summaryrefslogtreecommitdiff
path: root/src
AgeCommit message (Collapse)Author
2011-06-07Use Data, not Data32 in XIPassiveGrabDevicePeter Hutterer
Data32 takes and iterates over an array of longs, thus skipping every 4 bytes on LP64. Here we only have arrays of ints, use the normal Data macro instead. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Reviewed-by: Jeremy Huddleston <jeremyhu@apple.com> (cherry picked from commit 9faab2bc0bdd4d98a04e572a7a5201bfcd3bdc70)
2011-06-07XIChangeHierarchy: Return Success early if no actual changes are requested.Peter Hutterer
Do the same for negative num_changes. Found by static analyzer. Reported-by: Jeremy Huddleston <jeremyhu@apple.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Reviewed-by: Jeremy Huddleston <jeremyhu@apple.com> (cherry picked from commit cc6feecae23b321952921cf927bc965386844c8c)
2011-06-07Allocate enough memory for raw events + extra data.Peter Hutterer
Necessary space was calculated, but not actually used to allocate memory. As a result, valuator data would overwrite the allocated memory. ==4166== Invalid write of size 1 ==4166== at 0x4C29F04: memcpy (mc_replace_strmem.c:497) ==4166== by 0x8F39180: ??? (in /usr/lib/libXi.so.6.1.0) ==4166== by 0x7433D48: _XCopyEventCookie (in /usr/lib/libX11.so.6.3.0) ==4166== by 0x7425166: XPeekEvent (in /usr/lib/libX11.so.6.3.0) ==4166== by 0x49C3E3: process_key (x11_be.c:1065) ==4166== by 0x49EA5C: event_key_release (x11_be.c:2201) ==4166== by 0x49DD6E: x11_be_process_events (x11_be.c:1892) ==4166== by 0x4A38F4: x11_be_main_loop (x11_be.c:4353) ==4166== by 0x4A39E1: x11_be_thread_main (x11_be.c:4385) ==4166== by 0x87549C9: start_thread (pthread_create.c:300) ==4166== by 0x8A516FC: clone (clone.S:112) ==4166== Address 0x168afe80 is 0 bytes after a block of size 96 alloc'd ==4166== at 0x4C284A8: malloc (vg_replace_malloc.c:236) ==4166== by 0x8F390BD: ??? (in /usr/lib/libXi.so.6.1.0) ==4166== by 0x7433D48: _XCopyEventCookie (in /usr/lib/libX11.so.6.3.0) ==4166== by 0x7425166: XPeekEvent (in /usr/lib/libX11.so.6.3.0) ==4166== by 0x49C3E3: process_key (x11_be.c:1065) ==4166== by 0x49EA5C: event_key_release (x11_be.c:2201) ==4166== by 0x49DD6E: x11_be_process_events (x11_be.c:1892) ==4166== by 0x4A38F4: x11_be_main_loop (x11_be.c:4353) ==4166== by 0x4A39E1: x11_be_thread_main (x11_be.c:4385) ==4166== by 0x87549C9: start_thread (pthread_create.c:300) Reported-by: Roger Cruz <roger.cruz@virtualcomputer.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Reviewed-by: Jeremy Huddleston <jeremyhu@apple.com> Reviewed-by: Daniel Stone <daniel@fooishbar.org> (cherry picked from commit af65475b1f6b7209750220a74aaad9968d54aaf7)
2011-06-07Fix XISelectEvents on 64 bits, strict alignement architectures.Matthieu Herrb
Use Data() to send the struct xXIEventMask on the wire instead of Data32() which expects a pointer to a 64bits value on LP64 architectures. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> (cherry picked from commit 18177afd4fb3934d0a9083c599fb96701eec2ad9)
2011-03-17Force alignment with sizeof(Atom) for XIButtonClassInfoPeter Hutterer
The memory layout of an XIButtonClassInfo is [struct XIButtonClassInfo][mask][labels] With the mask being currently 4-byte aligned and labels a list of Atoms. On LP64, Atoms are 8 byte, leading to unaligned access for some mask lengths. Force the alignment to be sizeof(Atom). Reported-by: Christian Weisgerber <naddy@mips.inka.de> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Tested-by: Christian Weisgerber <naddy@mips.inka.de> Reviewed-by: Adam Jackson <ajax@redhat.com>
2011-03-08Don't discard extra data for passive grabs.Peter Hutterer
Failed modifier data was discarded, causing assertions inside xcb. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2011-03-03Rename XSetCPtr.c to XISetCPtr.c.Peter Hutterer
XISetClientPointer is an XI2.0 call and should be named accordingly. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Reviewed-by: Daniel Stone <daniel@fooishbar.org>
2011-03-03Require XI 2 for all XI 2 calls.Peter Hutterer
Each XI2 call should check _XiCheckExtInit for XI version 2.0 or higher. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Reviewed-by: Daniel Stone <daniel@fooishbar.org>
2011-02-23Fix typo in comment.Peter Hutterer
Reported-by: Julien Cristau <jcristau@debian.org> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2011-02-23Fix invalid read in XIGrabDevice.Peter Hutterer
