Re: Random keystrokes lost
On Wed, 27 Nov 2002, martin f krafft wrote:
> also sprach Karl E. Jorgensen <karl@jorgensen.com> [2002.11.27.0147 +0100]:
> > Bad luck for you :-) For me the system would freeze less than .5
> > seconds every 3 or 4 seconds. Just enough to make me wonder whether I
> > was given de-caf by mistake...
>
> The phenomenon of the mouse freezing or keyboard input being delayed
> does coincide with IDE/hdd activity.
>
> But the problem of forgotten keystrokes (and randomly repeating
> letters[1]) does not need to happen with a hdd access.
>
> 1. i *touch* 'i' and suddenly, iiiii appears, or similar. Happens
> for all keys rarely but noticably, and probably has something to
> do with the random keystrokes being lost.
Damn. I don't think I will ever touch Dell laptops ever again. My own
I4000 has had problems, such as 2 wonky HD's, but they just seem so damn
unreliable, and have a lot of bios bugs.
My own thought about the KB is - with mine, If I use the tleds package to
blink the keyboard leds AND have an externel keyboard plugged in, then
whenever one of the leds change state, then the KB won't register a
keypress within x milliseconds of the led changing state. This manifests
itself as a loss of every 20th or so keypress. It seems the Dell KB
controllers just can't think of 2 things at once - I have never had this
problem on any other computer.
Also - your hdd statement: try unmasking interrupts using hdparm. This
helps modem traffic a lot, and might also help the KB - if there is a lot
of hdd activity (and laptop disks are very slow), then the KB just never
gets a chance to signal to the CPU that "HEY! LOOK HERE. I HAVE A
KEYSTROKE FOR YOU. FEED ME!". :)
As for the temperature thing, can you cat /proc/apm and the i8k file under
/proc (I can't remember what is is called - perhaps you might have to do a
strace on i8kutils when it reads the temperature from the kernel, if it is
not obviously under /proc) when there is a problem, and then again when
there is not a problem?
Is this flavour laptop aimed more at APM or ACPI for power management? The
newer ones tend to work better on ACPI than APM.
--
TimC -- http://astronomy.swin.edu.au/staff/tconnors/
Bus error -- driver executed.
Reply to: