Elantech touchpad detected as Logitech PS2 Wheel Mouse

Hi,

I'm running debian wheezy with latest kernel (uname -a gives me: Linux ehehehpc 3.2.0-2-686-pae #1 SMP Sun May 13 07:51:23 UTC 2012 i686 GNU/Linux).
While I'm expecting the kernel to correctly detect my touchpad, it's detected as a mouse. No scrolling, no multitouch.
Below there's a snippet of my /proc/bus/input/devices:

[...]
I: Bus=0011 Vendor=0002 Product=0001 Version=0063
N: Name="PS/2 Logitech Wheel Mouse"
P: Phys=isa0060/serio4/input0
S: Sysfs=/devices/platform/i8042/serio4/input/input7
U: Uniq=
H: Handlers=mouse0 event7
B: PROP=0
B: EV=7
B: KEY=70000 0 0 0 0 0 0 0 0
B: REL=103
[...]

The notebook I'm using is a Fujitsu Siemens Amilo pi2515. I also tried kernel version 3.3.6 with same results. In both cases grep ELANTECH .config gives me:

CONFIG_MOUSE_PS2_ELANTECH=y

so I suppose it shouldn't be a kernel configuration problem. I opened a bug on http://bugs.debian.org/673588
You can find more information and preliminary testing there. Thank you very much for your attention.
Regards.

Nicola

Kernel log:

http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=15;filename=dmesg.txt;att=3;bug=673588

[...]

Can you get a log with the following patch applied?

Thanks,
Jonathan

index d2c0db159b18..a1d59e11e2d3 100644
@@ -10,6 +10,7 @@
  * Trademarks are the property of their respective owners.
  */
 
+#define DEBUG
 #include <linux/delay.h>
 #include <linux/slab.h>
 #include <linux/module.h>

I applied the patch. Now my kernel (3.3.6) log says:

[    6.734445] psmouse serio4: elantech: unexpected magic knock result

0x3c, 0x03, 0x3c.

I hope this can help.

Nicola

Il 20/05/12 21:20, Jonathan Nieder ha scritto:

Yes, thanks much.

One more question: how do you know this touchpad is an Elantech pad? E.g., is there another OS with a driver that detects it as Elantech?

Based on [1] it seems the Amilo Pi 2515 ships with a variety of different touchpad models.

Jonathan

[1] https://bugzilla.kernel.org/show_bug.cgi?id=10942

Hi Jonathan and Nicola,

           ^^^^^^^^^^^^^^^^^

Yes, from this, the touchpad is probably Elan.

But it is meant to be used as such, a basic pointing device. Some NB vendors may not require the multi-finger functions, for cost reason or others. Elan touchpad reports this magic knock when it only supports the basic mouse functions. With these touchpads, the multi-finger functons are NOT TESTED and thus NOT SUPPORTED.

We used to have a "force_elantech" module param to force the TP to be detected as Elan, but I removed it when I was updating elantech.c to support newer models. Maybe you can convince Dmitry to add it back, but still, this kind of Elan touchpads is not supported as multi-touch pointing device, so I wouldn't recommand it.

I hope this helps.

Best regards,
-JJ

Neat.

Definitely; thanks for explaining. Nicola, I'm still curious about the following:

If another OS offers more functionality, then we would know the hardware is capable of it.

The patch JJ described follows, in case you want to experiment.

-- >8 --
Subject: Input: elantech - allow forcing Elantech protocol when knock fails

This does not quite revert 4af61e90270e35bafa9e99a3c48fb3f363da592d ("Input: elantech - remove module parameter force_elantech") --- the old force_elantech parameter was about ignoring an unrecognized response to ETP_FW_VERSION_QUERY, while this one is about ignoring unrecognized responses to the magic knock.

This patch is just for experimentation and is not meant for inclusion.

---
 drivers/input/mouse/elantech.c |   16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

index d2c0db159b18..55ac1aa5e943 100644
@@ -27,6 +27,10 @@
 					fmt, ##__VA_ARGS__);		\
 	} while (0)
 
+static bool force_elantech;
+module_param_named(force_elantech, force_elantech, bool, 0644);
+MODULE_PARM_DESC(force_elantech, "Force the Elantech PS/2 protocol extension to be used, 1 = enabled, 0 = disabled (default).");
+
 /*
  * Send a Synaptics style sliced query command
  */

@@ -1195,10 +1199,14 @@ int elantech_detect(struct psmouse *psmouse, bool set_properties)
 	 */
 	if (param[0] != 0x3c || param[1] != 0x03 ||
 	    (param[2] != 0xc8 && param[2] != 0x00)) {
-		psmouse_dbg(psmouse,
-			    "unexpected magic knock result 0x%02x, 0x%02x, 0x%02x.\n",
-			    param[0], param[1], param[2]);
-		return -1;
+		if (!force_elantech) {
+			psmouse_dbg(psmouse,
+				    "unexpected magic knock result 0x%02x, 0x%02x, 0x%02x.\n",
+				    param[0], param[1], param[2]);
+			return -1;
+		}
+		psmouse_warning(psmouse, "Unexpected magic knock result.\n");
+		psmouse_info(psmouse, "Enabling anyway due to force_elantech.\n");
 	}
 
 	/*

1.7.10.2

Hi,

I'd like to thank you all for the time you dedicated to this problem. I'll try to experiment with the patch and I'll report if something interesting happens.

@Jonathan: there are two things that make me think to have a elantech touchpad. First, a sticker from the manufacturer below laptop's battery pack (in the slot, you can see it only if you remove the battery) labeled "ELANTEK". Second, I think more important, I've played around (about two months ago) with dkms and patches to psmouse. More in detail, I tried the patch from Seth Forshee (see
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/681904/comments/111).
Through it didn't fix the issue (the touchpad was totally unusable due to continuous "sync lost" messages) the touchpad was recognized as an elantech (I don't remember the details, but I could try to build the module again and mail some info).

Thanks again,

Nicola

Il 21/05/12 10:28, Jonathan Nieder ha scritto: