HDMI Modes Problem Firmware 1.0 vs 2.0
Open, HighPublic


Atomos Ninja works with the Beta signal in Firmware version 1.0 but not in 2.0.

The problem seems to be in the "pic_jtag_pcie.py" script. When running the "axiom-start.sh" it gives the following error:


+ ./pic_jtag_pcie.py 0x92 0x92
Traceback (most recent call last):
  File "./pic_jtag_pcie.py", line 28, in <module>
    jtag_tms(i2c, base, "01100")      # goto Shift-IR
  File "/opt/axiom-firmware/software/scripts/jtag.py", line 108, in jtag_tms
    jtag_seq(i2c, base, [0x02, 0x02], data, mod, val)
  File "/opt/axiom-firmware/software/scripts/jtag.py", line 70, in jtag_seq
    i2c.write_i2c_block_data(last + 1, mod, [val])
OSError: [Errno 6] No such device or address

If I get the "pic_jtag_pcie.py" and "jtag.py" scripts from the 1.0 firmware, the HDMI works correctly.

I attach the diff from 1.0 vs 2.0 firmware.

sebastian triaged this task as High priority.
sebastian renamed this task from HDMI Modes 1.0 vs 2.0 Firmware to HDMI Modes Problem Firmware 1.0 vs 2.0.
sebastian updated the task description. (Show Details)
sebastian updated the task description. (Show Details)May 24 2020, 9:55 AM
sebastian updated the task description. (Show Details)May 24 2020, 2:20 PM
sebastian added a comment.EditedMay 25 2020, 6:54 PM

Sergio tested the new Firmware 2.0 with this file
modified to
base = 0x10
and reports the HDMI works OK!

This is because the PIC firmware on the Power Board changed (addresses changed) and Sergios PICs still have the old firmware.

Actually it is the I2C addresses the PICs on the Mainboard
respond to, which originally was 0x10-0x1F and 0x30-0x3F but
after we got a mux which responds at address 0x30 (PCA9543)
we had to change that to 0x40-0x4F and 0x60-0x6F.

This change can also be implemented on the Beta, but it
requires to flash new firmware into both routing fabric PICs.

Why these symptoms also appear on Bertrands Beta still needs to be verified but the problem is likely a different one.
Bertl suspects: could be very well related to the smbus_cffi changes

Issue has been identified and a fix is available, now the next task is how to detect and fix the behaviour with the right hardware version in the upstream version: https://github.com/apertus-open-source-cinema/axiom-firmware/issues/179