similar to: [ANNOUNCE] xf86-input-evdev-2.0.0

Adam Jackson (1): evdev 1.99.3 Peter Hutterer (7): Fail if the device cannot be grabbed during the probe. Check for XINPUT ABI, parameters of InitValuatorClassRec have changed. Revert "Check for XINPUT ABI, parameters of InitValuatorClassRec have changed." Check for XINPUT ABI < 3 (corrected version) Plug a memory leak, we allocated pEvdev twice,

First snapshot for evdev 2.2. Lots of cleanup, and - most notably - general axis/button support. For those running an X server from master, evdev will label axes and buttons for you. shortlog is a bit longer than it actually is, it includes a number of patches that were cherry-picked onto 2.1 and have since been released. Cheers, Peter Shortlog since xf86-input-evdev-2.1.0: Adam Jackson

Adam Jackson (20): Unlibcwrap. Support the new DIX motion history API Ignore button events above BTN_TASK for now. Parse Option "Path" as well as Option "Device". Re-add XKB support. Nuke the keyboard bell code. Add absolute coordinate event support. Force maximum keycode to be 255 Handle the alternate spellings of the xkb

Hi again, Some further digging revealed that NVPlatformProbe is called with the flag PLATFORM_PROBE_GPU_SCREEN, so the flag XF86_ALLOCATE_GPU_SCREEN is passed to xf86AllocateScreen. After xf86AllocateScreen I end up with xf86GPUScreens == 1 and xf86NumScreens == 0. The xf86GPUScreens seems to be ignored by all the following code, only xf86NumScreens is evaluated afterwards in xf86BusConfig:

It's surprising that it's coming up as a platform device and not a pci device. Perhaps xorg was built without pci support? (Not sure if that's even possible.) It should definitely not be coming up as a GPU screen, but as a regular screen. -ilia On Thu, Jul 7, 2016 at 1:15 PM, Clemens Eisserer <linuxhippy at gmail.com> wrote: > Hi again, > > Some further digging

One significant change that affects packagers: the config snippet has been renamed from 90-libinput.conf to 60-libinput.conf. This should not show any immediate effects on most user's machines but there is a slight potential of user configurations to change if their custom snippets now sort after libinput instead of before. This change is part of the two-step solution to fix the handling of

evdev 1.99.3 had a nasty bug that makes the server crash on DEVICE_CLOSE. Here's a fixed version. Peter Hutterer (3): Remove wakeup handlers when device is closed. If Emulate3Buttons is specified in the config, don't auto-deactivate it. evdev 1.99.4 git tag: xf86-input-evdev-1.99.4 http://xorg.freedesktop.org/archive/individual/driver/xf86-input-evdev-1.99.4.tar.bz2

Alan Coopersmith (1): Remove xorgconfig & xorgcfg from See Also list in man page Matt Helsley (2): rename NBITS to NLONGS to reflect its actual meaning Fix FOO_MAX off-by-one Michael Chapman (1): Disable middle mouse button emulation when a middle mouse button event is registered Peter Hutterer (2): Finalize MB emu (and wakeup handlers) before closing the fd.

Expect this to be the last evdev 2.0.x release. Julien Cristau (1): Fix TestBit() on 64bit Peter Hutterer (1): evdev 2.0.8 git tag: xf86-input-evdev-2.0.8 http://xorg.freedesktop.org/archive/individual/driver/xf86-input-evdev-2.0.8.tar.bz2 MD5: 7f9ace5bcca1427b0da40827a74cbc33 xf86-input-evdev-2.0.8.tar.bz2 SHA1: 57a88717a621957d82169b7508ef293214f4b952

Julien Cristau (1): Set pInfo->fd to -1 on DEVICE_CLOSE Peter Hutterer (3): Fix "Device reopened after N attempts" message. Don't post keycodes > 255. evdev 2.0.7 git tag: xf86-input-evdev-2.0.7 http://xorg.freedesktop.org/archive/individual/driver/xf86-input-evdev-2.0.7.tar.bz2 MD5: cda012a05bcbe8fbad46ea338ad1716e xf86-input-evdev-2.0.7.tar.bz2

No blockers reported since the last RC, so let's get it out of the way. The new evdev 2.1 is here, and it's sparkly and shiny and whatnot. Note that this version does not grab the event device anymore. If you're updating from evdev 2.0, ensure you have [1] in the server to avoid events being sent to the TTY. Other patches of interest [2-4] are recommended to avoid duplicate events.

This one merges input-hotplug support. Daniel Stone: remove XFree86LOADER usage prune device list on DEVICE_CLOSE pass number of axes to InitValuatorClassDeviceStruct pass maxval correctly to xf86InitValuatorAxisDeviceStruct fix InitValuatorClassDeviceStruct call key: remove usage of OS keyboard layer for bell brain: call xf86DeleteInput when removing

Two fixes in this release, one to support Android MT devices, one to avoid an array index overflow. Colin B. Macdonald (1): Workaround lack of ABS_X on MT devices (#80470) Peter Hutterer (1): evdev 2.9.2 Tobias Himmer (1): Check for incoming MT slot indices exceeding the allocated number of slots. git tag: xf86-input-evdev-2.9.2

Only one minor fix since RC1. In xorg.conf static configurations, a wrong error code would prevent the server from loading additional devices if a device couldn't be opened. This is fixed now and only happens when we're out of memory (as intended). Looks like there's no major bugs out there, so unless I see a showstopper in before Monday, I'll get the release out next week. Peter

evdev 2.10 is now available. Plenty of bugfixes accumulated over time, one new option is now available: Resolution. This enables a user to set a mouse device's native resolution, if set the device's deltas are scaled to a normalized 1000dpi resolution. This can be used to make high-resolution mice less sensitive without having to actually change the acceleration in the desktop

Only one change over 2.10, affecting some devices with absolute axes but not ABS_X/ABS_Y. Previously, those devices had the first to axes mapped to axes 0 and 1 and thus some random axes was interpreted as x/y coordinate. With this release, axes 0 and 1 are reserved for x/y axes only. Peter Hutterer (2): Only map x and y to axes 0 and 1 evdev 2.10.1 git tag: xf86-input-evdev-2.10.1

This release restores wheel emulation on absolute devices and drops the forced direction locking for scroll buildup during wheel emulation since it made it almost impossible to actually scroll in both directions. Since horizontal scrolling is disabled by default anyway, we don't need this lock. Peter Hutterer (5): man: add a warning that wheel emu inertia must be set and it isn't

Only one fix over 2.10.2 restoring tablets' functionality. A bug in the wheel emulation rework caused non-x/y axes to stop updating. Interaction with tablets thus resulted in a zero pressure sent to the clients an the tablet being unable to draw. Peter Hutterer (2): Restore non-xy axes updates evdev 2.10.3 git tag: xf86-input-evdev-2.10.3

Just the changes to support server 1.19, nothing exciting otherwise. Keith Packard (3): Report errors when attempting to write keyboard controls Support XINPUT ABI version 23 (threaded input) Support XINPUT ABI 24 (remove select mask from block/wakeup handlers) Peter Hutterer (1): evdev 2.10.4 git tag: xf86-input-evdev-2.10.4

Two fixes, one new feature for the evdev driver. The new feature is the ability to change the emulated middle button number (e.g. useful on devices with physical middle button as well). Alexander Volkov (1): Don't send pointer events in addition to touch events David Strobach (1): Middle emulation - make the emulated button number configurable Peter Hutterer (2): Fix