PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Neil Brown]

Hi,
 The following adds support for the ALPS glidepoint/dualpoint pointing
 devices to the mouse driver in 2.5.7x

 It "works-for-me" but there are issues that probably need to be
 addressed.

 1/ The code is based on other code fragments I have collected off the
   internet.  I have found no documentation and the one request I made to
   ALPS has so-far been unanswered.  So I cannot be sure it is right,
   but as I say it seems to work.

 2/ It appears (but see 1) that it is not possible to reliably detect
    an ALPS device.  There is a sequence where you send 3 SetRes2:1
    commands, and then a GetStatus command and you get something which
    isn't really a status, but I don't know what range of status
    values mean "ALPS".  I tried checking for "any status which must
    be wrong" i.e. any status that say the device is in 1:1 mode, or
    is enabled etc.  But a Logitech mouse seem to respond
    interestingly to that sequence too.

    Also, there is no guarantee that the reply will come from the ALPS
    device.  For example, on my Dell Latitude D800, if I have a
    logitech mouse plugged in the expansion port, the GetStatus reply
    comes from the logitech and not the ALPS.  So it would seem that
    reliable detection is impossible.

    So the current code always sends the ALPS set-absolute-mode
    sequence (4 disables before the enable) unless a
    non-3-byte-protocol device was detected.

    There are two consequences of always assuming an ALPS that may not
    be good.
     1) The mouse always claims to generate various ABS events even
       when there might not be any ABS-generating device behind the
       mouse.
     2) The driver could misinterpret a normal mouse event that
       overflowed in the negative direction for both X and Y as part
       of an ALPS absolute event.  This is because ALPS absolute
       events are detected by checking if the top 5 bits of the first
       byte are all one.   I doubt this is a real problem as double
       overflows are very unlikely (aren't they?)

  3/ I haven't set the Min and Max absolute values (though both X and
     Y seem to range from 0 to 100 in practice on my notebook).
     This was because declaring a Min and Max causes the mousedev
     driver to scale values to fit a supposed screen size, and I don't
     think that is really appropriate for a touchpad.  Would there be
     some other way to decide when to scale?  I would like to be able
     to include Min and Max so that a post-processor (possibly in
     mousedev) would be able to differentiate edge based activity
     (scroll regions, corner taps, etc).


  This patch also includes a fix for mousedev_event that allows it to
  work sensibly with a touchpad in absolute mode.  With a touchpad, if
  you lift your finger and place it down again you don't want that to
  be interpreted as movement, but mousedev_event currently will.
  I have changed it so that if ABS_PRESSURE is an available event,
  then a new ABS_{X,Y} event while the current ABS_PRESSURE is zero
  will not generate any movement.

 ----------- Diffstat output ------------
 ./drivers/input/mouse/psmouse-base.c |   83 ++++++++++++++++++++++++++++++++++-
 ./drivers/input/mouse/psmouse.h      |    1 
 ./drivers/input/mousedev.c           |   26 +++++++---
 3 files changed, 100 insertions(+), 10 deletions(-)

diff ./drivers/input/mouse/psmouse-base.c~current~ ./drivers/input/mouse/psmouse-base.c
--- ./drivers/input/mouse/psmouse-base.c~current~	2003-06-26 13:49:26.000000000 +1000
+++ ./drivers/input/mouse/psmouse-base.c	2003-06-26 13:51:49.000000000 +1000
@@ -99,6 +99,48 @@ static void psmouse_process_packet(struc
 }
 
 /*
+ * ALPS abolute Mode
+ * byte 0: 1 1 1 1 1 mid0 rig0 lef0
+ * byte 1: 0 x6 x5 x4 x3 x2 x1 x0
+ * byte 2: 0 x10 x9 x8 x7 0 fin ges
+ * byte 3: 0 y9 y8 y7 1 mid1 rig1 lef1
+ * byte 4: 0 y6 y5 y4 y3 y2 y1 y0
+ * byte 5: 0 z6 z5 z4 z3 z2 z1 z0
+ *
+ * On a dualpoint, {mid,rig,lef}0 are the stick, 1 are the pad.
+ * We just 'or' them together for now.
+ * We also send 'ges'ture as BTN_TOUCH
+ */
+
+static void ALPS_process_packet(struct psmouse *psmouse, struct pt_regs *regs)
+{
+	unsigned char *packet = psmouse->packet;
+	struct input_dev *dev = &psmouse->dev;
+	int x,y,z;
+
+	input_regs(dev, regs);
+
+	x = (packet[1] & 0x7f) | ((packet[2] & 0x78)<<(7-3));
+	y = (packet[4] & 0x7f) | ((packet[3] & 0x70)<<(7-4));
+	z = packet[5];
+
+	if (z > 0) {
+		input_report_abs(dev, ABS_X, x);
+		input_report_abs(dev, ABS_Y, y);
+		input_report_abs(dev, ABS_PRESSURE, z);
+	} else
+		input_report_abs(dev, ABS_PRESSURE, 0);
+
+	input_report_key(dev, BTN_LEFT,   ((packet[0] | packet[3])     ) & 1);
+	input_report_key(dev, BTN_MIDDLE, ((packet[0] | packet[3]) >> 2) & 1);
+	input_report_key(dev, BTN_RIGHT,  ((packet[0] | packet[3]) >> 1) & 1);
+
+	input_report_key(dev, BTN_TOUCH,  packet[2] & 1);
+
+	input_sync(dev);
+}
+
+/*
  * psmouse_interrupt() handles incoming characters, either gathering them into
  * packets or passing them to the command routine as command output.
  */
@@ -139,6 +181,19 @@ static irqreturn_t psmouse_interrupt(str
 	psmouse->last = jiffies;
 	psmouse->packet[psmouse->pktcnt++] = data;
 
