uinput: don't try to set the syspath twice

If we have it, stop searching for it. Otherwise a second device with the
same name would overwrite the first, causing a leak.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Return -EBADF for functions that need the fd initialized

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Return EBADF when trying to read from an uninitalized device

All other functions that check the fd for validity return EBADF, which also makes it
easier to debug if the actual device goes away.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Use ENOMEM instead of ENOSPC

From errno(3):
   ENOMEM          Not enough space (POSIX.1)
   ENOSPC          No space left on device (POSIX.1)

when we run out memory the reason is a failed malloc, for which ENOMEM
seems more appropriate.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

test: update to test for the various _MAX values

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

memcpy instead of invidual bittoggle

The ioctls return the number of bytes copied into the destination, so just
copy them into the device state instead of individually flipping bits.

For easier review: rc is the return value of the EVIOCG* ioctl, which is
the number of bytes copied.

state variables must be initialized to 0 now, in case the kernel's FOO_MAX
is smaller than libevdev's FOO_MAX. If not initialized to 0, the bytes
between the two max values is undefined and we may end up generating bogus
events.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

test: add a test for uinput device properties

Specifically, test for INPUT_PROP_MAX, which is a valid property value

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Fix off-by-one errors when dealing with *_MAX values.

LED_MAX, KEY_MAX, ABS_MT_MAX, etc. are all valid event codes

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Use AC_USE_SYSTEM_EXTENSIONS

Defines _GNU_SOURCE for us.
http://www.gnu.org/software/autoconf/manual/autoconf-2.67/html_node/Posix-Variants.html

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: fix ev_rep test for new uinput implementation

We can actually set EV_REP values now, though with limitations

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

test: switch udev backend over to new libevdev-uinput bits

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: add uinput creation tests

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Add support for uinput device creation

This lets libevdev provide a relatively generic interface for the
creation of uinput devices so we don't need to duplicate this across
multiple projects.

Most of this is lifted from the current test implementation, with a
couple of minor changes.

EV_REP needs special handling:
   Kernel allows to set the EV_REP bit, it doesn't set REP_* bits (which we
   wrap anyway) but it will also set the default values (500, 33).

Device node is guessed based on the sysfs path:
   The sysfs path contains a eventN file, that corresponds to our
   /dev/input/eventN number. Use it so clients can quickly get the device
   node, without a libudev dependency.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Fix two signed vs unsigned int warnings

dev->num_slots is -1 if we don't have ABS_MT_SLOT.

Set dev->grabbed to the right field type.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Actually use the gcc CFLAGS set in configure

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Mark all external symbols with LIBEVDEV_EXPORT

GCC_CFLAGS was set by configure, but never actually used. So we didn't build with hidden symbols,
rather just weeded them out later through libtool.

Do this properly now, mark the symbols as visibility default.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Fix some compiler warnings about maybe uninitialized values

These are just to shut the compiler up, in all three cases we only access
the respective values if they're defined.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

configure.ac: Drop a superfluous variable

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add C++ header guards

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: add test for libevdev_is_event_code(EV_SYN...)

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Simplify getter/setter macros

Now that we've dropped the deprecated API, we can simplify the macros a bit.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Drop deprecated API

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Fix a few signed/unsigned int comparisons

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Add a note that libevdev_is_event_type/code have compile-time ranges

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

libevdev_is_event_type() needs to check for < EV_CNT

EV_MAX is a valid (though unused) value.

Reported-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Rewrite libevdev_is_event_code to avoid signed/unsigned comparison

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Use libevdev_get_event_type_max from libevdev_get_event_code_name

This will check for invalid types for us, and we can do a better check
for signed/unsigned comparison.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Axis values must be int, not unsigned int

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Ignore -Woverride-init in event-names.h

Having the declarations in the form of [0...EV_MAX] = NULL together
with the actual definitions causes warnings for every true definition if
-Woverride-init is enabled.

We can drop them for most as they're zero, but still need them for
ev_max (which defaults to -1), not zero. So use the GCC pragma to
disable the warnings for this file.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Change event name generate script to be python 2/3 compatible

Python 3 doesn't have a print statement, only a print function.

Fixed with:
  2to3 make-event-names.py | git apply

Print as function requires Python 2.6 which is reasonable enough given
that even RHEL6 ships that.

