Hello all
A while back there was quite a discussion on this issue and then
specifically "i8042 timing issues". I refer you to
http://lkml.org/lkml/2005/1/27/11 for more detail.
It turns out that it was the case that the i8042 controller responds too
late on commands, for example, we issue a AUX_LOOP command, whilst
waiting for the response it would time out, however, when issuing
AUX_TEST we will get the response for AUX_LOOP. It turns out this is
also true for the MUX and KBD parts of the driver. The attached patch
resolves this issue by providing a configurable option to issue commands
that expects results twice (not sure whether this is a good idea - but
it seems all commands either set some value or query some value).
Between the two commands it calls i8042_flush() to make sure that there
are not any earlier responses in the buffer.
I've built the patch against linux-2.6.11.6 and it should apply cleanly
(I've attached the patch cause it is about 80 lines in length and I fear
mozilla will break the formatting).
The patch fixes the problem for me - not sure how many other people out
there has the same issue. It provides a configurable option which can
be selected in the kernel configuration to actually enable this (albeit
enabling it should only result in a minor performance hit should it be
enabled needlessly - I think). Should the option not be enabled the
code remains 100% unchanged (except for one extra level of indentation).
I'm not sure what the policy regarding #ifdefs are, especially
considering they enclose the endpoints of a loop, please review and comment.
Jaco
--
There are only 10 kinds of people in this world,
those that understand binary and those that don't.
http://www.kroon.co.za/
diff -rau linux-2.6.11.6.orig/drivers/input/serio/Kconfig
linux-2.6.11.6/drivers/input/serio/Kconfig
--- linux-2.6.11.6.orig/drivers/input/serio/Kconfig 2005-03-26
05:28:21.000000000 +0200
+++ linux-2.6.11.6/drivers/input/serio/Kconfig 2005-04-04 12:01:01.000000000
+0200
@@ -31,6 +31,19 @@
To compile this driver as a module, choose M here: the
module will be called i8042.
+config SERIO_I8042_DELAYEDRESPONSEKLUDGE
+ tristate "i8042 delayed response kludge"
+ default n
+ depends on SERIO_I8042
+ ---help---
+ The Toshiba Satellite P10 series has a delayed response on the i8042
+ device which controls the mouse/keyboard. Whilst the keyboard can
+ function without this kludge it is required for the touchpad. Whilst
+ the kludge should not break other systems I highly recommend saying N
+ on all systems but those who suffer this problem.
+
+ If unsure, say N.
+
config SERIO_SERPORT
tristate "Serial port line discipline"
default y
diff -rau linux-2.6.11.6.orig/drivers/input/serio/i8042.c
linux-2.6.11.6/drivers/input/serio/i8042.c
--- linux-2.6.11.6.orig/drivers/input/serio/i8042.c 2005-03-26
05:28:15.000000000 +0200
+++ linux-2.6.11.6/drivers/input/serio/i8042.c 2005-04-04 12:26:15.000000000
+0200
@@ -185,24 +185,41 @@
unsigned long flags;
int retval = 0, i = 0;
+#ifdef CONFIG_SERIO_I8042_DELAYEDRESPONSEKLUDGE
+ int kludge = 0;
+
+ if((command >> 8) & 0xf)
+ kludge = 1;
+#endif
+
if (i8042_noloop && command == I8042_CMD_AUX_LOOP)
return -1;
-
+
spin_lock_irqsave(&i8042_lock, flags);
- retval = i8042_wait_write();
- if (!retval) {
- dbg("%02x -> i8042 (command)", command & 0xff);
- i8042_write_command(command & 0xff);
- }
-
- if (!retval)
- for (i = 0; i < ((command >> 12) & 0xf); i++) {
- if ((retval = i8042_wait_write())) break;
- dbg("%02x -> i8042 (parameter)", param[i]);
- i8042_write_data(param[i]);
+#ifdef CONFIG_SERIO_I8042_DELAYEDRESPONSEKLUDGE
+ do {
+#endif
+ retval = i8042_wait_write();
+ if (!retval) {
+ dbg("%02x -> i8042 (command)", command & 0xff);
+ i8042_write_command(command & 0xff);
}
+ if (!retval)
+ for (i = 0; i < ((command >> 12) & 0xf); i++) {
+ if ((retval = i8042_wait_write())) break;
+ dbg("%02x -> i8042 (parameter)", param[i]);
+ i8042_write_data(param[i]);
+ }
+
+#ifdef CONFIG_SERIO_I8042_DELAYEDRESPONSEKLUDGE
+ if(kludge)
+ i8042_flush();
+
+ } while(!retval && kludge--);
+#endif
+
if (!retval)
for (i = 0; i < ((command >> 8) & 0xf); i++) {
if ((retval = i8042_wait_read())) break;
