Broken pipe

[Anonymous]

Had KonePure for over a year and did work well with built rpms of roccat-tools 2.1.0. With later versions I needed to replug the mouse every time I booted for it to work (otherwise stuck mid-scren).

After a fresh install of my OP I've been trying to get it to work with compiled versions of 3.7.1. After initial install, logout/login and replug all works well in that the mouse works, I get a working GUI issuing roccatkonepureconfig as root (but not as user) and can change the colour, key functions etc. Reboot the computer and after login the mouse is dead, there's no light and it doesn't even function generically. A broken pipe seems to be the problem.

The discussion at http://sourceforge.net/p/roccat/discussion/989581/thread/0bb1219f/ seems to be rather old and, sorry if I'm wrong, but I'd venture to suggest there's nothing there to provide a solution for me.

Fedora 22, Kernel 4.2.3-200.fc22.x86_64

Terminal outputs:

[root@localhost ****]# export G_MESSAGES_DEBUG=all
[root@localhost ****]# roccateventhandler
(roccateventhandler:4090): GamingGear-DEBUG: Uhid GaminggearUhidKeyboard has file descriptor 5
(roccateventhandler:4090): GamingGear-DEBUG: Uhid GaminggearUhidMouse has file descriptor 6
(roccateventhandler:4090): GamingGear-DEBUG: Uhid GaminggearUhidMultimedia has file descriptor 7
(roccateventhandler:4090): Roccat-DEBUG: Roccat Device: 0x1e7d:0x2dbe (Kone Pure)
(roccateventhandler:4090): Roccat-DEBUG: Identifier: 1-9.4
(roccateventhandler:4090): Roccat-DEBUG: Syspath: (null)
(roccateventhandler:4090): Roccat-DEBUG: Roccat: (null)
(roccateventhandler:4090): Roccat-DEBUG: Interfaces: 2
(roccateventhandler:4090): Roccat-DEBUG: Hidraw[0]: /dev/hidraw2
(roccateventhandler:4090): Roccat-DEBUG: Hidraw[1]: /dev/hidraw3
(roccateventhandler:4090): GamingGear-DEBUG: File /dev/hidraw2 for device key hidraw0 has file descriptor 16

(roccateventhandler:4090): KonePureEventhandler-WARNING **: Could not read actual profile: Could not issue ioctl HIDIOCGFEATURE: Broken pipe

[root@localhost ****]# roccatkonepureconfig
(roccatkonepureconfig:4203): Roccat-DEBUG: Roccat Device: 0x1e7d:0x2dbe (Kone Pure)
(roccatkonepureconfig:4203): Roccat-DEBUG: Identifier: 1-9.4
(roccatkonepureconfig:4203): Roccat-DEBUG: Syspath: (null)
(roccatkonepureconfig:4203): Roccat-DEBUG: Roccat: (null)
(roccatkonepureconfig:4203): Roccat-DEBUG: Interfaces: 2
(roccatkonepureconfig:4203): Roccat-DEBUG: Hidraw[0]: /dev/hidraw2
(roccatkonepureconfig:4203): Roccat-DEBUG: Hidraw[1]: /dev/hidraw3
(roccatkonepureconfig:4203): GamingGear-DEBUG: File /dev/hidraw2 for device key hidraw0 has file descriptor 8

[Stefan Achatz]

First: Don't ever start the tools as root. Create a user roccat and add yourself to it as described in the install instructions. Best is to remove the files under /var/lib/roccat and rebuild the folder as described.
Regarding your problem, my KonePure is working flawlessly on Fedora22 with 4.2.3 kernel.
Maybe setting the device to factory defaults helps?

roccatkonepurecontrol --reset

[Anonymous]

I added the user roccat and did everything exactly as in the instructions. I removed everything including the /var/lib/roccat folder with Krusader running as root (except of course the kernel modules) and then did a repeat install with the same result.

I've now removed the latest editions of libgaminggear and libgaminggear-devel, cleared out the latest editions of roccat tools as above, installed libgaminggear and libgaminggear-devel 0.4.0-1 with the rpms from Matthias Summer's repository and built the rpms using rpmbuild from roccat tools 2.1.0:

$ rpmbuild -ta --clean roccat-tools-2.7.1.tar.bz2

Everything seems to work fine with the installation from these rpms (even with just user privileges). A 1+ year old mouse needs the 1+ year old software?

[Stefan Achatz]

A 1+ year old mouse needs the 1+ year old software?

No. As I said, the combination KonePure, roccat-tools-3.7.1, Fedora 22 with kernel 4.2.3 works like a charm for me. But as long as I can't reproduce your issue I can't do something against it.

