Testing needed for potential Beagle image release to use as starting point for GSoC

Google Summer of Code
1

Mentors, students,

If you can provide some testing and feedback on the images at Index of /rootfs/bb.org/testing/2021-03-29, it would be really great. If we can close any significant regressions, we and make it an official release as a starting point for this year’s GSoC students.

2

So far I’ve built/tested on black using u-boot/kernels and rootfs from Robert’s wiki page and encountered a few issues.

  1. focal rootfs has network issues and fails on network up/ttyG, then freezes without a console prompt (kernel 5.4 or less) whereas bionic boots successfully.
  2. with latest debian flashed to emmc, using uEnv.txt on both emmc and sdcard has some issues with newer kernels; 5.10 kernel has mmc0/1 swapped from both earlier kernels and u-boot.
  3. beaglebone-yocto (dunfell branch) boots fine, uses vfat /boot and extlinux.conf
  4. using extlinux.conf on sdcard boots reliably (with debian/uEnv on emmc) using bionic and 5.10 kernel

Hopefully building from meta-ti should also work, but I didn’t find an image recipe or doc to get a nice TI dev image with PRU stuff, etc.

Right now I don’t have an answer for munging uEnv.txt on sdcard to make it happy with Robert’s default setup.

3

This is filtering down from mainline:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/arch/arm/boot/dts/am33xx.dtsi?id=9bbce32a20d6a72c767a7f85fd6127babd1410ac

		mmc0 = &mmc1;
		mmc1 = &mmc2;
		mmc2 = &mmc3;

Some of us might remember that is issues in 3.8.x. :wink: Well that same fix is now finally mainline…

Regards,

4

Once: [PATCH v3 00/10] Add support for extension boards detection and DT overlays application hit’s mainline, we should be able to convert to extlinux while keeping overlay support…

Regards,

5

I’m not sure I remember anything much from 3.8 … I can see the alias patch on the ti-linux-rt-5.10.y branch, but it does not fix the swapped root device issue if I use the “default” uEnv.txt with just uname_r=n.n.n in it. What exactly is the “simple” uEnv fix to get the right root mounted? With the above it loads kernel from sdcard but then mounts the emmc rootfs…

6

This is fixed… it was broken in 5.8/5.9/5.10, but it’s been fixed…

http://gfnd.rcn-ee.org:81/farm/uptime/pwr03-ser11-bbb-5.4.106-ti-r26.log
http://gfnd.rcn-ee.org:81/farm/uptime/pwr05-ser13-bbb-5.10.21-ti-r1.log

lsblk - 5.10.21-ti-r1.log
NAME         MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
mmcblk1      179:0    0  1.8G  0 disk 
mmcblk1boot0 179:256  0    1M  1 disk 
mmcblk1boot1 179:512  0    1M  1 disk 
mmcblk0      179:768  0 29.8G  0 disk 
└─mmcblk0p1  179:769  0 29.8G  0 part /

lsblk - 5.4.106-ti-r26.log
NAME         MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
mmcblk0      179:0    0 29.8G  0 disk 
└─mmcblk0p1  179:1    0 29.8G  0 part /
mmcblk1      179:256  0  1.8G  0 disk 
mmcblk1boot0 179:512  0    1M  1 disk 
mmcblk1boot1 179:768  0    1M  1 disk

Regards,

7

I literally just built it (again) last night and got 5.10.21-ti-r0 … Lemme update again.

8

Sorry, i “literally” fixed it last night, when i tagged ‘r1’ and added the “rt” version :wink:

Regards,

9

Ps, here’s a simple /boot/uEnv.txt:

uname_r=5.10.21-ti-r1
cmdline=coherent_pool=1M net.ifnames=0 lpj=1990656 rng_core.default_quality=100
#enable_uboot_overlays=1

Regards,

10

Okay, confirmed the root device fix on 5.10.21-ti-rt-r1 but ubuntu still has some issues:

  1. serial gettys still timeout/fail and console freezes with no login prompt:

    [DEPEND] Dependency failed for Serial Getty on ttyGS0.
    [ OK ] Started Generic Board Startup.
    [ OK ] Reached target Multi-User System.
    [ OK ] Reached target Graphical Interface.
    Starting Update UTMP about System Runlevel Changes…
    [ OK ] Started Update UTMP about System Runlevel Changes.

  2. still getting phy error on ethernet startup:

    [ 20.664757] cpsw 4a100000.ethernet: phy “/ocp/interconnect@4a000000/segment@0/target-module@100000/ethernet@0/mdio@1000/ethernet-phy@0” not found on slave 0`

11

Does this same issue occur with Debian?

Throw that board away, or mark it as having a bad PHY… Our phy port hack probably needs a tweak for later kernels… For bleeding edge, which is our 5.10-ti branch you’ll just be pulling hairs…

Regards,

12

Tested so far:

  • yocto beaglebone dunfell branch - PASS

Manual u-boot/kernel/sdcard build using https://www.digikey.com/eewiki/display/linuxonarm/BeagleBone+Black and kernel 5.10.21-ti-rt-r1:

  • debian - PASS
  • ubuntu - network issues

Latest test image:

  • bone-eMMC-flasher-debian-10.9-iot-armhf-2021-03-29-4gb.img.xz - PASS

Ubuntu issues (mainly network/tty-related) vary with hardware and kernel version; note anything before the above version and later than 5.4.x will have mmc device issues. Occasional PHY errors (also depends on kernel and/or hardware).

13

Tested on the following hardware:

  • black 4 GB emmc
  • black 2 GB emmc
  • green wireless
1 Like