Miscalculation of length caused Data() to memcpy too many bytes. ==2865== Invalid read of size 1 ==2865== at 0x4A07480: memcpy (mc_replace_strmem.c:602) ==2865== by 0x544271E: XIGrabDevice (XIGrabDevice.c:69) ==2865== by 0x400B0A: main (gnome642481.c:56) ==2865== Address 0x642f614 is 0 bytes after a block of size 20 alloc'd ==2865== at 0x4A04896: calloc (vg_replace_malloc.c:418) ==2865== by 0x54425D3: XIGrabDevice (XIGrabDevice.c:65) ==2865== by 0x400B0A: main (gnome642481.c:56) SetReqLen() expects 4-byte units. Data() expects bytes. Gnome Bug 642481 <https://bugzilla.gnome.org/show_bug.cgi?id=642481> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2011-02-23Mark sourceid in Raw events as bug and force to 0..Peter Hutterer
The protocol does not provide a source ID for raw events, so this value is always 0. It shouldn't really be there, but the past is so hard to change. Reported-by: Mark Dokter <dokter@icg.tugraz.at> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2011-01-26Fill in mods/group->effective in XIQueryPointer()Carlos Garnacho
the other XIModifierState/XIGroupState fields are being set correctly, but the "effective" field was being left as undefined memory. Signed-off-by: Carlos Garnacho <carlosg@gnome.org> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2011-01-11Fix passive grabs.Philipp Reh
_XIPassiveGrabDevice, which is called by alle the passive grab functions, wrongly returns an error when it shouldn't. The attached patch adds the missing "not" to properly test the error condition of _XReply. Signed-off-by: Philipp Reh <sefi@s-e-f-i.de> Reviewed-by: Daniel Stone <daniel@fooishbar.org> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2010-12-14WireToEvent: Set display member of all events as wellDaniel Stone
All events were getting random uninitialised garbage for display; fix that. Signed-off-by: Daniel Stone <daniel@fooishbar.org>
2010-10-06Purge cvs tags.Jesse Adkins
Signed-off-by: Jesse Adkins <jesserayadkins@gmail.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
2010-10-07Fix typo when converting raw events from the wire.Carlos Garnacho
The raw values were being miscalculated, containing only the integral part of the FP3232, meanwhile normal valuators were mistakenly added the fractional part of its corresponding raw value. Signed-off-by: Carlos Garnacho <carlosg@gnome.org> Reviewed-by: Jeremy Huddleston <jeremyhu@apple.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2010-08-26Fix build with gcc 2.95Matthieu Herrb
Signed-off-by: Matthieu Herrb <matthieu.herrb@laas.fr> Reviewed-by: Jeremy Huddleston <jeremyhu@apple.com> Reviewed-by: Matt Turner <mattst88@gmail.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2010-07-13Always unlock display correctlyPauli Nieminen
XISelectEvents and XIGetSelectedEvents were not unlocking display in all return paths. Reported-by: Julien Cristau <jcristau@debian.org> Signed-off-by: Pauli Nieminen <ext-pauli.nieminen@nokia.com>
2010-07-13Use single error path in XGetDeviceControlPauli Nieminen
This reduces code duplication and fixes possible leak of d. d would leak if allocation of Device fails. Signed-off-by: Pauli Nieminen <ext-pauli.nieminen@nokia.com> Reviewed-by: Julien Cristau <jcristau@debian.org>
2010-07-13Use single error path in XGetFeedbackControlPauli Nieminen
This reduces code duplication and fixes possible leak of f. f would leak if allocation of Feedback fails. Signed-off-by: Pauli Nieminen <ext-pauli.nieminen@nokia.com> Reviewed-by: Julien Cristau <jcristau@debian.org>
2010-07-13Use single error path in XQueryDeviceStatePauli Nieminen
This reduces code duplication and fixes possible leak of data. data would leak if allocation of state fails. Signed-off-by: Pauli Nieminen <ext-pauli.nieminen@nokia.com> Reviewed-by: Julien Cristau <jcristau@debian.org>
2010-07-13Fix memory leak in XIGetSelectedEventsPauli Nieminen
mask_in was leaking for every successfull XIGetSelectedEvents. Signed-off-by: Pauli Nieminen <ext-pauli.nieminen@nokia.com> Reviewed-by: Julien Cristau <jcristau@debian.org>
2010-07-13Fix usage of uninitialized valuePauli Nieminen
In error case length of extra data could be uninitialized. This would result randomly sized request later in function. Signed-off-by: Pauli Nieminen <ext-pauli.nieminen@nokia.com> Reviewed-by: Julien Cristau <jcristau@debian.org>
2010-07-08Purge macros NEED_EVENTS and NEED_REPLIESFernando Carrijo
Signed-off-by: Fernando Carrijo <fcarrijo@yahoo.com.br> Acked-by: Tiago Vignatti <tiago.vignatti@nokia.com> Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com>
2010-04-09Don't unlock the Display twice.Jamey Sharp
_xiQueryVersion's caller must wrap it in a LockDisplay/UnlockDisplay pair, so it shouldn't call UnlockDisplay itself. Signed-off-by: Jamey Sharp <jamey@minilop.net>
2010-02-22Initialize extension with the right number of events.Peter Hutterer
