XO-1.75 Open Firmware reports continuous Keyboard interrupt if a GPS with PL2303 is attached

[Quozl]

A regression over XO-1.5 and XO-1. First mentioned in #11871.

[Quozl]

Summary: data sent by the GPS, such as the end of line marker (0x0d or 0x0a), if sent at a low baud rate that is not 115200 baud, will be received as a NUL (0x00) or BREAK, and generate the Keyboard interrupt.

Tested with an XO-1.75 Q4D14ja, a USB serial adapter based on PL2303, a serial DE9 female to DE9 female crossover "null modem" cable, to another computer.

With the other computer configured for 115200 baud, using screen /dev/ttyS0 115200, the screen instance becomes a keyboard for the XO-1.75, and Ctrl/Space is treated by Open Firmware as a Keyboard interrupt.

With the other computer reconfigured for 1200 baud, pressing enter is treated as a Keyboard interrupt. Open Firmware initialises the USB serial adapter to 115200 baud. This is hard coded, but can be changed before inserting the adapter, like this:

\ boot
\ insert adapter
probe-usb
dev /usb/serial
patch 2580 1c200 pl2303-set-baud
?usb-keyboard

I suggest that in an application where a USB serial adapter is permanently connected with a device that is not generating data at 115200 baud, that a custom build of Open Firmware be prepared with a suitable change to pl2303-set-baud. The olpc-firmware git repository has the procedures for building.

We might also restore the behaviour to XO-1 and XO-1.5, as mentioned in #11871.

[Quozl]

Mitch has agreed we can remove the USB serial keyboard support from all models.

[Quozl]

Fixed in svn 2985.

[Quozl]

Released in Q4D15 and is in os12.

[Quozl]

pgf: wanna test? or just close, thanks.

[pgf]

sorry for the delay -- lost my gps dongle for a week. :-) verified fixed, closing.