[Anonymous]

I just uninstalled the older files again, added Matthias Summer's repository to yum and installed that way. Pretty much the same result as in my first post. Removed the repository, the files thus installed and re-installed the older rpms as in my second post – mouse controls working perfectly again.

Are you using Gnome? I tried Gnome Classic rather than Gnome Wayland, but that didn't seem to alter the situation.

[Stefan Achatz]

It's a shot in the dark, but maybe you wanna try this patch with 3.7.1:

diff --git a/konepure/libkonepureeventhandler/konepure_eventhandler.c b/konepure/libkonepureeventhandler/konepure_eventhandler.c
index 4bdefbf..120ed8b 100644
--- a/konepure/libkonepureeventhandler/konepure_eventhandler.c
+++ b/konepure/libkonepureeventhandler/konepure_eventhandler.c
@@ -463,7 +463,6 @@ static void device_add_cb(RoccatDeviceScannerInterface *interface, RoccatDevice
        KonepureEventhandler *eventhandler = KONEPURE_EVENTHANDLER(user_data);
        KonepureEventhandlerPrivate *priv = eventhandler->priv;
        GError *local_error = NULL;
-       KonepureProfileSettings *profile_settings;
        guint driver_state;

        if (priv->device) {
@@ -497,30 +496,20 @@ static void device_add_cb(RoccatDeviceScannerInterface *interface, RoccatDevice
        if (!priv->timer_note)
                priv->timer_note = roccat_notification_timer_new(roccat_eventhandler_host_get_notificator(priv->host), KONEPURE_DEVICE_NAME);

-       priv->actual_profile_index = koneplus_actual_profile_read(priv->device, &local_error);
-       if (local_error) {
-               g_warning(_("Could not read actual profile: %s"), local_error->message);
-               g_clear_error(&local_error);
-               priv->actual_profile_index = 0; /* guessing */
-       } else {
-               profile_settings = konepure_profile_settings_read(priv->device, priv->actual_profile_index, &local_error);
-               if (local_error) {
-                       g_warning(_("Could not read hardware profile data: %s"), local_error->message);
-                       g_clear_error(&local_error);
-                       priv->actual_sensitivity_x = (ROCCAT_SENSITIVITY_MIN + ROCCAT_SENSITIVITY_MAX) >> 1; /* guessing */
-               }
-               else {
-                       priv->actual_sensitivity_x = profile_settings->sensitivity_x;
-                       g_free(profile_settings);
-               }
-       }
-       
        profile_data_load(eventhandler, &local_error);
        if (local_error) {
                g_warning(_("Could not read rmp: %s"), local_error->message);
                g_clear_error(&local_error);
        }

+       priv->actual_profile_index = koneplus_actual_profile_read(priv->device, &local_error);
+       if (local_error) {
+               g_warning(_("Could not read actual profile: %s"), local_error->message);
+               g_clear_error(&local_error);
+               priv->actual_sensitivity_x = (ROCCAT_SENSITIVITY_MIN + ROCCAT_SENSITIVITY_MAX) >> 1;
+       } else
+               priv->actual_sensitivity_x = koneplus_rmp_get_sensitivity_x(priv->rmp[priv->actual_profile_index]);
+
        priv->gfx = koneplus_gfx_new(priv->device);

        // FIXME create in init and just disconnect/reconnect?

If this doesn't help you could try to delete /etc/Xdg/autostart/roccateventhandler.desktop and check if, after reboot or X-restart, the mouse still hangs after login. Maybe roccateventhandler isn't itself the culprit.

[Anonymous]

OK, thanks. The patch seems to have fixed it. Used Yum Extender to remove the old installation from Matthias Summer's repository and removed var/lib/roccat. Compiled and installed libgaminggear-0.11.2 from SourceForge. Downloaded roccat-tools-3.7.1.tar.bz2 from SourceForge, extracted it and made the patch edits to /konepure/libkonepureeventhandler/konepure_eventhandler.c and then compiled and installed as per INSTALL instructions. Logged out and back in and re-plugged mouse and it worked. Rebooted the computer and (as I write!) it's still working as user clicking on desktop icons to bring up GUI.

[Stefan Achatz]

Well, it's crazy my assumption was right. So it seems to be a quirk of the firmware triggered by your machine maybe being faster than my 8 year old nag. I'll implement this fix for all devices possibly affected for the next release.

[Anonymous]

Yeah, I'm now running an all solid state computer with a 6x3.50GHz i7 processor. Occasionally it misses both the keyboard and the mouse altogether, almost consistently recently if I reboot from Fedora into Ubuntu on my dual boot setup.

The problem with the need to re-plug the mouse with the 2.2.1 Roccat tools and later existed on the 9+ years old 2x2.13GHz Core-2-Duo it has superseded though.