+	/* ALPS absolute mode packets start with 0b11111mrl
+	 * Normal mouse packets are extremely unlikely to overflow both
+	 * x and y 
+	 */
+	if (!psmouse_noext && psmouse->type < PSMOUSE_GENPS &&
+	    (psmouse->packet[0] & 0xf8)== 0xf8) {
+		if (psmouse->pktcnt == 6) {
+			ALPS_process_packet(psmouse, regs);
+			psmouse->pktcnt = 0;
+		}
+		goto out;
+	}
+			
 	if (psmouse->pktcnt == 3 + (psmouse->type >= PSMOUSE_GENPS)) {
 		psmouse_process_packet(psmouse, regs);
 		psmouse->pktcnt = 0;
@@ -424,6 +479,17 @@ static void psmouse_set_resolution(struc
  * psmouse_initialize() initializes the mouse to a sane state.
  */
 
+static inline void set_abs_params(struct input_dev *dev, int axis, int min, int max, int fuzz, int flat)
+{
+	dev->absmin[axis] = min;
+	dev->absmax[axis] = max;
+	dev->absfuzz[axis] = fuzz;
+	dev->absflat[axis] = flat;
+
+	set_bit(axis, dev->absbit);
+}
+
+
 static void psmouse_initialize(struct psmouse *psmouse)
 {
 	unsigned char param[2];
@@ -453,8 +519,23 @@ static void psmouse_initialize(struct ps
 
 /*
  * Last, we enable the mouse so that we get reports from it.
- */
+ * If it is a 3-byte setting and we are allowed to use extensions,
+ * then it could be an ALPS Glidepoint, so send the init sequence just
+ * incase. i.e. 4 consecutive "disable"s before the "enable"
+ */
+	if (psmouse->type < PSMOUSE_GENPS && !psmouse_noext) {
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+
+		set_bit(BTN_TOUCH, psmouse->dev.keybit);
+		set_bit(EV_ABS, psmouse->dev.evbit);
+		set_abs_params(&psmouse->dev, ABS_X, 0, 0, 0, 0);
+		set_abs_params(&psmouse->dev, ABS_Y, 0, 0, 0, 0);
+		set_abs_params(&psmouse->dev, ABS_PRESSURE, 0, 127, 0, 0);
 
+	}
 	if (psmouse_command(psmouse, NULL, PSMOUSE_CMD_ENABLE))
 		printk(KERN_WARNING "psmouse.c: Failed to enable mouse on %s\n", psmouse->serio->phys);
 

diff ./drivers/input/mouse/psmouse.h~current~ ./drivers/input/mouse/psmouse.h
--- ./drivers/input/mouse/psmouse.h~current~	2003-06-26 13:49:26.000000000 +1000
+++ ./drivers/input/mouse/psmouse.h	2003-06-26 13:50:16.000000000 +1000
@@ -9,6 +9,7 @@
 #define PSMOUSE_CMD_GETID	0x02f2
 #define PSMOUSE_CMD_SETRATE	0x10f3
 #define PSMOUSE_CMD_ENABLE	0x00f4
+#define	PSMOUSE_CMD_DISABLE	0x00f5
 #define PSMOUSE_CMD_RESET_DIS	0x00f6
 #define PSMOUSE_CMD_RESET_BAT	0x02ff
 

diff ./drivers/input/mousedev.c~current~ ./drivers/input/mousedev.c
--- ./drivers/input/mousedev.c~current~	2003-06-26 13:49:26.000000000 +1000
+++ ./drivers/input/mousedev.c	2003-06-26 13:50:22.000000000 +1000
@@ -53,7 +53,7 @@ struct mousedev_list {
 	struct fasync_struct *fasync;
 	struct mousedev *mousedev;
 	struct list_head node;
-	int dx, dy, dz, oldx, oldy;
+	int dx, dy, dz, oldx, oldy, finger;
 	signed char ps2[6];
 	unsigned long buttons;
 	unsigned char ready, buffer, bufsiz;
@@ -79,6 +79,7 @@ static void mousedev_event(struct input_
 	struct mousedev **mousedev = mousedevs;
 	struct mousedev_list *list;
 	int index, size, wake;
+	int dx, dy;
 
 	while (*mousedev) {
 
@@ -93,23 +94,30 @@ static void mousedev_event(struct input_
 						case ABS_X:	
 							size = handle->dev->absmax[ABS_X] - handle->dev->absmin[ABS_X];
 							if (size != 0) {
-								list->dx += (value * xres - list->oldx) / size;
-								list->oldx += list->dx * size;
+								dx = list->dx + (value * xres - list->oldx) / size;
+								list->oldx += dx * size;
 							} else {
-								list->dx += value - list->oldx;
-								list->oldx += list->dx;
+								dx = list->dx + value - list->oldx;
+								list->oldx += dx;
 							}
+							if (list->finger || !test_bit(ABS_PRESSURE, handle->dev->absbit))
+								list->dx = dx;
 							break;
 
 						case ABS_Y:
 							size = handle->dev->absmax[ABS_Y] - handle->dev->absmin[ABS_Y];
 							if (size != 0) {
-								list->dy -= (value * yres - list->oldy) / size;
-								list->oldy -= list->dy * size;
+								dy = list->dy - (value * yres - list->oldy) / size;
+								list->oldy -= dy * size;
 							} else {
-								list->dy -= value - list->oldy;
-								list->oldy -= list->dy;
+								dy = list->dy - (value - list->oldy);
+								list->oldy -= dy;
 							}
+							if (list->finger || !test_bit(ABS_PRESSURE, handle->dev->absbit))
+								list->dy = dy;
+							break;
+						case ABS_PRESSURE:
+							list->finger = value;
 							break;
 					}
 					break;

Re: PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Felipe Alfaro Solana]

On Fri, 2003-06-27 at 04:38, Neil Brown wrote:
> Hi,
>  The following adds support for the ALPS glidepoint/dualpoint pointing
>  devices to the mouse driver in 2.5.7x
> 
>  It "works-for-me" but there are issues that probably need to be
>  addressed.
> 
>  1/ The code is based on other code fragments I have collected off the
>    internet.  I have found no documentation and the one request I made to
>    ALPS has so-far been unanswered.  So I cannot be sure it is right,
>    but as I say it seems to work.
> 
>  2/ It appears (but see 1) that it is not possible to reliably detect
>     an ALPS device.  There is a sequence where you send 3 SetRes2:1
>     commands, and then a GetStatus command and you get something which
>     isn't really a status, but I don't know what range of status
>     values mean "ALPS".  I tried checking for "any status which must
>     be wrong" i.e. any status that say the device is in 1:1 mode, or
>     is enabled etc.  But a Logitech mouse seem to respond
>     interestingly to that sequence too.
> 
>     Also, there is no guarantee that the reply will come from the ALPS
>     device.  For example, on my Dell Latitude D800, if I have a
>     logitech mouse plugged in the expansion port, the GetStatus reply
>     comes from the logitech and not the ALPS.  So it would seem that
>     reliable detection is impossible.
> 
>     So the current code always sends the ALPS set-absolute-mode
>     sequence (4 disables before the enable) unless a
>     non-3-byte-protocol device was detected.
> 
>     There are two consequences of always assuming an ALPS that may not
>     be good.
>      1) The mouse always claims to generate various ABS events even
>        when there might not be any ABS-generating device behind the
>        mouse.
>      2) The driver could misinterpret a normal mouse event that
>        overflowed in the negative direction for both X and Y as part
>        of an ALPS absolute event.  This is because ALPS absolute
>        events are detected by checking if the top 5 bits of the first
>        byte are all one.   I doubt this is a real problem as double
>        overflows are very unlikely (aren't they?)
> 
>   3/ I haven't set the Min and Max absolute values (though both X and
>      Y seem to range from 0 to 100 in practice on my notebook).
>      This was because declaring a Min and Max causes the mousedev
>      driver to scale values to fit a supposed screen size, and I don't
>      think that is really appropriate for a touchpad.  Would there be
>      some other way to decide when to scale?  I would like to be able
>      to include Min and Max so that a post-processor (possibly in
>      mousedev) would be able to differentiate edge based activity
>      (scroll regions, corner taps, etc).
> 
> 
>   This patch also includes a fix for mousedev_event that allows it to
>   work sensibly with a touchpad in absolute mode.  With a touchpad, if
>   you lift your finger and place it down again you don't want that to
>   be interpreted as movement, but mousedev_event currently will.
>   I have changed it so that if ABS_PRESSURE is an available event,
>   then a new ABS_{X,Y} event while the current ABS_PRESSURE is zero
>   will not generate any movement.

Is there any trick to force enabling ALPS support? I'm using a NEC
Chrom@ laptop with an ALPS GlidePoint touchpad, 2.5.73-mm1 and your
patch, but I can't seem to get the enhanced functionality of my
touchpad, like using the edges of the touchpad to simulate the wheel or
else. It seems to behave like a normal PS/2 mouse.

Also, on dmesg, I can't see any reference to an ALPS input device being
detected. Any ideas?

Thanks!

Re: PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Andrey Panin]

[-- Attachment #1: Type: text/plain, Size: 525 bytes --]

On 178, 06 27, 2003 at 09:41:28AM +0200, Felipe Alfaro Solana wrote:
> 
> Is there any trick to force enabling ALPS support? I'm using a NEC
> Chrom@ laptop with an ALPS GlidePoint touchpad, 2.5.73-mm1 and your
> patch, but I can't seem to get the enhanced functionality of my
> touchpad, like using the edges of the touchpad to simulate the wheel or
> else. It seems to behave like a normal PS/2 mouse.

DMI ?

-- 
Andrey Panin		| Linux and UNIX system administrator
pazke@donpac.ru		| PGP key: wwwkeys.pgp.net

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Neil Brown]

On  June 27, felipe_alfaro@linuxmail.org wrote:
> 
> Is there any trick to force enabling ALPS support? I'm using a NEC
> Chrom@ laptop with an ALPS GlidePoint touchpad, 2.5.73-mm1 and your
> patch, but I can't seem to get the enhanced functionality of my
> touchpad, like using the edges of the touchpad to simulate the wheel or
> else. It seems to behave like a normal PS/2 mouse.

Well, if it behaves like a normal PS/2 mouse, it is quite possibly
working :-)  
That patch didn't add any obvious new functionality.  It just change
things to the ALPS device was used in absolute mode.
If you manage to find evtest.c, you can watch events on
/dev/input/event1 (or similar) and see the absolute event.

The next step is adding scroll-edge functionality and similar things
to mousedev.c 
I've almost got it so that when yor finger hits the edge of the
touchpad, the mouse keeps moving, and moves faster if you press
harder.   Once I'm happy with that I'll post it and start on the
scroll-wheel thing.

> 
> Also, on dmesg, I can't see any reference to an ALPS input device being
> detected. Any ideas?

No, the device isn't detected exactly.  See point 2 in the original
mail.

NeilBrown

Re: PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Johan Braennlund]

At first I tried posting this to linux.kernel, since I thought that was a
bidirectional gateway, but the post hasn't showed up. Apologies if you see
this twice.

Neil Brown <neilb@cse.unsw.edu.au> wrote:

> Hi,
>  The following adds support for the ALPS glidepoint/dualpoint pointing
>  devices to the mouse driver in 2.5.7x

>  It "works-for-me" but there are issues that probably need to be
>  addressed.

Hi! Thank you for the patch. It looks interesting, but unfortunately it
doesn't work very well for me. I have an Acer Aspire laptop with a
four-button Alps touchpad (left,right and two up/down scroll buttons). The
"down" button functions as the middle mouse button, but I've never been
able to get the "up" button properly recognized under linux.

If I apply your patch to 2.5.73, none of the buttons work (except for
tapping for left-click), neither in X nor with gpm, but touchpad movement
works fine, with increased sensitivity compared to the standard driver.
Unpatched 2.5.73 works as expected.

gpm commandline: gpm -t ps/2 -m /dev/psaux

XF86Config:

Section "InputDevice"
        Identifier      "Configured Mouse"
        Driver          "mouse"
        Option          "CorePointer"
        Option          "Device"                "/dev/psaux"
        Option          "Protocol"              "GlidePointPS/2"
        Option          "Buttons"               "4"
EndSection

My C skills are pretty meager but if there's anything I can do in the way
of debugging, I'd be happy to.

Johan

Re: PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Neil Brown]

On Sunday June 29, spahmtrahp@yahoo.com wrote:
> 
> Hi! Thank you for the patch. It looks interesting, but unfortunately it
> doesn't work very well for me. I have an Acer Aspire laptop with a
> four-button Alps touchpad (left,right and two up/down scroll buttons). The
> "down" button functions as the middle mouse button, but I've never been
> able to get the "up" button properly recognized under linux.
> 
> If I apply your patch to 2.5.73, none of the buttons work (except for
> tapping for left-click), neither in X nor with gpm, but touchpad movement
> works fine, with increased sensitivity compared to the standard driver.
> Unpatched 2.5.73 works as expected.

Hmmm... the joys of no documentation....

Can you
  grab  http://www.cse.unsw.edu.au/~neilb/D800/mouseplay.c
  compile it
  boot into a 2.4 kernel
  Disable gpm and get out of X
  run mouseplay capturing the output 
    e.g. mouseplay | tee /tmp/m.out
  perform various mouse action, particularly the buttons
  annotate /tmp/m.out to tell me what you did when,
  send m.out to me.

Thanks,
NeilBrown

 

Re: PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Peter Osterlund]

Neil Brown <neilb@cse.unsw.edu.au> writes:

> On  June 27, felipe_alfaro@linuxmail.org wrote:
> > 
> > Is there any trick to force enabling ALPS support? I'm using a NEC
> > Chrom@ laptop with an ALPS GlidePoint touchpad, 2.5.73-mm1 and your
> > patch, but I can't seem to get the enhanced functionality of my
> > touchpad, like using the edges of the touchpad to simulate the wheel or
> > else. It seems to behave like a normal PS/2 mouse.
> 
> Well, if it behaves like a normal PS/2 mouse, it is quite possibly
> working :-)  
> That patch didn't add any obvious new functionality.  It just change
> things to the ALPS device was used in absolute mode.
> If you manage to find evtest.c, you can watch events on
> /dev/input/event1 (or similar) and see the absolute event.
> 
> The next step is adding scroll-edge functionality and similar things
> to mousedev.c 
> I've almost got it so that when yor finger hits the edge of the
> touchpad, the mouse keeps moving, and moves faster if you press
> harder.   Once I'm happy with that I'll post it and start on the
> scroll-wheel thing.

Hi!

I have updated your patch for kernel 2.6.0-test6-bk6 and made it
report events compatible with the synaptics touchpad kernel driver.
This should make it possible to use an ALPS device with the XFree86
synaptics driver:

        http://w1.894.telia.com/~u89404340/touchpad/index.html

Using this driver will give you edge scrolling and similar things.

