On 17/6/20 8:34 am, Chris Johns wrote: > On 16/6/20 10:55 pm, Joel Sherrill wrote: >> On Tue, Jun 16, 2020 at 7:08 AM <jan.som...@dlr.de >> <mailto:jan.som...@dlr.de>> >> wrote: >> We had this problem trying to set up the embedded PC we use for testing in >> the >> new lab. >> When we tried the same hardware with FreeDOS, it worked. When we tried an >> old Dell >> desktop as the RTEMS PC, the serial port didn't work. > > I have a legacy small form factor ITX computer with a standard COM1 RS232 > only port.
I have debugged the problem on my hardware and the issue seems to be no serial port initialisation during start up. There was no call being made to the legacy ns16550 intialise call ns16550_init. I am wondering if the improvements to remove code that is not used and the removal of the console in tests means nothing is left to make the call and we have depended on it being there. I have a patch I am testing which I will post soon to add the initialisation to the BSP printk input and output functions. I am now wondering how many other BSPs could be suffering from the same problem? Any BSP where the bootloader or some initialisation process handles the printk set up will not see a problem. For example the xilinx zynq BSP will not see a problem because the systemZ ps7init set up intialises the UART. Chris _______________________________________________ devel mailing list devel@rtems.org http://lists.rtems.org/mailman/listinfo/devel