x15: weekly snapshot updates




opencl enabled out of box


better support for the netinstall


I’m assuming you know about this:

[ 12.720913] ------------[ cut here ]------------
[ 12.720935] WARNING: CPU: 0 PID: 4 at drivers/remoteproc/remoteproc_core.c:873 rproc_handle_intmem+0x3c/0x1c4()
[ 12.720941] RSC_INTMEM is deprecated. Please do not use this resource type to support loading into internal memories.
[ 12.720947] Modules linked in: rpmsg_rpc virtio_rpmsg_bus snd_soc_simple_card ti_vpe videobuf2_dma_contig ti_vpdma v4l2_mem2mem videobuf2_memops snd_soc_omap_hdmi_audio videobuf2_core v4l2_common videodev media snd_soc_davinci_mcasp pruss_remoteproc snd_soc_edma omap_rng rng_core snd_soc_tlv320aic3x omap_remoteproc debugss_kmodule(O) uio_pdrv_genirq uio sg
[ 12.721037] CPU: 0 PID: 4 Comm: kworker/0:0 Tainted: G O 4.1.10-ti-r24 #1
[ 12.721044] Hardware name: Generic DRA74X (Flattened Device Tree)
[ 12.721056] Workqueue: events request_firmware_work_func
[ 12.721076] [] (unwind_backtrace) from [] (show_stack+0x20/0x24)
[ 12.721090] [] (show_stack) from [] (dump_stack+0x8c/0xcc)
[ 12.721102] [] (dump_stack) from [] (warn_slowpath_common+0x98/0xc8)
[ 12.721113] [] (warn_slowpath_common) from [] (warn_slowpath_fmt+0x40/0x48)
[ 12.721124] [] (warn_slowpath_fmt) from [] (rproc_handle_intmem+0x3c/0x1c4)
[ 12.721136] [] (rproc_handle_intmem) from [] (rproc_handle_resources+0x78/0x140)
[ 12.721146] [] (rproc_handle_resources) from [] (rproc_boot+0x32c/0x51c)
[ 12.721157] [] (rproc_boot) from [] (rproc_virtio_find_vqs+0x1f8/0x250)
[ 12.721172] [] (rproc_virtio_find_vqs) from [] (rpmsg_probe+0xc4/0x3bc [virtio_rpmsg_bus])
[ 12.721195] [] (rpmsg_probe [virtio_rpmsg_bus]) from [] (virtio_dev_probe+0x20c/0x2f0)
[ 12.721207] [] (virtio_dev_probe) from [] (driver_probe_device+0x214/0x440)
[ 12.721218] [] (driver_probe_device) from [] (__device_attach+0x50/0x54)
[ 12.721228] [] (__device_attach) from [] (bus_for_each_drv+0x74/0xa8)
[ 12.721238] [] (bus_for_each_drv) from [] (device_attach+0x84/0x98)
[ 12.721248] [] (device_attach) from [] (bus_probe_device+0x94/0xb8)
[ 12.721257] [] (bus_probe_device) from [] (device_add+0x44c/0x5e4)
[ 12.721267] [] (device_add) from [] (device_register+0x24/0x28)
[ 12.721276] [] (device_register) from [] (register_virtio_device+0xb8/0xf8)
[ 12.721287] [] (register_virtio_device) from [] (rproc_add_virtio_dev+0x54/0xa8)
[ 12.721298] [] (rproc_add_virtio_dev) from [] (rproc_handle_vdev+0x1e8/0x298)
[ 12.721309] [] (rproc_handle_vdev) from [] (rproc_handle_resources+0x78/0x140)
[ 12.721319] [] (rproc_handle_resources) from [] (rproc_fw_config_virtio+0xf0/0xf8)
[ 12.721329] [] (rproc_fw_config_virtio) from [] (request_firmware_work_func+0x40/0x60)
[ 12.721341] [] (request_firmware_work_func) from [] (process_one_work+0x138/0x450)
[ 12.721351] [] (process_one_work) from [] (worker_thread+0x54/0x4bc)
[ 12.721363] [] (worker_thread) from [] (kthread+0xec/0x104)
[ 12.721375] [] (kthread) from [] (ret_from_fork+0x14/0x2c)
[ 12.721382] —[ end trace 47e62c10c0368342 ]—

This occurs because the DSP firmware authors need to move away from usage of RSC_INTMEM in the resource table. That approach has been rejected upstream, and I will be reworking by TI to not rely on this in the next month.


Thanks for the info, John.

Robert, do you want us to test the weekly updates or concentrate on the production image from before?

So, for the first x15 production run it sounds like "2015-10-25" (1)
has been choose by CircuitCo, the deadline was friday (2015-10-30)

I've incorporated all the feedback from the other threads, so 10-25
should be in pretty good shape..

Due to this part being a beast, i'm still planning weekly snapshot
updates of things we missed in 10-25:

x15: pru_rpmgs:

John just fixed this in the kernel, it'll be enabled by default with
the r25 kernel.

x15: video decode/encoder:
needs libdce, which needs ipcdev's headers *

I took the short cut with opencl, to use ipcdev's "AF_RPMSG=41" kernel
header flag, which allows us to NOT have to build ipcdev against every
kernel update...

However, it turns out libdce requires something not under the
"AF_RPMSG" flag, so i need to fix ipcdev

x15: sgx

linux module is installable via apt:
little bit of wayland/egl userspace hacking and it should be working..
(ti's sdk has matrix-gui running on wayland, sgx works)

1: https://rcn-ee.com/rootfs/bb.org/testing/2015-10-25/





remoteproc_pruss work from: John Syne

sudo cp /opt/scripts/device/x15/pru-rpmsg_client_sample/am57xx* /lib/firmware/
sudo reboot

sudo modprobe rpmsg_client_sample

Check journalctl for "hello world"


For shutdown, I've noticed

sudo systemctl poweroff

is just more reliable then "sudo halt" as halt doesn't seem to turn
everything off..


Hi John -

I've been testing out the new kernel and example PRU code you wrote. Is that code available anywhere to see how you did it? I want to use the PRU to control the different serial interfaces and McSPI but can't seem to figure it out on the x15.

Hi Yogi,

First, I didn’t write any of this code; all were written by TI. All I did was to pull pieces from the various TI repo’s and SDKs and then get them working on the x15 and BBB. When I did this on the bb-kernel repo, the changes were considerable, but when I did this on the ti-linux-kernel-dev repo, most of the patches had already been applied. For the source code, you will find it here: