libpruio (fast and easy D/A - I/O)

not_me · March 22, 2019, 11:53am

Scratch that, the problem occurs repetively.
After a random while the network over USB disappears and connecting to the wifi does not let me reconnect to the cloud9 ide.
I got the startup error again, but the problems occurs even if the error isn’t there.

Message from syslogd@beaglebone at Mar 21 18:33:07 …
kernel:[ 412.210610] Internal error: : 1008 [#1] PREEMPT THUMB2

kernel:[ 412.326905] Process irq/118-wl18xx (pid: 1474, stack limit = 0xdc57a210)

kernel:[ 412.333632] Stack: (0xdc57bce0 to 0xdc57c000)

kernel:[ 412.338012] bce0: c100cb70 c1004088 00000000 18000000 c011eea1 c011eea1 db53b22c c011dbb7

kernel:[ 412.346227] bd00: c100a758 00000000 c110c6bc 23d5fe02 00000000 c100cb70 00000000 c110c6bc

kernel:[ 412.354440] bd20: 18000000 c011ddb9 c1004088 c011e12d c100cb70 c100cb70 000c0013 dc28a400

kernel:[ 412.362653] bd40: 00000000 c011e133 dc28a880 00000001 00000000 c011ee6f dc286010 00000000

kernel:[ 412.370867] bd60: 00000000 c011eeb1 00000000 dc286010 00000000 c05f46f5 dc569a00 dc286010

kernel:[ 412.379082] bd80: ffffe000 00208040 00000000 c011eea1 c1004088 db53b22c 00000000 c05f47ef

kernel:[ 412.387296] bda0: dc286010 c101dc80 dc246010 c05f42a1 dc57bdb0 dc57bdb0 00000000 dc57bdbc

kernel:[ 412.395531] bdc0: dc57bdbc 23d5fe02 dc57be0c dc286010 00000004 400c0013 00000002 dc57a000

kernel:[ 412.403743] bde0: c1004088 c05f4581 400c0013 db53b000 00000001 c07174db db53b22c ffffe000

kernel:[ 412.411959] be00: 00000000 dc569a00 c0147bb9 00000100 00000200 23d5fe02 00000000 db5cbc00

kernel:[ 412.420172] be20: c1004088 0001fffc db218c10 00000004 daf1e840 00000000 bf92f171 bf92f1a3

kernel:[ 412.428387] be40: dad87088 c013b5d3 20000000 00000000 c1004088 23d5fe02 dad86d00 c1004088

kernel:[ 412.436602] be60: c101dc80 00000000 00006d90 bf99b680 00000001 bf9860d3 00000000 00000000

kernel:[ 412.444816] be80: 00000000 dc57be84 dc57be84 23d5fe02 c1017a90 dad86d00 dad86d38 dad86ec8

kernel:[ 412.453031] bea0: 1c800000 c1004088 c0165a39 dc518e58 dad86d00 bf97b761 00000000 c0146683

kernel:[ 412.461245] bec0: dc57bed8 c0146683 00000002 bf99b680 02213071 00000001 db42b400 23d5fe02

kernel:[ 412.469459] bee0: c1017a90 c1004088 dad86ec8 c1004088 00000000 c013b6ad 400c0013 dc57bf08

kernel:[ 412.477673] bf00: c0165a39 23d5fe02 c094366b dad86d00 dad86d38 dad86ec8 1c800000 c1004088

kernel:[ 412.485888] bf20: c0165a39 dc518e58 dc11bdf8 bf97becf bf97bdfd dc5183c0 daa1d600 daa1d600

kernel:[ 412.494102] bf40: c0165911 c0165925 dc5183c0 dc57a000 daa1d600 c0165b17 00000000 00000000

kernel:[ 412.502318] bf60: c0165999 23d5fe02 c0165a39 dc518e40 00000000 dc518980 dc57a000 dc5183c0

kernel:[ 412.510533] bf80: c0165a39 c013fe1f dc57a000 dc518980 c013fd45 00000000 00000000 00000000

kernel:[ 412.518746] bfa0: 00000000 00000000 00000000 c0106a59 00000000 00000000 00000000 00000000

kernel:[ 412.526960] bfc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000

kernel:[ 412.535174] bfe0: 00000000 00000000 00000000 00000000 00000013 00000000 00000000 00000000

kernel:[ 412.697028] Code: 6d6a fa12 f383 d1ec (681b) 07da

Any idea why this would happen?

CM1 · March 26, 2019, 8:50pm

I have encountered some issues with analog input and libpruio. Last week, I had a working version of some code that is a modification of rb_file.

When I run it (or any of the examples) now, I get a Bus error. I am not sure what I did to cause this.

