Re: Repeated or missed keys after upgrading from 6.2 to 7.0

From: Mike Silbersack <silby_at_silby.com>
Date: Tue, 27 May 2008 00:17:48 -0500 (CDT)
On Mon, 24 Dec 2007, Mike Silbersack wrote:

> In order to eat my own dog food, I upgraded my laptop from 6.2 to 7.0. This 
> seemed to have gone well, until I started writing a long e-mail while sitting 
> on the couch today.  As I was typing the e-mail, I noticed that my typing 
> skills seemed to have gone missing; there were words missing 2-3 letters, and 
> other places where I was apparently holding down keyyyys. Heh, that's a real 
> example of the phenomenon right there.
>
> After a while I realized that I was not typing sloppily, but that in fact 
> keys are being lost in certain cases and duplicated in others.  Since I did 
> not rebuild any ports or packages, I'm convinced that this is directly 
> related to the 7.0 upgrade.
>
> This behavior has shown up when running a local copy of pine (inside 
> konsole), chatting in ksirc, and in a few other programs.  (I'm running KDE.) 
> I think it happens more when on battery than when plugged into an outlet. 
> I'm running xbattbar, so it could be querying the battery status and causing 
> problems.  This is using the laptop's built-in keyboard (non-USB.)
>
> I'm going to try to track this down, although I don't know how successful 
> I'll be.  I'd like to know if anyone else has seen this problem and if they 
> have any additional information that might help me track it down faster.
>
> Thanks,
>
> Mike "Silby" Silbersack

For anyone still interested (and I suspect those who have Acer laptops 
will be), I've finally found a fix for this problem.

The same problem was reported with some versions of Linux:

http://bugzilla.kernel.org/show_bug.cgi?id=9998

To summarize the bug report, the problem is that the ACPI Embedded 
Controller on some Acer laptops handles both keyboard I/O and the 
communication channel to the smart battery.  If you talk to the battery 
too quickly, the chip will start dropping keystrokes.  To deal with this, 
the Linux acpi maintainers added back some delays that had been present in 
the past.

I tried a similar approach, and found it to be effective on FreeBSD 7.0. 
Applying the attached patch and setting debug.acpi.ec.extradelay=1000 
seems to completely cure the keyboard problems for me.

Anyone interested in reviewing the patch?

Thanks,

Mike "Silby" Silbersack
Received on Tue May 27 2008 - 03:44:31 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:31 UTC