I recently grovelled through the ps2 mouse code (my new mouse wasn't
working - my fault).  There doesn't seem to be anything in netBSD to
support 'hot plugging' of keyboards and mice.

I've written such code (for a different OS), it isn't exactly hard!
Indeed you can tell whether you have a mouse or keyboard attached.
(running with 2 ps2 mice, and a touchscreen, and a serial mouse...)

Unfortunately the way netbsd runs its hardware drivers seems to rely on
an initial grope....  (Which looks like it is broken in two separate
places...)

    David

 
----- Original Message ----- 
From: Matthew Fincham <matthewf@cat.co.za>
To: NetBSD-Users <netbsd-users@netbsd.org>; <tech-kern@netbsd.org>
Sent: Tuesday, December 04, 2001 7:36 AM
Subject: Keyboard difficulties


> Hello
> 
> I have already posted this message on netbsd-help, but haven't had any
> feedback, so I am trying again, on tech-kern, and netbsd-users. :-)
> 
> I am having difficulty getting the keyboard to work when the system boots up
> without a keyboard, and then has the keyboard plugged in after the boot-up
> sequence is complete.
> 
> The actions I take are these:
> 
> 1. unplug the keyboard, and reboot the system. Leave the keyboard unplugged.
> 2. once the system has booted up, plug the keyboard in and try to login.
> 
> If the keyboard is not working, then repeat step 1. What I find is that the
> keyboard works on alternate reboots (works, doesn't work, works, ....).
> 
> I have added debug info to the kernel, and have included the relevant (??)
> parts in the e-mail. Essentailly, there are three cases to look at:
> a) boot with keyboard plugged in
> b) boot without keyboard plugged in, and when plugged in, keyboard works
> c) boot without keyboard plugged in, and when plugged in, keyboard does not
> work
> 
> a) Keyboard Plugged In
> =================
> pckbc0 at isa0 port 0x60-0x64
> pckbdprobe: reset succeeded!
> pckbdprobe:resp[0]=0xaa -> continuing
> pckbdprobe: flushing kbd buffer
> pckbd: the 8042 is translating -> use AT codes
> pckbdprobe: succeeded setting xtscancode
> pckbd0 at pckbc0 (kbd slot)
> pckbc0: using irq 1 for kbd slot
> 
> wskbd_match
> Console unspecified
> wskbd0 at pckbd0
> wskbd_attach: console keyboard
> wskbd_add_mux: 0 dmux0 0xc0866e00
> wskbd_set_display()
> Ebusy 1, using wsdisplay0
> pmsi0 at pckbc0 (aux slot)
> pckbc0: using irq 12 for aux slot
> wsmouse0 at pmsi0wsmux_attach: n=0
> mux 0
> 
> b) Boot without keyboard plugged in, and when plugged in, keyboard works
> ======================================================
> pckbc0 at isa0 port 0x60-0x64
> pckbdprobe: reset error 5
> pckbdprobe: returning 1
> pckbd0 at pckbc0 (kbd slot)
> pckbc0: using irq 1 for kbd slot
> 
> wskbd_match
> Console unspecified
> wskbd0 at pckbd0
> wskbd_attach: console keyboard
> wskbd_add_mux: 0 dmux0 0xc0866e00
> wskbd_set_display()
> Ebusy 1, using wsdisplay0
> pmsi0 at pckbc0 (aux slot)
> pckbc0: using irq 12 for aux slot
> wsmouse0 at pmsi0wsmux_attach: n=0
> mux 0
> 
> c) Boot without keyboard plugged in, and when plugged in, keyboard does not
> work
> =========================================================
> pckbc0 at isa0 port 0x60-0x64
> pckbdprobe: reset succeeded!
> pckbdprobe:resp[0]=0xaa -> continuing
> pckbdprobe: flushing kbd buffer
> pckbd: the 8042 is translating -> use AT codes
> pckbd: error setting scanset 2
> pckbd0 at pckbc0 (kbd slot)
> pckbc0: using irq 1 for kbd slot
> 
> wskbd_match
> Console unspecified
> wskbd0 at pckbd0
> wskbd_attach: console keyboard
> wskbd_add_mux: 0 dmux0 0xc0866e00
> wskbd_set_display()
> Ebusy 1, using wsdisplay0
> pmsprobe: reset error 5
> pmsiprobe: reset error 5
> 
> 
> The most obvious thing I can see is that in case (c), when the keyboard is
> not present and when plugged in fails, the reset command works! This seems
> strange, because the reset command polls for a response...but there is no
> keyboard. In case (b) the reset command fails, which makes a lot more sense.
> 
> The situation this needs to be solved for is where a number of systems use a
> single montior and keyboard with a switch-box. We need the systems to be
> able to boot without the switch-box switched to the booting machine. (the
> tests done here did not involve the switch-box, but just a single keyboard
> and system, so I don't think it is an issue with the switch-box)
> 
> I have tried this on NetBSD 1.4 and 1.5 with the same results. The
> motherboard being used is a CS62-TC DFI motherboard . On other motherboards
> (e.g. the GIGABYTE GA-6BX7) this has not been a problem.
> 
> Does anyone have any ideas why case (c) behaves as it does? Any help would
> be appreciated.
> 
> Thank you
> Matthew Fincham
> 
> 
> 
> 
> 
>