I changed some systemd services that I have calling my code, and I increased the amount of samples my rb_file code is recording.

It doesn’t seem like either of these should have affected libpruio.

debian@beaglebone:/boot$ sudo /opt/scripts/tools/version.sh
git:/opt/scripts/:[24f2e3495113a63c2dcc2c0bdc0d7660b54e4e8f]
eeprom:[A335BNLTBLA21722EL002623]
model:[TI_AM335x_BeagleBone_Blue]
dogtag:[BeagleBoard.org Debian Image 2018-10-07]
bootloader:[microSD-(push-button)]:[/dev/mmcblk0]:[U-Boot 2018.09-00002-g0b54a51eee]:[location: dd MBR]
bootloader:[eMMC-(default)]:[/dev/mmcblk1]:[U-Boot 2018.09-00002-g0b54a51eee]:[location: dd MBR]
kernel:[4.19.28-bone28]
nodejs:[v6.16.0]
uboot_overlay_options:[enable_uboot_overlays=1]
uboot_overlay_options:[uboot_overlay_addr4=/lib/firmware/BB-I2C1-00A0.dtbo]
uboot_overlay_options:[uboot_overlay_addr5=/lib/firmware/BB-W1-BLUE-00A0.dtbo]
uboot_overlay_options:[uboot_overlay_addr6=/lib/firmware/libpruio-0A00.dtbo]
uboot_overlay_options:[uboot_overlay_pru=/lib/firmware/AM335X-PRU-UIO-00A0.dtbo]
uboot_overlay_options:[enable_uboot_cape_universal=1]
pkg check: to individually upgrade run: [sudo apt install --only-upgrade ]
pkg:[bb-cape-overlays]:[4.4.20190123.0-0rcnee0~stretch+20190123]
pkg:[bb-wl18xx-firmware]:[1.20180517-0rcnee0~stretch+20180517]
pkg:[kmod]:[23-2rcnee1~stretch+20171005]
pkg:[librobotcontrol]:[1.0.4-git20190107.0-0rcnee0~stretch+20190108]
pkg:[firmware-ti-connectivity]:[20180825+dfsg-1rcnee1~stretch+20181217]
groups:[debian : debian adm kmem dialout cdrom floppy audio dip video plugdev users systemd-journal i2c bluetooth netdev cloud9ide gpio pwm eqep admin spi tisdk weston-launch xenomai]
cmdline:[console=ttyO0,115200n8 bone_capemgr.uboot_capemgr_enabled=1 root=/dev/mmcblk0p1 ro rootfstype=ext4 rootwait coherent_pool=1M net.ifnames=0 quiet]
dmesg | grep remote
[ 0.992529] remoteproc remoteproc0: wkup_m3 is available
[ 1.091092] remoteproc remoteproc0: powering up wkup_m3
[ 1.091113] remoteproc remoteproc0: Booting fw image am335x-pm-firmware.elf, size 217168
[ 1.095377] remoteproc remoteproc0: remote processor wkup_m3 is now up
dmesg | grep pru

DTJF · March 27, 2019, 5:07am

Hi CM!

CM1 · March 27, 2019, 2:25pm

Hello,

I have tried a few implementations, with the number of samples: 30,000 → 300,000 → 600,000. I had no issues with 30,000 or 300,000, as far as I can tell.

I haven’t made any changes to extram_pool_sz, and don’t actually know where the default for this is set. If you could please tell me where that is, I can answer that question.

I also wondered if I should change my uEnv.txt reference from AM335X-PRU-UIO-00A0.dtbo to AM335X-PRU-UIO-4-19-00A0.dtbo, but it was running on the 4.19 kernel previously without issue.

Thanks!

RobertCNelson · March 27, 2019, 2:31pm

don't use AM335X-PRU-UIO-4-19-00A0.dtbo, it was only used for testing briefly

https://github.com/beagleboard/bb.org-overlays/commit/35dddbc4bec1a61beddfd199601a784745124d3b

Regards,

DTJF · March 27, 2019, 2:52pm

Are you using error checks (as in the examples)? You didn’t change the extram_pool_sz, so you should get the message “out of memory” for sample amounts (= Adc->ChAz x Adc->Samp) greater than 128 kSamples (= default extram_pool_sz).

You can check the current size of the external memory block in variable PruIo->ESize. A guide for manipulation is in the docs at

http://users.freebasic-portal.de/tjf/Projekte/libpruio/doc/html/ChaMemory.html#SecERam

Regards

DTJF · March 27, 2019, 2:56pm

In version 0.6.4 some speed optimizations in the ARM code are done. This may lead to race conditions between ARM and PRU in sequences of fast ???->setValue() function calls. The bug is fixed in version 0.6.4b, which I uploaded today. Also the re-muxing in the DTOR is fine again.