I don't have an ALPS GlidePoint so I haven't been able to test this
patch at all. Test reports are appreciated. You probably need to
change a few parameters in the X configuration, like edge parameters
and finger pressure thresholds. Also note that the auto detection will
not work with an ALPS device, so you have to use Protocol="event" and
Device="/dev/input/eventN" for some value of N.


 linux-petero/drivers/input/mouse/psmouse-base.c |   86 ++++++++++++++++++++++++
 linux-petero/drivers/input/mouse/psmouse.h      |    1 
 2 files changed, 87 insertions(+)

diff -puN drivers/input/mouse/psmouse-base.c~alps drivers/input/mouse/psmouse-base.c
--- linux/drivers/input/mouse/psmouse-base.c~alps	2003-10-05 09:00:49.000000000 +0200
+++ linux-petero/drivers/input/mouse/psmouse-base.c	2003-10-05 18:30:40.000000000 +0200
@@ -106,6 +106,48 @@ static void psmouse_process_packet(struc
 }
 
 /*
+ * ALPS abolute Mode
+ * byte 0: 1 1 1 1 1 mid0 rig0 lef0
+ * byte 1: 0 x6 x5 x4 x3 x2 x1 x0
+ * byte 2: 0 x10 x9 x8 x7 0 fin ges
+ * byte 3: 0 y9 y8 y7 1 mid1 rig1 lef1
+ * byte 4: 0 y6 y5 y4 y3 y2 y1 y0
+ * byte 5: 0 z6 z5 z4 z3 z2 z1 z0
+ *
+ * On a dualpoint, {mid,rig,lef}0 are the stick, 1 are the pad.
+ * We just 'or' them together for now.
+ * We also send 'ges'ture as BTN_TOUCH
+ */
+
+static void ALPS_process_packet(struct psmouse *psmouse, struct pt_regs *regs)
+{
+	unsigned char *packet = psmouse->packet;
+	struct input_dev *dev = &psmouse->dev;
+	int x,y,z;
+
+	input_regs(dev, regs);
+
+	x = (packet[1] & 0x7f) | ((packet[2] & 0x78)<<(7-3));
+	y = (packet[4] & 0x7f) | ((packet[3] & 0x70)<<(7-4));
+	z = packet[5];
+
+	if (z > 0) {
+		input_report_abs(dev, ABS_X, x);
+		input_report_abs(dev, ABS_Y, y);
+	}
+	input_report_abs(dev, ABS_PRESSURE, z);
+	input_report_key(dev, BTN_TOOL_FINGER, z > 0);
+
+	input_report_key(dev, BTN_LEFT,   ((packet[0] | packet[3])     ) & 1);
+	input_report_key(dev, BTN_MIDDLE, ((packet[0] | packet[3]) >> 2) & 1);
+	input_report_key(dev, BTN_RIGHT,  ((packet[0] | packet[3]) >> 1) & 1);
+
+	input_report_key(dev, BTN_TOUCH,  packet[2] & 1);
+
+	input_sync(dev);
+}
+
+/*
  * psmouse_interrupt() handles incoming characters, either gathering them into
  * packets or passing them to the command routine as command output.
  */
@@ -180,6 +222,19 @@ static irqreturn_t psmouse_interrupt(str
 		goto out;
 	}
 
+	/* ALPS absolute mode packets start with 0b11111mrl
+	 * Normal mouse packets are extremely unlikely to overflow both
+	 * x and y
+	 */
+	if (!psmouse_noext && psmouse->type < PSMOUSE_GENPS &&
+	    (psmouse->packet[0] & 0xf8)== 0xf8) {
+		if (psmouse->pktcnt == 6) {
+			ALPS_process_packet(psmouse, regs);
+			psmouse->pktcnt = 0;
+		}
+		goto out;
+	}
+
 	if (psmouse->pktcnt == 3 + (psmouse->type >= PSMOUSE_GENPS)) {
 		psmouse_process_packet(psmouse, regs);
 		psmouse->pktcnt = 0;
@@ -463,6 +518,17 @@ static void psmouse_set_rate(struct psmo
  * psmouse_initialize() initializes the mouse to a sane state.
  */
 
+static inline void set_abs_params(struct input_dev *dev, int axis, int min, int max, int fuzz, int flat)
+{
+	dev->absmin[axis] = min;
+	dev->absmax[axis] = max;
+	dev->absfuzz[axis] = fuzz;
+	dev->absflat[axis] = flat;
+
+	set_bit(axis, dev->absbit);
+}
+
+
 static void psmouse_initialize(struct psmouse *psmouse)
 {
 	unsigned char param[2];
@@ -486,6 +552,26 @@ static void psmouse_initialize(struct ps
  */
 
 	psmouse_command(psmouse, param, PSMOUSE_CMD_SETSTREAM);
+
+/*
+ * If it is a 3-byte setting and we are allowed to use extensions,
+ * then it could be an ALPS Glidepoint, so send the init sequence just
+ * incase. i.e. 4 consecutive "disable"s before the "enable"
+ */
+	if (psmouse->type < PSMOUSE_GENPS && !psmouse_noext) {
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+
+		set_bit(BTN_TOUCH, psmouse->dev.keybit);
+		set_bit(BTN_TOOL_FINGER, psmouse->dev.keybit);
+		set_bit(EV_ABS, psmouse->dev.evbit);
+		set_abs_params(&psmouse->dev, ABS_X, 0, 0, 0, 0);
+		set_abs_params(&psmouse->dev, ABS_Y, 0, 0, 0, 0);
+		set_abs_params(&psmouse->dev, ABS_PRESSURE, 0, 127, 0, 0);
+
+	}
 }
 
