**** BEGIN LOGGING AT Mon May 23 02:59:58 2016 May 23 06:21:38 I think people widely overestimate the usefulness of a TPM in the typical (i.e. insecure) way it is integrated into the overall system May 23 06:25:46 hi, I want to run my script.sh file from crontab. It is successfully running standalone but not through crontab -e . I have googled it and almost have done every possible solution mentioned there. I have set environment variable, resolved permission issue. But still it is not working. where I am lacking? May 23 06:26:08 have you checked journal for errors? May 23 06:26:32 btw making it a systemd service would probably make a good alternative May 23 06:26:51 those offer a lot more control over the environment in which it is run May 23 06:28:15 (you'd first make a service file for the script, verify it works when you manually start it, and then make a timer unit file for it) May 23 06:31:54 ok..thanku for your suggestion... i am working on it... May 23 07:18:53 hey there, i'd like to ask how many cores are there on BBB? May 23 07:18:57 zmatt: have you written a systemd unit to go to the store and do your work for you yet? May 23 07:19:12 randomdude_: single core for the am335x May 23 07:19:24 depends on how you count "cores" May 23 07:19:38 ohh.... because i'd like to implement a hypervisor on BBB May 23 07:19:55 since it has only 1 core, i cant really implement my project on this platform May 23 07:19:55 ok its got one cpu a7 for runnign linux .. and an m3 core for supervisor functions May 23 07:20:19 yeahno. you -could- but it wouldn't be usable :D May 23 07:20:34 but it would be a challenge probably May 23 07:20:34 i'm needd to have at least 2 cores, 1 core for linux another for freeRTOS May 23 07:20:39 yeah May 23 07:20:49 searching for a platform that uses ARMv7 May 23 07:21:00 i cant use Rpi2 because it doesnt have GIC in it May 23 07:22:01 randomdude_: check out imx6 series .. wandboard / panda (iirc) / others May 23 07:22:19 pi3 might work .. or there are allwinner cpu's .. m3 is quad May 23 07:23:14 thanks, i'll chech it out :) May 23 07:23:33 check* May 23 07:23:45 i've a pi3 May 23 07:23:54 not sure whether GIC's working May 23 07:28:52 eyyyy May 23 07:34:17 make install May 23 07:34:22 the best thing ever May 23 07:35:54 i figured doing an LFS run on my desktop was a good way to learn more about linux... and eventually take a look at how it works on ARM May 23 08:12:30 makh: http://gerbil.xs4all.nl/hello-systemd.tar.gz May 23 08:13:28 veremit: annoyingly if TI left trustzone freely usable (as it is on the rpi) then the cortex-a8 would be a suitable platform May 23 08:14:13 hypervisor in secure privileged, rtos in secure user, linux in public privileged+user May 23 08:15:14 fiq can be redirected to hypervisor (and even is on GP devices, even though there it ends in a deadloop) May 23 08:15:30 very fast context switching May 23 08:17:56 oh, he claims to need GIC for some unspecified reason May 23 08:25:30 I'm doing some boot time optimizations on BBB and i considered using initramfs creates longer boot time, is it safe to just ignore initramfs on rcn kernels? my kernel (4.4.10) May 23 08:25:58 kernel doesn't understand UUID afaik May 23 08:26:16 and stable numbering of mmc block devices seems to have only been introduced in 4.6 kernels May 23 08:27:19 but at least 4.6 seems to boot fine without initramfs May 23 08:27:36 (thanks to ds2 for making me realize there's really no need for initramfs here) May 23 08:32:42 thanks zmatt: but i didn't understand completely, In my case can i remove it from boot process? May 23 08:33:07 I deliberately made my example *not* start at boot May 23 08:33:24 you start the timer manually (systemctl start hello-world.timer) May 23 08:33:32 and then it'll run the service every 10 seconds May 23 08:33:54 to enable the timer at boot you need to uncomment one line in hello-world.timer (in the [Install] section) May 23 08:34:14 read the systemd.timer manpage, there are lots of ways to configure the timer May 23 08:35:19 and even more ways to configure the service ;) May 23 