Regards

CM1 · March 27, 2019, 3:56pm

Are you using error checks (as in the examples)? You didn’t change the extram_pool_sz, so you should get the message “out of memory” for sample amounts (= Adc->ChAz x Adc->Samp) greater than 128 kSamples (= default extram_pool_sz).

Yes, I am still using the error checks per the examples. When I run as root I get a bus error, otherwise the error message is:

constructor failed (cannot open /dev/uio5)

Maybe the number of samples is a secondary issue, because I cannot even run any of the examples, e.g. 1 returns a segmentation fault.

DTJF · March 27, 2019, 5:58pm

This is a LINUX issue. The kernel driver uio_pruss doesn’t load properly. Perhaps it gets listed by lsmod, but it didn’t create the interrupt devices /dev/uio[0-7]. (From user space there is no file /dev/uio5, so opening fails. As root it creates such a device, but without any function. → Don’t use root privileges for development!)

Follow Roberts instruction in this post, in order to provide the necessary debugging informations.

Regards

CM1 · March 27, 2019, 7:44pm

Follow Roberts instruction in this post, in order to provide the necessary debugging informations.

I have read enough of this board to know that question was coming, so I posted the output of /opt/scripts/tools/version.sh in my initial post. Here it is in case it wasn’t clear:

debian@beaglebone:/boot$ sudo /opt/scripts/tools/version.sh
git:/opt/scripts/:[24f2e3495113a63c2dcc2c0bdc0d7660b54e4e8f]
eeprom:[A335BNLTBLA21722EL002623]
model:[TI_AM335x_BeagleBone_Blue]

dogtag:[BeagleBoard.org Debian Image 2018-10-07]
bootloader:[microSD-(push-button)]:[/dev/mmcblk0]:[U-Boot 2018.09-00002-g0b54a51eee]:[location: dd MBR]

bootloader:[eMMC-(default)]:[/dev/mmcblk1]:[U-Boot 2018.09-00002-g0b54a51eee]:[location: dd MBR]
kernel:[4.19.28-bone28]
nodejs:[v6.16.0]
uboot_overlay_options:[enable_uboot_overlays=1]
uboot_overlay_options:[uboot_overlay_addr4=/lib/firmware/BB-I2C1-00A0.dtbo]
uboot_overlay_options:[uboot_overlay_addr5=/lib/firmware/BB-W1-BLUE-00A0.dtbo]
uboot_overlay_options:[uboot_overlay_addr6=/lib/firmware/libpruio-0A00.dtbo]

uboot_overlay_options:[uboot_overlay_pru=/lib/firmware/AM335X-PRU-UIO-00A0.dtbo]
uboot_overlay_options:[enable_uboot_cape_universal=1]
pkg check: to individually upgrade run: [sudo apt install --only-upgrade ]

pkg:[bb-cape-overlays]:[4.4.20190123.0-0rcnee0~stretch+20190123]

pkg:[bb-wl18xx-firmware]:[1.20180517-0rcnee0~stretch+20180517]
pkg:[kmod]:[23-2rcnee1~stretch+20171005]

pkg:[librobotcontrol]:[1.0.4-git20190107.0-0rcnee0~stretch+20190108]
pkg:[firmware-ti-connectivity]:[20180825+dfsg-1rcnee1~stretch+20181217]
groups:[debian : debian adm kmem dialout cdrom floppy audio dip video plugdev users systemd-journal i2c bluetooth netdev cloud9ide gpio pwm eqep admin spi tisdk weston-launch xenomai]
cmdline:[console=ttyO0,115200n8 bone_capemgr.uboot_capemgr_enabled=1 root=/dev/mmcblk0p1 ro rootfstype=ext4 rootwait coherent_pool=1M net.ifnames=0 quiet]
dmesg | grep remote
[ 0.992529] remoteproc remoteproc0: wkup_m3 is available
[ 1.091092] remoteproc remoteproc0: powering up wkup_m3
[ 1.091113] remoteproc remoteproc0: Booting fw image am335x-pm-firmware.elf, size 217168
[ 1.095377] remoteproc remoteproc0: remote processor wkup_m3 is now up
dmesg | grep pru

W1-BLUE is a custom dtbo for 1 wire.
libpruio dtbo is probably not actually referencing anything (I think I put it there as a misstep toward troubleshooting this issue)

Another thought I had is that perhaps there is a conflict with librobotcontrol using the PRU when librpuio wants to.

Thank you for your help!