 /*
diff -puN drivers/input/mouse/psmouse.h~alps drivers/input/mouse/psmouse.h
--- linux/drivers/input/mouse/psmouse.h~alps	2003-10-05 09:00:49.000000000 +0200
+++ linux-petero/drivers/input/mouse/psmouse.h	2003-10-05 09:00:49.000000000 +0200
@@ -9,6 +9,7 @@
 #define PSMOUSE_CMD_GETID	0x02f2
 #define PSMOUSE_CMD_SETRATE	0x10f3
 #define PSMOUSE_CMD_ENABLE	0x00f4
+#define	PSMOUSE_CMD_DISABLE	0x00f5
 #define PSMOUSE_CMD_RESET_DIS	0x00f6
 #define PSMOUSE_CMD_RESET_BAT	0x02ff
 

_

-- 
Peter Osterlund - petero2@telia.com
http://w1.894.telia.com/~u89404340

Re: PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Vojtech Pavlik]

On Sun, Oct 05, 2003 at 06:55:31PM +0200, Peter Osterlund wrote:
 
> Hi!
> 
> I have updated your patch for kernel 2.6.0-test6-bk6 and made it
> report events compatible with the synaptics touchpad kernel driver.
> This should make it possible to use an ALPS device with the XFree86
> synaptics driver:
> 
>         http://w1.894.telia.com/~u89404340/touchpad/index.html
> 
> Using this driver will give you edge scrolling and similar things.
> 
> I don't have an ALPS GlidePoint so I haven't been able to test this
> patch at all. Test reports are appreciated. You probably need to
> change a few parameters in the X configuration, like edge parameters
> and finger pressure thresholds. Also note that the auto detection will
> not work with an ALPS device, so you have to use Protocol="event" and
> Device="/dev/input/eventN" for some value of N.

Very nice. Could you also make it a separate file? I think it's enough
code to make that worth it ...

>  linux-petero/drivers/input/mouse/psmouse-base.c |   86 ++++++++++++++++++++++++
>  linux-petero/drivers/input/mouse/psmouse.h      |    1 
>  2 files changed, 87 insertions(+)

-- 
Vojtech Pavlik
SuSE Labs, SuSE CR

Re: PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Peter Osterlund]

Vojtech Pavlik <vojtech@suse.cz> writes:

> On Sun, Oct 05, 2003 at 06:55:31PM +0200, Peter Osterlund wrote:
>  
> > I have updated your patch for kernel 2.6.0-test6-bk6 and made it
> > report events compatible with the synaptics touchpad kernel driver.
> > This should make it possible to use an ALPS device with the XFree86
> > synaptics driver:
> > 
> >         http://w1.894.telia.com/~u89404340/touchpad/index.html
> > 
> > Using this driver will give you edge scrolling and similar things.
> > 
> > I don't have an ALPS GlidePoint so I haven't been able to test this
> > patch at all. Test reports are appreciated. You probably need to
> > change a few parameters in the X configuration, like edge parameters
> > and finger pressure thresholds. Also note that the auto detection will
> > not work with an ALPS device, so you have to use Protocol="event" and
> > Device="/dev/input/eventN" for some value of N.
> 
> Very nice. Could you also make it a separate file? I think it's enough
> code to make that worth it ...

Here is a new patch that moves the ALPS code to a separate file. The
logic in ALPS_process_packet() has also gone through a few iterations
with Johan Braennlund as a tester, so that it actually even works now,
at least with his touchpad.

Note though that the potential problems mentioned by Neil Brown in his
initial announcement have not been dealt with:

    It appears (but see 1) that it is not possible to reliably detect
    an ALPS device.
    ...
    So the current code always sends the ALPS set-absolute-mode
    sequence (4 disables before the enable) unless a
    non-3-byte-protocol device was detected.

    There are two consequences of always assuming an ALPS that may not
    be good.
     1) The mouse always claims to generate various ABS events even
       when there might not be any ABS-generating device behind the
       mouse.
     2) The driver could misinterpret a normal mouse event that
       overflowed in the negative direction for both X and Y as part
       of an ALPS absolute event.  This is because ALPS absolute
       events are detected by checking if the top 5 bits of the first
       byte are all one.   I doubt this is a real problem as double
       overflows are very unlikely (aren't they?)


 linux-petero/drivers/input/mouse/Makefile       |    2 
 linux-petero/drivers/input/mouse/alps.c         |  131 ++++++++++++++++++++++++
 linux-petero/drivers/input/mouse/alps.h         |   17 +++
 linux-petero/drivers/input/mouse/psmouse-base.c |    7 +
 linux-petero/drivers/input/mouse/psmouse.h      |    1 
 5 files changed, 157 insertions(+), 1 deletion(-)

diff -puN drivers/input/mouse/Makefile~alps drivers/input/mouse/Makefile
--- linux/drivers/input/mouse/Makefile~alps	2003-10-08 18:30:40.000000000 +0200
+++ linux-petero/drivers/input/mouse/Makefile	2003-10-08 18:30:40.000000000 +0200
@@ -14,5 +14,5 @@ obj-$(CONFIG_MOUSE_PC9800)	+= 98busmouse
 obj-$(CONFIG_MOUSE_PS2)		+= psmouse.o
 obj-$(CONFIG_MOUSE_SERIAL)	+= sermouse.o
 
