Arch Linux

Well, I don't see the error in that message. It's a diagnostic message printed by the i8042 driver. Mine says this:
[ 0.990364] i8042: PNP: PS/2 Controller [PNP0303:PS2K,PNP0f13:PS2M] at 0x60,0x64 irq 1,12

i'm sorry i have to dig deeper here. my suggestion that the log level was wrong was not good.

however, i wonder if i8042 should not be compiled as a module (CONFIG_SERIO_I8042=m), as well as libps2 and serio

I think it is in the kernel due to have the possibility to have an keyboard on initrd.

No, i8042 is too common and important to make this change now. Sorry, but you'll have to live with that message.

Any chance of reassessing this bug now that we have the new 'keyboard' hook in mkinitcpio? The new hook treats all keyboards the same, and will include whatever module is necessary in the initamfs.

I think PS/2 will be getting more and more obsolete, so I think it makes sense to not give it special treatment (if it really is necessary to have the keyboard driver compiled in, then we probably want to start compiling in more modules).

A post-upgrade message from the kernel package might be necessary to inform people that they need to add the 'keyboard' hook if they want a keyboard in their initrd, but apart from that building i8042 as a module should not have any adverse effects (and udev will anyway load the module eventually even if it is not included in the initramfs).

Some data: https://www.archlinux.de/?page=ModuleStatistics

I don't know how many people would still use i8042 (as the data is only available for things built as modules), but it appears usbhid is currently used by more than two thirds of our users, so at least in principle they might not need i8042. In practice I suppose most people actually have both, at least for now.

PS/2 keyboard with USB mouse still loads usbhid, so that's not really an indication of "no ps/2 keyboard".

Yeah, I have no way of knowing how large percentage actually don't use PS/2 at all (only way to find out is to make it a module ;-) ).

Tom, did you even try this? I can see no way to convince Kbuild to compile this as a module - it is selected by several options which you cannot even change on x86.

Closing this one it is not possible on X86 to have it as module.

Reopening to answer the question:

Yes I tried this (of course). I have been running with this as a module since I first brought it up. I remember it being a bit tricky to convince Kconfig, but now I cannot reproduce the problem here. I have:

Symbol: SERIO_I8042 [=m]

Selected by: KEYBOARD_ATKBD [=m] && !UML && INPUT [=y] && INPUT_KEYBOARD [=y] && X86 [=y] ||
MOUSE_PS2 [=m] && !UML && INPUT [=y] && INPUT_MOUSE [=y] && X86 [=y]

Okay, I found it now. Configuring the KEYBOARD_ATKBD option requires CONFIG_EXPERT=y, let me paste the help for that option:

menuconfig EXPERT
bool "Configure standard kernel features (expert users)"
# Unhide debug options, to make the on-by-default options visible
select DEBUG_KERNEL
help
This option allows certain base kernel options and settings
to be disabled or tweaked. This is for specialized
environments which can tolerate a "non-standard" kernel.
Only use this if you really know what you are doing.

We are not in a "specialized environment" and I would really advise against enabling this option.

Seems to me this should be changed upstream. the CONFIG_EXPERT=y requirement probably goes back to the time when there existed no USB keyboards and no one bothered changing that since then.

I'll submit a patch upstream once I get my git send-email to work again. I'll ping this report again when/if it gets accepted.

What is up? I am still waiting for it to be fixed ...

@djszapi: I encourage you to scratch your own itch if this is really bothering you to the point of whinging here. It really should be a trivial patch, as it doesn't even require writing any C.

Any update on the situation?

This will be fixed upstream for 3.13. I submitted one fix [0] and two trivial buildsys changes [1,2] that have been applied to the input tree for 3.13, but have not yet reached Linus.

[0]: http://www.spinics.net/lists/linux-input/msg27317.html
[1]: http://www.spinics.net/lists/linux-input/msg27298.html
[2]: http://www.spinics.net/lists/linux-input/msg27299.html

For those of you waiting with bated breath for this to move forward: my patches have now reached Linus' tree, so we'll get rid of this warning by 3.13... yay \o/

Any reason this issue is still open? Sounds like it's fixed already