08:45:57 shiny, looks like there's good support for doing DMA from userspace in the works, complete with the ability to map a buffer as cacheable and then use ioctls to perform the appropriate manual cache coherence ops May 23 08:46:36 yay voor graphics people, although that's probably not what I'd use it for O:) May 23 08:47:47 https://software.intel.com/en-us/blogs/2016/05/13/sharing-cpu-and-gpu-buffers-on-linux May 23 08:51:22 although you quickly get buried knee-deep in acronym-hell May 23 09:03:56 what's the usage of "uuid" in /boot/uEnv.txt , it's not referenced anywhere during boot ! May 23 10:49:37 hi, I just have booted Arch Linux distro (4.6.0-3-ARCH kernel), and can't enable UART access. May 23 10:50:13 I have tried adding "capemgr.enable_partno=BB-UART4" in /boot/uEnv.txt but nothing happens May 23 10:50:51 I have also installed "core/linux-am33x" and "core/linux-am33x-headers" packages May 23 10:51:05 but still no /dev/ttyO* files May 23 10:51:15 Any ideas how to enable them? May 23 11:18:46 anyone? May 23 11:32:43 danbilzerian: are you sure you edited the active uEnv.txt .. what does cat /proc/cmdline say? and are you sure the omap-serial is the one you're looking for (ie. ttyOx) and not the new 8250 driver which will be ttySx? May 23 11:33:17 for (1) you should be able to load the UART on the command-line, and see if it fails in 'dmesg' May 23 11:36:19 @veremit the output of "cat /proc/cmdline": console=ttyO0,115200n8 coherent_pool=1M capemgr.enable_partno=BB-UART4 root=PARTUUID=08c01409-01 rw rootwait fixrtc May 23 11:38:09 veremit I think it's omap-serial, at least this was in the old Debian distro May 23 11:38:35 I'm thinking that the 4.6 kernel it's unusual to be using omap... May 23 11:39:11 can you pastebin/bpaste your dmesg response? May 23 11:40:29 I can't because the old command "echo BB-UART1 > /sys/devices/bone_capemgr.9/slots" May 23 11:40:36 won't work here May 23 11:40:48 "/sys/devices/bone_capemgr.9/slots: No such file or directory" May 23 11:41:15 there's nothing starting with "bone" in /sys/devices May 23 11:45:21 yes I was wondering whether the capemgr was present in that kernel .. May 23 11:45:49 so this means that capemgr is not present? May 23 11:46:43 quite possibly .. when you deviate from the official images, you have to be sure you're copying the BBB specific patches over if you choose a different distro. Sure it will work, but not necessarily with all the functions May 23 11:47:19 cape manager has come and gone a lot in the last ~major version of kernel May 23 11:47:25 I don't think its part of mainline May 23 11:47:57 you can do it 'new-school' via device-tree fairly easily May 23 11:48:39 can you please give me some tutorial? May 23 11:52:57 here's the dmesg log: https://bpaste.net/show/151d0dfc08da May 23 11:54:05 capemgr lives at /sys/devices/platform/bone_capemgr/ May 23 11:54:08 or something ike that May 23 11:54:23 but there's also a newer mechanism May 23 11:55:01 ah weird .. there is a cape_mgr loading .. May 23 11:55:27 capemgr is afaik part of all standard rcn kernels and DTs May 23 11:55:28 zmatt: I also tried "echo BB-UART1 > /sys/devices/platform/bone_capemgr/slots", same result May 23 11:55:50 zmatt: didn't think it went mainline .. and this is an Arch kernel .. apparently. May 23 11:56:00 veremit: they use rcn too May 23 11:56:07 danbilzerian: you sure it isn't BBB- May 23 11:56:18 zmatt: ah thats ok then May 23 11:56:39 danbilzerian: can you pastebin the contents of slots ? May 23 11:56:49 it's BBB rev C May 23 11:56:52 just a sec May 23 11:57:12 and grep -vP '^#|^$' /boot/uEnv.txt May 23 11:57:31 zmatt: nice grep May 23 11:58:21 zmatt: https://bpaste.net/show/e42de7322b8b May 23 11:58:22 I also fairly recently made some utilities that make using the new configfs mechanism easier (instead of abusing capemgr for non-CAPEs), and also building overlays much nicer May 23 11:58:47 http://pastebin.com/Mduupacd here I gave someone an example to enable pwm1 May 23 11:58:50 (both outputs) May 23 11:59:32 ok, so the uart looks loaded May 23 11:59:39 the overlay May 