#include <hallo.h> Herbert Xu wrote on Sat Mar 23, 2002 um 01:21:21PM:
> > though to have seen it work), and I got the old failure. So Martin has > > the same problem as me, you may remember the small flamewar some months > > ago. Long description: > > What happens with 2.4.18-bf2.4 where you've got USB completely modularised? Same story as with your kernels. The box bootes, the keyboard stops working. If modules are loaded, it is useable again. > > - if there are no USB drivers in the kernel, the keyboard may work, or > > even NOT. > > - If it does work, I noted that as soon as any keyboard controller > > driver is loaded, the keyboard locks up immeditely. > > That's understandable. > > We've got two cases: > > 1. BIOS is in legacy mode. It is all the time. Without it, you would not be able to go trough LILO menu, syslinux prompt etc. etc. The keyboard works fine in DOS. > You should not load the usbkbd module. Well, I do not. After the first steps of hardware detection, I can not do anything. No matter with which kernel, even kernel-image-2.4.18-k7 has the same symptoms. > 2. BIOS is not in legacy mode. > > The usbkdb and associated modules must be loaded for keyboard to function. Same story here - no keyboard, no control, no chance to load them. > >From what you're saying, it seems that 1 fails on some recent machines. > However, have you seen any machine where 2 does not work? I did not try to disable the legacy emulation, but I expect the keyboard to work there as soon as the modules are loaded. > If 2 always works, but fails when the BIOS is in legacy mode, then the > solution is to get the boot floppies to differentiate the two modes and > operate differently. This can be done either through a argument passed > via LILO, or perhaps a timeout event (i.e., press ENTER to continue, > if that doesn't work, the usb modules can be loaded after a certain > period of time). This is possible, in fact I will do this for boot floppies now. Unfortunately, this will not help to load the root disk. (*) > And I'm yet to hear any reports where 2 does not work. As far as I can say, the only method to make it work is to load the module. > > - BTW, note that is it not possible to use the initrd filesystem when > > the keyboard is not functional, and not install Debian, just not do > > anything. > > If you're talking about the initrd used for my 2.4 kernels, that's a > different problem. The shell in the initrd is meant to be used for > debugging only since if you need it to boot, there must be a bug in I do not explicitely need it, but I wish to have a working keyboard without loading the modules first, even for init=/bin/sh tricks. > mkinitrd. And if you're debugging things, it's trivial to load any > modules you need for the console to operate. > > > - Everything works fine if the drivers are loaded as modules, usbkbd as > > well as keybdev. > > Does this work even in legacy mode? If it does, then the problem is solved Yes, modules seem not to care about previous keyboard state. > and I don't understand what the bug report is about. Okay, detach your keyboard and type "modprobe usbkbd". Have fun. (*) Could you find the patch for syslinux again? There is still the problem with loading the rootdisk when using the floppy. Gruss/Regards, Eduard. -- Wenn einer tr�umt, bleibt es ein Traum. Wenn viele tr�umen, beginnt der Traum, Wirklichkeit zu werden. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
