summaryrefslogtreecommitdiff
path: root/src
AgeCommit message (Collapse)Author
2020-10-10Update the list of supported XI versionsPovilas Kanapickas
2020-10-10Add support for XI 2.4 gesture eventsPovilas Kanapickas
2019-06-19Replace open-coded FP3232_TO_DOUBLEMatt Turner
Signed-off-by: Matt Turner <mattst88@gmail.com>
2019-06-17Fix the FIXME in XIValuatorClass case of copy_classes function in XExtInt.cAlexander Bersenev
In addition fix FP3232_TO_DOUBLE macro to correctly compute the fractional part. This fixes glitchy scrolling in Qt applications when the application was just activated or was scrolled in the backgroud. Qt uses XIQueryDevice call to synchronize internal scroll location with an actual one. Bug: https://gitlab.freedesktop.org/xorg/lib/libxi/issues/10
2018-02-06_XIPassiveGrabDevice needs to set time valueJeff Smith
When setting up a XIPassiveGrabDevice request, the time field is not being set, leading to improper data being passed 'over the wire'. Accept a time value into _XIPassiveGrabDevice and use it to set the time field in the request. Since the the functions calling _XIPassiveGrabDevice are part of the API, and they do not accept time values, they can just pass CurrentTime. Signed-off-by: Jeff Smith <whydoubt@gmail.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2016-12-29Fix possible free of uninitialized pointerEmilio Pozuelo Monfort
If the _XReply() call fails, we'll try to free an uninitialized pointer. https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=849026 Reported-by: Thomas Walker <thwalker3@gmail.com> Signed-off-by: Emilio Pozuelo Monfort <pochu@debian.org> Reviewed-by: Julien Cristau <jcristau@debian.org> Tested-by: Thomas Walker <thwalker3@gmail.com> Signed-off-by: Julien Cristau <jcristau@debian.org>
2016-10-26Check that allocating a buffer succeededEmilio Pozuelo Monfort
Since we are going to write into the buffer, we should make sure the allocation didn't fail. Reported-by: Julien Cristau <jcristau@debian.org> Signed-off-by: Emilio Pozuelo Monfort <pochu@debian.org> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2016-10-26Plug a memory leakEmilio Pozuelo Monfort
Introduced in commit 19a9cd6. Reported-by: Julien Cristau <jcristau@debian.org> Signed-off-by: Emilio Pozuelo Monfort <pochu@debian.org> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2016-10-13XListInputDevices: don't touch ndevices in case of errorPeter Hutterer
We used to always set *ndevices to the number of devices returned by the server. This magically worked because we pretty much never returned an error except on faulty server or library implementations. With 19a9cd60 we now have more chances of getting an error, so the polite thing is to just leave *ndevices alone when we error out. Document it as such in the man page, just in case someone accidentally reads it. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> CC: Niels Ole Salscheider <niels_ole@salscheider-online.de> Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
2016-10-13SizeClassInfo can return 0 even without an errorNiels Ole Salscheider
Catch the error case separately. Commit 19a9cd607d added length checking to SizeClassInfo but re-used the return value of 0 for an error. A device without classes (as is initialized by xf86-input-libinput for tablets) can legitimately return 0 and erroneously triggers an error. Fix this by using a separate value for the error. Reproducible by calling XListInputDevices() with a tablet attached. This fixes a regression introduced in commit 19a9cd607d. Signed-off-by: Niels Ole Salscheider <niels_ole@salscheider-online.de> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
2016-09-25Properly validate server responses.Tobias Stoeckmann
By validating length fields from server responses, out of boundary accesses and endless loops can be mitigated. Signed-off-by: Tobias Stoeckmann <tobias@stoeckmann.org> Reviewed-by: Matthieu Herrb <matthieu@herrb.eu>
2015-11-03Fix const compiler warningsJavier Pello
When invoking Data, Data16 and Data32 from XChangeDeviceProperty, we must cast the data pointer to the right type, but we do not need to cast constness away. This change allows to enable -Wcast-qual on the build and have it complete without warnings. Signed-off-by: Javier Pello <javier.pello@urjc.es> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2015-10-19Don't use raw serial numbers in XIEventsPeter Hutterer
cookie->serial is an Xlib contoction, provided by _XSetLastRequestRead(). This serial may be different to the raw serial number from the wire protocol. This causes issues when the raw serial is used to e.g. compare the event to other non-XI events. Use the cookie's serial number instead. https://bugzilla.gnome.org/show_bug.cgi?id=756649 See also https://bugs.freedesktop.org/show_bug.cgi?id=64687 Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2015-02-24Fix version check in _XIAllowEventsCosimo Cecchi
Commit 5810d0797160a97012664ffe719a59e1b288a525 changed _XIAllowEvents() to use _XiCheckVersion() instead of _XiCheckExtInit() to avoid a double display unlock, but it failed to correctly check for the version, since we should set have_XI22 to True for every version greater or equal to 2.2. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-11-04XIGrabDevice: Unlock display in error path.Michal Srb
Signed-off-by: Michal Srb <msrb@suse.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-11-03Refactor XGetExtensionVersion.Michal Srb
_XiGetExtensionVersion was called from XGetExtensionVersion and from _XiCheckExtInit. When called from _XiCheckExtInit, nothing accounted for the fact that it can return ((XExtensionVersion *) NoSuchExtension) in case of error. Also it recursively calls _XiCheckExtInit potentionally causing multiple unlocks if _XiCheckExtInit fails. -> Remove it and call directly _XiGetExtensionVersionRequest and only call _XiCheckExtInit only from XGetExtensionVersion. Signed-off-by: Michal Srb <msrb@suse.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-11-03Fix logic in _XIAllowEvents and prevent double unlock.Michal Srb