23 11:59:43 zmatt: https://bpaste.net/show/bfb01bddf720 May 23 12:00:09 ah thats uart1 though May 23 12:00:14 ohwait .. haha May 23 12:00:20 I doubt that's the only enabled line from that grep May 23 12:00:34 also, optargs? what? May 23 12:00:49 oh wait, arch, might be doing weird things with bootloader May 23 12:00:51 that was there by default May 23 12:01:04 I only added "capemgr.enable_partno=BB-UART4" May 23 12:01:07 not on any debian system May 23 12:01:13 ok so loading uart1 manually seems to work .. May 23 12:01:21 can you manually load uart4?? May 23 12:02:00 I loaded it now May 23 12:02:04 do you actually have a capemgr startup service that pais attention to "capemgr.yada" directives in your kernel cmdline? because I don't think the driver itself does May 23 12:02:08 it's present in *slots file "5: P-O-L- 1 Override Board Name,00A0,Override Manuf,BB-UART4" May 23 12:02:50 zmatt: I don't think I have May 23 12:04:17 note that CAPEmgr really has one goal: to automatically detect and identify CAPEs via i²c-2 and load their associated overlays to make CAPEs (in theory) work "plug and play" May 23 12:04:46 using it to load arbitrary overlays is technically abuse, though for quite a while there was no alternative way hence it's become a habit May 23 12:04:57 configfs is the way to go May 23 12:05:18 hmm, I think it's working now May 23 12:05:29 moreover, if you don't have any *actual* CAPEs you can disable capemgr entirely and free up the two pins it reserves for i2c May 23 12:05:30 I can use UART4 via /dev/ttyS4 now May 23 12:06:37 zmatt: you always were able to force-load overlays .. like UARTs .. it worked before .. should work now May 23 12:06:48 danbilzerian: ok so it seems to be a boot issue May 23 12:07:44 veremit: would be ok if I make a sh script that automatically runs this commands at boottime? May 23 12:08:34 veremit: it still works yes May 23 12:08:48 veremit: but it's kind of deprecated since the new mechanism May 23 12:09:09 like, capemgr also requires that you invent bogus metadata (cape name and version number) for your "CAPE" May 23 12:09:27 and it gets a "slot" (albeit one invalid for any real cape) May 23 12:09:38 configfs has none of that nonsense May 23 12:09:53 zmatt: yes, we know its a fudge .. but it was a -working- fudge :P esp for newbies :D May 23 12:10:12 as long as they can find an overlay that does *exactly* what they want May 23 12:10:18 since writing those things is horror May 23 12:10:25 zmatt: its an efff'ing uart .. May 23 12:10:44 that doesn't make pinmux constants less magical :P May 23 12:11:06 zmatt: just cos you understand DT and pinmuxes backwards :P doesnt mean everyone else does May 23 12:11:07 or the overlay syntax less revolting (and really without any good reason) May 23 12:11:14 that's my point May 23 12:11:23 *I* still haven't written any DT ... lol May 23 12:11:24 the old way was pulling magic hex numbers out of your ass May 23 12:11:36 have you seen what it looks like with my toolage? http://pastebin.com/Mduupacd May 23 12:12:21 positively pleasant!? :D May 23 12:14:34 ok RN says that cmdline still should work in 4.1, 4.4 kernels via https://github.com/beagleboard/bb.org-overlays May 23 12:15:17 or if you prefer syntax highlighting http://gerbil.xs4all.nl/pwm1.dtsi.html May 23 12:15:34 yeah, assuming you have his startup scripts May 23 12:15:48 ah those are there May 23 12:16:35 zmatt: *gasp* .. https://github.com/beagleboard/bb.org-overlays/blob/master/src/arm/BB-UART4-RS485-00A0.dts May 23 12:17:51 ok so they still live in /lib/firmware right ... May 23 12:19:13 ugh I need to get my bbb permanently powered May 23 12:19:15 but not today May 23 12:19:24 http://gerbil.xs4all.nl/uart4.dtsi.html similar source code for uart4 May 23 12:20:17 if the rts485 stuff matters to you, you could obviously add those to the overlay May 23 12:20:37 zmatt: that's horribly simple .. why?! :P May 23 12:20:41 at least rcn uses cpp early May 23 12:20:46 *at least May 23 12:20:51 most people don't even do that May 23 12:21:22 because DT *is* horribly simple... but it was designed to