Even though it's not needed for 2.6, use
from __future__ import print_function
to avoid accidentally introducing a print statement in the future.
With this line, print "blah" is now a syntax error in python 2.
This line was added manually, after the 2to3 conversion.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

doc: actually package generated documentation

Previous find command wouldn't actually list the generated documentation
files, so they'd be missing in the tarball.

...and add a dist-hook to make sure this won't happen during release.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Add support for EV_SW

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Avoid under-allocation of array for syncing key/led state

No functional effect as the size would end up being the same anyway
due to the values of KEY_MAX and LED_MAX.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Remove unneeded include of linux/uinput.h

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Merge branch 'led-handling'

Rename enums to match lower_case format

This is technically an API, but not an ABI change.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>

Store the abs value after handling mt events

This way any ABS_MT_ event value that comes in will also be stored in abs_info.
That always corresponds to "current slot", so if a user calls
libevdev_set_event_value() or libevdev_get_event_value() they're actually
modifying the current slot value.

When the current slot changes, sync the state back into the absinfo values.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

write EV_SYN input_event along with led events to device

Other clients of an evdev device need to have the events they receive
be separated, in moment in time, from other events by an EV_SYN/
SYN_REPORT. This is the responsibility of the client who writes events
into the stream.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Fix build without MSC_TIMESTAMP

To build with older input.h

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Pass --no-print-directory to make

The "Entering/Leaving directroy /xyz" messages only add noise to
make-output. Even without we still get short hints which directory is
currently processed. Compare the two outputs below:

With --no-print-directory added:
  make --no-print-directory all-recursive
  Making all in doc
  doxygen libevdev.doxygen
  Making all in libevdev
  python2 ./make-event-names.py --output=c > event-names.h
  make  all-am
    CC       libevdev.lo
    CCLD     libevdev.la
  Making all in tools
    CC       libevdev-events.o
    CCLD     libevdev-events
  Making all in test
    CC       test-main.o
    CC       test-event-names.o
    CC       test-libevdev-init.o
    CC       test-libevdev-has-event.o
    CC       test-int-queue.o
    CC       test-libevdev-events.o
    CC       libevdev.o
    CC       test-common-uinput.o
    CC       test-common.o
    CCLD     test-libevdev

Without it:
  make  all-recursive
  make[1]: Entering directory `/home/david/dev/libevdev'
  Making all in doc
  make[2]: Entering directory `/home/david/dev/libevdev/doc'
  doxygen libevdev.doxygen
  make[2]: Leaving directory `/home/david/dev/libevdev/doc'
  Making all in libevdev
  make[2]: Entering directory `/home/david/dev/libevdev/libevdev'
  python2 ./make-event-names.py --output=c > event-names.h
  make  all-am
  make[3]: Entering directory `/home/david/dev/libevdev/libevdev'
    CC       libevdev.lo
    CCLD     libevdev.la
  make[3]: Leaving directory `/home/david/dev/libevdev/libevdev'
  make[2]: Leaving directory `/home/david/dev/libevdev/libevdev'
  Making all in tools
  make[2]: Entering directory `/home/david/dev/libevdev/tools'
    CC       libevdev-events.o
    CCLD     libevdev-events
  make[2]: Leaving directory `/home/david/dev/libevdev/tools'
  Making all in test
  make[2]: Entering directory `/home/david/dev/libevdev/test'
    CC       test-main.o
    CC       test-event-names.o
    CC       test-libevdev-init.o
    CC       test-libevdev-has-event.o
    CC       test-int-queue.o
    CC       test-libevdev-events.o
    CC       libevdev.o
    CC       test-common-uinput.o
    CC       test-common.o
    CCLD     test-libevdev
  make[2]: Leaving directory `/home/david/dev/libevdev/test'
  make[2]: Entering directory `/home/david/dev/libevdev'
  make[2]: Leaving directory `/home/david/dev/libevdev'
  make[1]: Leaving directory `/home/david/dev/libevdev'

We don't really win any useful information from these messages.

Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Remove deprecated SYMBOL_CACHE_SIZE from doxygen.in

SYMBOL_CACHE_SIZE was removed in newer doxygen releases. We use the
default value, so simply drop it.

Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add -I m4 to ACLOCAL_AMFLAGS

We put m4 files into a separate directory. If we ever put our own files in
there, we need to include it. Furthermore, this suppresses annoying
autotools warnings.

Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add functions to toggle LEDs on the device

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Keep the LED state and sync it after SYN_DROPPED

This enables libevdev_get_event_value(dev, EV_LED, LED_NUML); to check
if a LED is on or off.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Add setters for an event or slot value

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Rename libevdev_kernel_set_abs_value to libevdev_kernel_set_abs_info

Keep it in line with the non-kernel setter

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Sync the key state on startup

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Simplify a bit state setter

set_bit_state does exactly what the previous if/else did.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Minor documentation improvement

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: devices with abs axes need to be created properly

Simply enabling the bits is not enough, we need to provide axis
information too if we want to enable this properly.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Make the tests optional to drop unneeded dependencies

libevdev has no external dependencies and both check and libudev are
only required for running the unit-tests. Make them optional.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

test: change device for enabling/disabling bits to use REL_*

No real effect on the test, this check is just so that a follow-up patch to
fix the tests for the new uinput backend is a bit easier to review.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Fix stray * in documentation

libevdev 0.3

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Fix header guard, lowercased in some previous search/replace

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add setters for product/vendor/bustype/version

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Rename ID getters to have a consistent naming scheme

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Replace ID getter implementation with a define

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Provide setters for name, phys, uniq

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Add abs setters for changing a single value on an abs axis

Changing a single value on an abs axis is slightly more common than
having to enable that axis outright. Provide a set of accessors for
doing so.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Rename a few getters for consistency

Deprecated:
* libevdev_get_abs_min, libevdev_get_abs_max
* libevdev_get_input_prop_name

Will be removed in one or two versions.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Replace libevdev_get_abs_* with macro-generated ones

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Add setter for property bits

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

test: check for enabling REL_X with data

Only EV_REP and EV_ABS accept data.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Add libevdev_has_event_pending()

Returns non-zero if there are events avialable to be read.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

doc: fix css for notes and return values

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

When enabling EV_REP, set the delay/period values

Just enabling EV_REP sets them to zero, but when enabling them directly,
a value is required.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Make sure all EV_REP bits are set

Current code was skipping REP_PERIOD

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Set the size for repeat delays based on REP_CNT

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Don't try setting an ABS code without data

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

Allow enabling of EV_SYN codes

Has no effect, but allow it nonetheless to avoid superfluous
conditions in client code.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

test: drop some superfluous dev creation

copy/paste residue, had no effect

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

gitignore: ignore gcov, ctags, announce emails

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>

libevdev 0.2.1

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Fix name and version in pkgconfig file

evdev_read was an earlier name for this library.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

libevdev 0.2

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

doc: fix two doxygen include paths to be source-relative

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add libevdev.css to EXTRA_DIST

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Point to new API documentation URL

And fix up the tool to publish it.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Document mailing list

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Change github.com references to freedesktop.org

Except for the API documentation, that'll stay on github for now until
I figure out where to put it (and have the space to put it in).

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: add uinput prop bits and the test for it

Turns out I was looking at an old header file, UI_SET_PROPBIT
has existed for quite a while.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Don't include config.h in the public header

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

doc: couple of minor documentation fixes

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

doc: css changes for gnome web

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add a custom stylesheet to the documentation.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add documentation about the test suite

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: define a common device name

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: constify the input_absinfo argument

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: don't set the abs value when initializing an uinput device

The helper function uses abs.value for the axis code (to save a parameter).
Reset that to 0 before we write to the device

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: fix up the tests to test for resolution now too

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: write resolution for uinput devices after creating the device

EVIOCSABS allows us to set the device's abs axes to the desired resolution.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

doc: tweak a few doxygen options, mainly to display the version

The rest is rather insignificant

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Some minor documentation fixes

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: add test for event values (type and code)

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

test: unify test case creation in test-event-names.c

Simple rename from custom test case variable names to tc

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add two helper functions to check event type/code

These aren't really that useful since they're just one or two lines in
code, but it saves callers from accidentally misplacing braces, etc.

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Add a simple example to the documentation

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

Indent some documentation better

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>

README: Drop the details, refer to documentation instead

Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>