stuff that's automatically loaded at boot, auxiliary cores that get fired up
Sep 28 00:17:26 <Guest15> Just had a look at the beaglebone download page.  I don't see a non TIDL image for am5729?
Sep 28 00:17:42 <zmatt> the_person48: P8_31-46 should be NOPULL
Sep 28 00:18:35 <zmatt> Guest15: hmm, weird that for some reason there's no SD image, only an eMMC flasher ("AM5729 Debian 10.3 2020-04-06 4GB eMMC IoT Flasher")
Sep 28 00:19:40 <Guest15> There's a console image:  AM5729 Debian 10.3 2020-04-06 1GB SD console
Sep 28 00:19:45 <Guest15> Would that work?
Sep 28 00:20:24 <Guest15> I do need the video acceleration to work too btw.  Will that happen in non TIDL images?
Sep 28 00:20:32 <zmatt> I mean, if you want a console image... they're very minimalistic images meant for people comfortable with debian packagement
Sep 28 00:21:07 <zmatt> if by "video acceleration" you mean TIDL then no, if you mean something else then no idea
Sep 28 00:22:12 <Guest15> I'm not using any UI.  I just need the sgx drivers to work.
Sep 28 00:22:22 <zmatt> oh that has nothing to do with TIDL
Sep 28 00:22:41 <zmatt> nor with video for that matter, it's a 3d gpu :P
Sep 28 00:25:32 <Guest15> Ah, good.  Thanks ;)   Do you know if the DSP cores will be off as well, and is there any way to turn off one of the cpu cores?
Sep 28 00:26:32 <zmatt> the DSPs and EVEs are powered on in the TIDL images but shouldn't be in the non-TIDL images
Sep 28 00:28:10 <zmatt> the hardware supports powering off one of the cortex-a15 cores, I don't know if linux supports it
Sep 28 00:30:30 <Guest15> echo 0 > /sys/devices/system/cpu/cpu1/online
Sep 28 00:30:39 <Guest15> cpufreq-info shows the cpu is off
Sep 28 00:30:45 <zmatt> well there you gp
Sep 28 00:30:50 <Guest15> no noticeable change in current draw though
Sep 28 00:33:46 <zmatt> not super surprised, there shouldn't be if the cpu power management is doing its job
Sep 28 00:37:16 <Guest15> Okay, well thanks again.  Gonna go burn that new image.
**** ENDING LOGGING AT Tue Sep 28 02:59:56 2021