If the server supports a lower XI version than the client, the Xlib-internal event vector may be smashed. See libXext for more details. http://cgit.freedesktop.org/xorg/lib/libXext/commit/?id=83fdb27df4ddc2fb088ddf2ec65f0db6b7c57287 This patch queries the server for the supported XI extension before registering the extension with Xlib. The number of events registered depends on the server version. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Reviewed-by: Julien Cristau <jcristau@debian.org>
2010-02-20config: move CWARNFLAGS from configure.ac to Makefile.amGaetan Nadon
Compiler warning flags should be explicitly set in the makefile rather than being merged with other packages compiler flags. Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
2009-09-29Don't export sizeDeviceClassType, size_classes and copy_classesJulien Cristau
Signed-off-by: Julien Cristau <jcristau@debian.org>
2009-09-24XQueryDeviceState: correct length of XValuatorStatePeter Hutterer
The length is suppposed to be the complete class size, not just the struct size. Since XValuatorState is followed by multiple valuator values the size of the class is variable and dependent on the number of valuators. The server assembles the valuator state last. This bug is unlikely to affect clients as they should never go past the last class anyway. Tested-by: Thomas Jaeger Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-09-07mask_len in XIGrabDevice is in 4-byte units (#23708)Peter Hutterer
X.Org Bug 23708 <http://bugs.freedesktop.org/show_bug.cgi?id=23708> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-09-01Initialize send_event for cookie events. (#23609)Peter Hutterer
X.Org Bug 23609 <http://bugs.freedesktop.org/show_bug.cgi?id=23609>
2009-08-24Update XIWarpPointer to take doubles (inputproto >= 1.9.99.902)Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-08-05Update library version for new symbols.Peter Hutterer
XI2 adds a whole lot of new symbols but leaves the existing ones. Minor bump only. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-08-04Remove leftover RCS tags.Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-22Replace keysym grabs with keycode grabs.Peter Hutterer
Keysym grabs are tricky to get right for applications that are more complicated than demo applications. otoh, we know keycode grabs are working. So let's go with keycode grabs for now and add keysym grabs later when we've sorted out the details. Requires inputproto 1.9.99.15 Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-15Fix wrong raw_valuator wire copy, add fractional parts.Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-13Add flags to XIDeviceEvent and XIRawEvent.Daniel Stone
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-13Cater for new event-specific raw event types.Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-13Fix broken formatting, remove empty comment.Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-13Add the deviceid to XI2 property events.Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-13Plug memory leak in XIHierarchyEvent copy.Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-13Fix segfault in XIDeviceChangedEvent wire parsing.Peter Hutterer
Introduced by 225071e2e67fb65a0258397212f9826c9b25e078. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-12Add event copying for raw events.Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-12Switch to new XIEvent structures - no need for pointers anymore.Peter Hutterer
This removes all those pointers from event structures that were just there due to the pre-cookie struct size limit. Pointers remaining are only those that are of variable length (e.g. masks and valuator states). Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-12Remove XIFreeEventData - obsolete with cookie events.Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-12Add effective modifiers/group to events - require inputproto 1.9.99.13Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-08Split copy_classes and size_classes to avoid memory leakPeter Hutterer
The switch to cookie events introduced a memory leak in XIDeviceChangedEvents. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-08Use cookie events for XI2 events.Peter Hutterer
Instead of squashing the events directly into the XEvent use the new libX11 cookie events. This allows us to have larger events without the 96 byte restriction. Requires libX11 1.2.99.1 Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-07-08Add next_block() call to advance pointers over memory.Peter Hutterer
Simple call to get the pointer to the next field when operating on a wire protocol struct or on a single-memory-block Xlib structure. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2009-06-17Axis and button labels should be called labels, not names.Peter Hutterer
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-22 08:15:19 +0000