DTJF · March 27, 2019, 10:07pm

I have read enough of this board to know that question was coming, so I posted the output of /opt/scripts/tools/version.sh in my initial post. Here it is in case it wasn’t clear:

Hopefully Robert finds any hint here …

W1-BLUE is a custom dtbo for 1 wire.
libpruio dtbo is probably not actually referencing anything (I think I put it there as a misstep toward troubleshooting this issue)

Another thought I had is that perhaps there is a conflict with librobotcontrol using the PRU when librpuio wants to.

I don’t know librobotcontrol, but in a first glance it looks like it’s using the rproc driver. It’s impossible to use both, rproc and uio_pruss driver at the same time. libpruio seems to work well on your system, but it’s the LINUX driver loading that fails. Sorry, I cannot help. For libpruio you’ll need a proper uio_pruss driver loading (rproc isn’t fast enough). Check the output of ls -l /dev/uio* , that should list eight entries in case of success. Ask in a librobotcontrol forum if it’s possible to run that library with uio_pruss driver loaded.

Regards

CM1 · March 28, 2019, 2:03pm

Thank you for the help regarding the driver conflict. I feared that might be the case. I will have to look into running librobotcontrol with uio_pruss, as you suggested.

I also hope that Robert has some insight into my problem.

Also, thank you for your efforts on libpruio. The times I am able to use it, it works well!

RobertCNelson · March 28, 2019, 2:06pm

librobotcontrol "pru" function calls are tied to remoteproc_pruss (v4.14.x-ti)

What features of librobotcontrol are you using?

Regards,

CM1 · March 28, 2019, 3:14pm

Hello Robert,

I am using a colleague’s code to read the BeagleBone Blue’s IMU/barometer. He is using some of librobotcontrol in that. To be honest, I don’t know specifically what of librobotcontrol his code is using.

To my understanding, it includes

rc/math/kalman.h, filter.h, and quaternion.h
rc/time.h, bmp.h, and mpu.h

I am reading through the librobotcontrol source, but it doesn’t look like any of these use the PRU.

I appreciate your insight into this.

RobertCNelson · March 28, 2019, 3:17pm

You should be fine then, just disable remoteproc_pruss and use
uio_pruss and ignore any of the librobotcontrol pru warnings..

Regards,

CM1 · March 28, 2019, 3:58pm

You should be fine then, just disable remoteproc_pruss and use
uio_pruss and ignore any of the librobotcontrol pru warnings…

Forgive me if I’m oblivious, what are the steps I take to disable remoteproc_pruss and use uio_pruss?

Do you mean to comment out the PRU-RPROC dtbo files in uEnv.txt, blacklist them in /etc/modprobe.d, or something else entirely?

lsmod shows uio, uio_pdrv_genirq, and uio_pruss already loaded.

When I check the PRU availability, i.e. ls /dev/uio8, it shows uio0-7. All are yellow but uio5.

Thank you,
CM

RobertCNelson · March 28, 2019, 4:01pm

in /boot/uEnv.txt

only have this one uboot_overlay_pru define enabled:

uboot_overlay_pru=/lib/firmware/AM335X-PRU-UIO-00A0.dtbo

Regards,

CM1 · March 28, 2019, 5:47pm

Ok, I already have just AM335X-PRU-UIO-00A0.dtbo enabled.

So, if it’s safe to say the issue is not with librobotcontrol, that brings me back to the initial problem.

I cannot run rb_file (or the code I wrote as a modification of it) - getting “constructor failed (cannot open /dev/uio5)”
When I run the example 1 - I get a seg fault.

Any thoughts?

DTJF · March 29, 2019, 1:07am

Please post the full output of ls -l /dev/uio*

Do you (your user ID) have write access to that files?

Regards

CM1 · March 29, 2019, 1:06pm

Please post the full output of ls -l /dev/uio*

Do you (your user ID) have write access to that files?

debian@beaglebone:~/analog$ ls -l /dev/uio*

crw-rw---- 1 root users 240, 0 Nov 3 2016 /dev/uio0

crw-rw---- 1 root users 240, 1 Nov 3 2016 /dev/uio1
crw-rw---- 1 root users 240, 2 Nov 3 2016 /dev/uio2
crw-rw---- 1 root users 240, 3 Nov 3 2016 /dev/uio3
crw-rw---- 1 root users 240, 4 Nov 3 2016 /dev/uio4
-rw-r–r-- 1 root root 0 Mar 29 08:30 /dev/uio5
crw-rw---- 1 root users 240, 6 Nov 3 2016 /dev/uio6
crw-rw---- 1 root users 240, 7 Nov 3 2016 /dev/uio7

When I add write access to /dev/uio5, and run my code (or example 1), I get a Bus error.

/dev/uio5 is listed in white, which means the system doesn’t recognize it as a device, correct?