-psmouse-objs				:= psmouse-base.o logips2pp.o
+psmouse-objs				:= psmouse-base.o logips2pp.o alps.o
 psmouse-$(CONFIG_MOUSE_PS2_SYNAPTICS)	+= synaptics.o
diff -puN drivers/input/mouse/alps.c~alps drivers/input/mouse/alps.c
--- linux/drivers/input/mouse/alps.c~alps	2003-10-08 18:30:40.000000000 +0200
+++ linux-petero/drivers/input/mouse/alps.c	2003-10-08 18:30:40.000000000 +0200
@@ -0,0 +1,131 @@
+/*
+ * Logitech PS/2++ mouse driver
+ *
+ * Copyright (c) 2003 Neil Brown <neilb@cse.unsw.edu.au>
+ * Copyright (c) 2003 Peter Osterlund <petero2@telia.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published by
+ * the Free Software Foundation.
+ */
+
+#include "alps.h"
+#include <linux/input.h>
+#include <linux/serio.h>
+#include "psmouse.h"
+
+static inline void set_abs_params(struct input_dev *dev, int axis, int min, int max, int fuzz, int flat)
+{
+	dev->absmin[axis] = min;
+	dev->absmax[axis] = max;
+	dev->absfuzz[axis] = fuzz;
+	dev->absflat[axis] = flat;
+
+	set_bit(axis, dev->absbit);
+}
+
+/*
+ * If it is a 3-byte setting and we are allowed to use extensions,
+ * then it could be an ALPS Glidepoint, so send the init sequence just
+ * incase. i.e. 4 consecutive "disable"s before the "enable"
+ */
+
+void ALPS_initialize(struct psmouse *psmouse)
+{
+	if (psmouse->type < PSMOUSE_GENPS) {
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+		psmouse_command(psmouse, NULL, PSMOUSE_CMD_DISABLE);
+
+		set_bit(BTN_TOUCH, psmouse->dev.keybit);
+		set_bit(BTN_BACK, psmouse->dev.keybit);
+		set_bit(BTN_FORWARD, psmouse->dev.keybit);
+		set_bit(BTN_TOOL_FINGER, psmouse->dev.keybit);
+		set_bit(EV_ABS, psmouse->dev.evbit);
+		set_abs_params(&psmouse->dev, ABS_X, 0, 0, 0, 0);
+		set_abs_params(&psmouse->dev, ABS_Y, 0, 0, 0, 0);
+		set_abs_params(&psmouse->dev, ABS_PRESSURE, 0, 127, 0, 0);
+	}
+}
+
+/*
+ * ALPS abolute Mode
+ * byte 0: 1 1 1 1 1 mid0 rig0 lef0
+ * byte 1: 0 x6 x5 x4 x3 x2 x1 x0
+ * byte 2: 0 x10 x9 x8 x7 up1 fin ges
+ * byte 3: 0 y9 y8 y7 1 mid1 rig1 lef1
+ * byte 4: 0 y6 y5 y4 y3 y2 y1 y0
+ * byte 5: 0 z6 z5 z4 z3 z2 z1 z0
+ *
+ * On a dualpoint, {mid,rig,lef}0 are the stick, 1 are the pad.
+ * We just 'or' them together for now.
+ * We also send 'ges'ture as BTN_TOUCH
+ *
+ * The touchpad on an 'Acer Aspire' has 4 buttons:
+ *   left,right,up,down.
+ * This device always sets {mid,rig,lef}0 to 1 and
+ * reflects left,right,down,up in lef1,rig1,mid1,up1.
+ */
+
+static void ALPS_process_packet(struct psmouse *psmouse, struct pt_regs *regs)
+{
+	unsigned char *packet = psmouse->packet;
+	struct input_dev *dev = &psmouse->dev;
+	int x, y, z;
+	int left = 0, right = 0, middle = 0;
+
+	input_regs(dev, regs);
+
+	x = (packet[1] & 0x7f) | ((packet[2] & 0x78)<<(7-3));
+	y = (packet[4] & 0x7f) | ((packet[3] & 0x70)<<(7-4));
+	z = packet[5];
+
+	if (z > 0) {
+		input_report_abs(dev, ABS_X, x);
+		input_report_abs(dev, ABS_Y, y);
+	}
+	input_report_abs(dev, ABS_PRESSURE, z);
+	input_report_key(dev, BTN_TOOL_FINGER, z > 0);
+
+	if (z > 30) input_report_key(dev, BTN_TOUCH, 1);
+	if (z < 25) input_report_key(dev, BTN_TOUCH, 0);
+
+	left  |= (packet[2]     ) & 1;
+	left  |= (packet[3]     ) & 1;
+	right |= (packet[3] >> 1) & 1;
+	if (packet[0] == 0xff) {
+		input_report_key(dev, BTN_BACK,    (packet[3] >> 2) & 1);
+		input_report_key(dev, BTN_FORWARD, (packet[2] >> 2) & 1);
+	} else {
+		left   |= (packet[0]     ) & 1;
+		right  |= (packet[0] >> 1) & 1;
+		middle |= (packet[0] >> 2) & 1;
+		middle |= (packet[3] >> 2) & 1;
+	}
+
+	input_report_key(dev, BTN_LEFT, left);
+	input_report_key(dev, BTN_RIGHT, right);
+	input_report_key(dev, BTN_MIDDLE, middle);
+
+	input_sync(dev);
+}
+
+int ALPS_process_byte(struct psmouse *psmouse, struct pt_regs *regs)
+{
+	if (psmouse->type >= PSMOUSE_GENPS)
+		return -1;
+
+	/* ALPS absolute mode packets start with 0b11111mrl
+	 * Normal mouse packets are extremely unlikely to overflow both
+	 * x and y
+	 */
+	if ((psmouse->packet[0] & 0xf8) != 0xf8)
+		return -1;
+
+	if (psmouse->pktcnt == 6) {
+		ALPS_process_packet(psmouse, regs);
+		psmouse->pktcnt = 0;
+	}
+	return 0;
+}
diff -puN drivers/input/mouse/alps.h~alps drivers/input/mouse/alps.h
--- linux/drivers/input/mouse/alps.h~alps	2003-10-08 18:30:40.000000000 +0200
+++ linux-petero/drivers/input/mouse/alps.h	2003-10-08 20:23:41.000000000 +0200
@@ -0,0 +1,17 @@
+/*
+ * Logitech PS/2++ mouse driver
+ *
+ * Copyright (c) 2003 Peter Osterlund <petero2@telia.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published by
+ * the Free Software Foundation.
+ */
+
+#ifndef _ALPS_H
+#define _ALPS_H
+struct psmouse;
+struct pt_regs;
+void ALPS_initialize(struct psmouse *psmouse);
+int ALPS_process_byte(struct psmouse *psmouse, struct pt_regs *regs);
+#endif
diff -puN drivers/input/mouse/psmouse-base.c~alps drivers/input/mouse/psmouse-base.c
--- linux/drivers/input/mouse/psmouse-base.c~alps	2003-10-08 18:30:40.000000000 +0200
+++ linux-petero/drivers/input/mouse/psmouse-base.c	2003-10-08 18:30:40.000000000 +0200
@@ -21,6 +21,7 @@
 #include "psmouse.h"
 #include "synaptics.h"
 #include "logips2pp.h"
+#include "alps.h"
 
