Skip to contentSkip to conversation

Synaptics losing sync

DISCUSSION Ruben Puettmann 2 participants 4 messages Nov 1, 2003 – Nov 3, 2003 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw

You wrote in linux.kernel:

Quoted 21 lines

I've sent this report before.

I repeat it just in case someone found a workaround, and because 2.6.0-test9 gives other related errors as well (TSC error):

...
Synaptics driver lost sync at 1st byte
Synaptics driver lost sync at 1st byte
Synaptics driver lost sync at 1st byte
Synaptics driver lost sync at 4th byte
Synaptics driver lost sync at 1st byte
Synaptics driver lost sync at 1st byte
Synaptics driver lost sync at 1st byte
Synaptics driver lost sync at 1st byte
Synaptics driver resynced.
Losing too many ticks!
TSC cannot be used as a timesource. (Are you running with SpeedStep?) Falling back to a sane timesource.
Synaptics driver lost sync at 1st byte
...

Quoted 3 lines

The laptop is a Dell X200 with APM and cpufreq enabled, and IO-apic disabled.

Here the same with IBM Thinkpad R40 2722GDG with APM cause ACPI is for this laptop totaly broken.

Ruben

Signature · 3 lines
Ruben Puettmann
[email protected]
http://www.puettmann.net
#2 lore raw

Hi!

Quoted 9 lines · Ruben Puettmann
Quoted 9 lines

...
Synaptics driver lost sync at 1st byte
Synaptics driver resynced.
Losing too many ticks!
TSC cannot be used as a timesource. (Are you running with SpeedStep?) Falling back to a sane timesource.
Synaptics driver lost sync at 1st byte
...

I can observe the same "Synaptics lost sync" messages on Acer TravelMate 242XC with ACPI enabled (without APM even compiled) on 2.6.0-test[0-9]. But I silently ignore these messages - touchpad works ok. I use an USB mouse, but messages appear even without it plugged.

"Losing too many ticks" exists when I'm running with cpufreq [p4_clockmod] and clock=tsc (default). This is because of rescaling TSC pitch by cpufreq, I think. If I specify in bootloader clock=hpet, problem disappears. [Am I doing right?]

I can provide more information if somebody is interested.

Cheers
Szymon

#3 lore raw

On Sun, Nov 02, 2003 at 10:48:33AM +0100, Szymon Aceda?ski wrote:

                hy,
Quoted 5 lines

"Losing too many ticks" exists when I'm running with cpufreq [p4_clockmod] and clock=tsc (default). This is because of rescaling TSC pitch by cpufreq, I think. If I specify in bootloader clock=hpet, problem disappears. [Am I doing right?]

If I boot with clock=hpet the cpu MHz in cat /proc/cpuinfo is 0.

Kernel 2.6.0-tes8-bk1
IBM Thinkpad with PentiumM

Ruben

Signature · 3 lines
Ruben Puettmann
[email protected]
http://www.puettmann.net
#4 lore raw
Quoted 10 lines · you

On Sun, Nov 02, 2003 at 10:48:33AM +0100, Szymon Aceda?ski wrote:

                hy,
Quoted 4 lines

"Losing too many ticks" exists when I'm running with cpufreq [p4_clockmod] and clock=tsc (default). This is because of rescaling TSC pitch by cpufreq, I think. If I specify in bootloader clock=hpet, problem disappears. [Am I doing right?]

If I boot with clock=hpet the cpu MHz in cat /proc/cpuinfo is 0.

Oh, yes. HPET timer needs ACPI support to work (also HPET is not present on many machines). You can probably find 'Warning: clock= override failed' in your dmesg. The default timer is then set to PIT. Determination of CPU clock when using PIT is not implemented in the kernel, so it's reported as zero in /proc/cpufreq. You may also find some patches around the LKML adding support for ACPI timer, but it needs ACPI itself. I didn't try it.

Szymon