on screen debug: many keystrokes needed to get to next page of messages

[Adek336]

When on screen debug is selected at boot time, I have to press a key to see more and more pages of messages. However about the time AHCI messages appear, suddenly I need a lot of keystrokes to get the next pages of messages. It doesn't seem like this is just a matter of the kernel not issueing any messages for a significant time: the "Press any key to continue" text doesn't disappear, and when it finally does print the next page of messages they all come very quickly.

[PieterPanman]

See also #3129, I have the same thing. Is anyone NOT having this? Maybe it is due to some hardware component?

[Adek336]

proof of concept

[Adek336]

The interrupt handler debug_keyboard_interrupt() steals the keystroke from arch_debug_blue_screen_getchar() in x86 (perhaps this also affects other architectures).

To fix the problem, arch_debug_blue_screen_getchar() has to force the isr to not read from ps2 port, temporarily uninstall the isr, clear interrupts or mask the keyboard irq. kbd.1.diff is an example fix.

Also note that F12 / BREAK key isn't programmed to open the KDL, when pressed against a "Press key to continue" message.

[PieterPanman]

I can confirm your patch works on my laptop, it now responds perfectly. Very nice.

[Adek336]

Does the shift key work in the debugger for you with the patch applied?

[PieterPanman]

Nope it does not. I also cannot enter KDL using F12 while it is paging the debugging output :) It just goes to the next page...

[scottmc]

any progress on this one, it is still an issue? Perhaps you can verify that the patch still applies cleanly on a recent build of Haiku and double check for proper Haiku coding style. I believe the returns get their own line... there's probably a few other issues in there as well.

[pulkomandy]

Can anyone still reproduce this? (note: you need a PS/2 keyboard - USB won't have the same problem)

[waddlesplash]

No reply in 13 months, closing