be generated and consumed by machines, not by people May 23 12:21:43 zmatt: then why all the bikeshedding?! lol May 23 12:21:52 -shrug- May 23 12:22:00 hence the ugliness, redundancy, fussiness, lack of meaningful error reporting etc May 23 12:23:02 the differences between rcn's and my version btw (aparety from the rs485 properties) is 1. different choice of macro names/implementations 2. he uses pointlessly verbose identifiers here and there May 23 12:23:08 and most importantly May 23 12:23:58 I have a 50-line or so perl script that converts fragments written in normal dtsi syntax into the horror that the overlay mechanism demands May 23 12:25:34 but right now I can't talk, got stuff to do NOW... will be back in probably bunch of hours May 23 12:25:42 lol have fun May 23 18:39:21 Anyone know what package I need to add to the Jessie Console image to have my BBB available as a serial port on the PC. It works with the IOT images and LXQT. May 23 18:43:14 I mean serial over USB -- i.e. I want it to show up as Gadget Serial port COMx on my PC when I connect the USB to my PC. May 23 18:46:10 Geof: I think you want the g_serial kernel module May 23 18:46:39 So apt-get install g_serial? May 23 18:47:03 it should be in /lib/modules/`uname -r`/kernel/drivers/usb/gadget/ already May 23 18:47:51 oh, it might already be part of the gadget bundle May 23 18:47:56 do you have a /dev/ttyGS0? May 23 18:48:36 No /dev/ttyGS0 -- that's what I'm trying to start up. I think it's a service May 23 18:50:09 so g_multi doesn't show as loaded from lsmod? May 23 18:50:36 No, just g_ether May 23 18:54:16 hm, I have a /opt/scripts/boot/am335x_evm.sh I think that is what is loading g_multi for me May 23 18:54:27 but it didn't come from a package May 23 19:01:37 So you're implying that it's built into the kernel -- and not installable at run-time? May 23 19:02:01 not into the kernel, it looks like a boot script that runs at init May 23 19:02:21 not sure how it all fits together, but it has logic to load the various gadget modules May 23 19:03:33 Yes, and if I read that logic correctly, then if the modprobe works, it executes a systemctl start serial-getty@ttyGS0.service May 23 19:03:56 But what if the modprobe fails -- I think that means that the module was not included in the kernel build. May 23 19:04:27 And I'm not sure if a kernel module can be added at runtime. that's where I'm pretty novice. May 23 19:05:42 modules are parts of the kernel that can get loaded at runtime May 23 19:06:00 I added my kernel module at run time. I don't use modprobe, though, because it seemed more wonky to load/unload on demand. May 23 19:06:33 Aha. So what command would load the g_multi module at runtime? I think that might solve my problem May 23 19:06:39 Geof: it sounds like that script decided to load g_ether instead of g_multi for you May 23 19:06:54 I use insmod, but if you have a script I'd follow what it does. May 23 19:07:02 first you have to unload g_ether as you can only have one gadget loaded at a time May 23 19:08:31 Ok, so I find the script that installed g_ether and figure out why it didn't install g_multi. I'll go work on that for a little while. do you know what script that might be? May 23 19:08:50 try /opt/scripts/boot/am335x_evm.sh May 23 19:09:05 looks like /etc/init.d/boot_scripts.sh calls it May 23 19:22:29 ll May 23 19:55:50 Looks to me that if the file /var/local/usb_mass_storage.img exists, then it loads g_multi. I'll play with that... May 23 19:57:57 Actually that likely means a bunch of other stuff is needed. So for now, I'll stick to using the actual uart0 console port May 23 19:58:22 ddrown: Thanks for your help. I'm good for now. May 23 19:58:40 Geof: cool May 23 19:59:35 it looks like it first checks for /var/local/usb_mass_storage.img, then tries to find the root filesystem drive May 23 20:00:15 if the root filesystem drive isn't /dev/mmcblk0p1 or /dev/mmcblk1p1, then it uses g_multi May 23 20:07:03 Ok, thanks. May 23 23:01:32 https://github.com/gngeorgiev/jshell javascript based shell **** ENDING LOGGING AT Tue May 24 02:59:58 2016