 MODULE_AUTHOR("Vojtech Pavlik <vojtech@suse.cz>");
 MODULE_DESCRIPTION("PS/2 mouse driver");
@@ -180,6 +181,9 @@ static irqreturn_t psmouse_interrupt(str
 		goto out;
 	}
 
+	if (!psmouse_noext && !ALPS_process_byte(psmouse, regs))
+		goto out;
+
 	if (psmouse->pktcnt == 3 + (psmouse->type >= PSMOUSE_GENPS)) {
 		psmouse_process_packet(psmouse, regs);
 		psmouse->pktcnt = 0;
@@ -486,6 +490,9 @@ static void psmouse_initialize(struct ps
  */
 
 	psmouse_command(psmouse, param, PSMOUSE_CMD_SETSTREAM);
+
+	if (!psmouse_noext)
+		ALPS_initialize(psmouse);
 }
 
 /*
diff -puN drivers/input/mouse/psmouse.h~alps drivers/input/mouse/psmouse.h
--- linux/drivers/input/mouse/psmouse.h~alps	2003-10-08 18:30:40.000000000 +0200
+++ linux-petero/drivers/input/mouse/psmouse.h	2003-10-08 18:30:40.000000000 +0200
@@ -9,6 +9,7 @@
 #define PSMOUSE_CMD_GETID	0x02f2
 #define PSMOUSE_CMD_SETRATE	0x10f3
 #define PSMOUSE_CMD_ENABLE	0x00f4
+#define PSMOUSE_CMD_DISABLE	0x00f5
 #define PSMOUSE_CMD_RESET_DIS	0x00f6
 #define PSMOUSE_CMD_RESET_BAT	0x02ff
 

_

-- 
Peter Osterlund - petero2@telia.com
http://w1.894.telia.com/~u89404340

Re: PATCH - ALPS glidepoint/dualpoint driver for 2.5.7x

[Neil Brown]

On Sunday October 5, vojtech@suse.cz wrote:
> On Sun, Oct 05, 2003 at 06:55:31PM +0200, Peter Osterlund wrote:
>  
> > Hi!
> > 
> > I have updated your patch for kernel 2.6.0-test6-bk6 and made it
> > report events compatible with the synaptics touchpad kernel driver.
> > This should make it possible to use an ALPS device with the XFree86
> > synaptics driver:
> > 
> >         http://w1.894.telia.com/~u89404340/touchpad/index.html
> > 
> > Using this driver will give you edge scrolling and similar things.
> > 
> > I don't have an ALPS GlidePoint so I haven't been able to test this
> > patch at all. Test reports are appreciated. You probably need to
> > change a few parameters in the X configuration, like edge parameters
> > and finger pressure thresholds. Also note that the auto detection will
> > not work with an ALPS device, so you have to use Protocol="event" and
> > Device="/dev/input/eventN" for some value of N.
> 
> Very nice. Could you also make it a separate file? I think it's enough
> code to make that worth it ...

I would actually much rather not have this code in the kernel at all.
Given that I cannot find any documentation of the ALPS interface and
have no confidence of being able to detect an ALPS device or that
different ALPS devices behave the same way, I simply don't think this
stuff belongs in the kernel.

What I would much rather do is have /dev/psaux really be a char-dev
interface to the PS2 AUX port on the computer (rather than a faked
ps-aux look alike sythesised from all available mice) and have a
user-space program read that device, interpret it in a
user-configurable way, and send mouse events into /dev/input/uinput so
that other parts of the system see the appropriate moouse events.

I have code that does this but haven't had time to sort out a few
remaining little issues.

What I would *really* like to do is change /dev/psaux so that:
  If it is opened by user-space, it gets all chars from the AUX port,
  and input/mousedev doesn't see them.
  If it is not open, chars from /dev/psaux would bet processed by the
  current mouse driver.

Is there any chance that this might be accepted?

NeilBrown