Replacing the second _XiCheckExtInit with _XiCheckVersion prevents possible double unlock as _XiCheckExtInit actually unlocks the display when it returns -1. Signed-off-by: Michal Srb <msrb@suse.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-11-03XIGetClientPointer: Return False on error.Michal Srb
Not NoSuchExtension which is 1 = True! Signed-off-by: Michal Srb <msrb@suse.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-11-03Do not return NoSuchExtension casted to pointer as an error.Michal Srb
Several functions were returning NoSuchExtension casted to a pointer in case of an error. Often in parallel with returning NULL in case of another error. It is undocumented and certainly wrong. Signed-off-by: Michal Srb <msrb@suse.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-11-03XIChangeHierarchy: Add missing unlock.Michal Srb
When num_changes <= 0 or Xmalloc fails, the display has to be unlocked. Signed-off-by: Michal Srb <msrb@suse.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-11-03Fix double unlock when _XiCheckExtInit return -1.Michal Srb
_XiCheckExtInit unlocks the display if it fails and returns -1. Most callers account for it properly, but few didn't. Signed-off-by: Michal Srb <msrb@suse.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-10-28Advance the request buffer by the right amount in XIChangeHierarchyJulien Cristau
c->length is in 4-byte units, dptr is a char *, so we need to advance dptr by 4 * length to get the position of the next HierarchyChangeInfo. Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net> Signed-off-by: Julien Cristau <jcristau@debian.org>
2014-07-14Fix locking bugs with XIAllowTouchEvents() and XIUngrabTouchBegin()Owen W. Taylor
Fix two places where the display was double locked when an API function chained to an implementation that also locks the display. Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-07-10XIPassiveGrab: Fix completely broken locking in XIGrabTouchBeginJasper St. Pierre
_XIPassiveGrabDevice calls LockDisplay as the first thing it does. That means that it expects the display to be unlocked. XIGrabTouchBegin locks the display to check for the XI extension, and then never unlocks it. Effectively, this meant that anybody that called XIGrabTouchBegin after XInitThreads just got a deadlock. Cool. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2014-07-10XIPassiveGrab: Fix display locking inside _XIPassiveGrabDevice for error pathsJasper St. Pierre
The code here before would just leave the display locked on error, which is all sorts of broken. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-11-25Remove fallback for _XEatDataWords, require libX11 1.6 for itMichael Joost
_XEatDataWords was orignally introduced with the May 2013 security patches, and in order to ease the process of delivering those, fallback versions of _XEatDataWords were included in the X extension library patches so they could be applied to older versions that didn't have libX11 1.6 yet. Now that we're past that hurdle, we can drop the fallbacks and just require libX11 1.6 for building new versions of the extension libraries. Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-06-28Remove check that can never be true.Thomas Klausner
clang warns: warning: comparison of constant 268435455 with expression of type 'CARD16' (aka 'unsigned short') is always false Signed-off-by: Thomas Klausner <wiz@NetBSD.org> Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-06-27Include limits.h to prevent build error: missing INT_MAXPeter Hutterer
Introduced in 4c8e9bcab459ea5f870d3e56eff15f931807f9b7. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-06-27If the XGetDeviceDontPropagateList reply has an invalid length, return 0Peter Hutterer
If we skip over the reply data, return 0 as number of event classes. Follow-up to 6dd6dc51a2935c72774be81e5cc2ba2c30e9feff. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-06-27Change size += to size = in XGetDeviceControlPeter Hutterer
size += blah is technically correct but it implies that we're looping or otherwise incrementing the size. Which we don't, it's only ever set once. Change this to avoid reviewer confusion. Reported-by: Dave "color-me-confused" Airlie <airlied@redhat.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-06-27Fix potential corruption in mask_len handlingPeter Hutterer
First: check for allocation failure on the mask. XI2 requires that the mask is zeroed, so we can't just Data() the mask provided by the client (it will pad) - we need a tmp buffer. Make sure that doesn't fail. Second: req->mask_len is a uint16_t, so check against malicious mask_lens that would cause us to corrupt memory on copy, as the code always allocates req->mask_len * 4, but copies mask->mask_len bytes. Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-24Don't overwrite the cookies serial numberPeter Hutterer
serial != sequenceNumber, see _XSetLastRequestRead() cookie->serial is already set at this point, setting it again directly from the sequenceNumber of the event causes a bunch of weird issues such as scrollbars and text drag-n-drop breaking. https://bugzilla.redhat.com/show_bug.cgi?id=965347 Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23sign extension issue in XListInputDevices() [CVE-2013-1995]Alan Coopersmith
nptr is (signed) char, which can be negative, and will sign extend when added to the int size, which means size can be subtracted from, leading to allocating too small a buffer to hold the data being copied from the X server's reply. v2: check that string size fits inside the data read from the server, so that we don't read out of bounds either Reported-by: Ilja Van Sprundel <ivansprundel@ioactive.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23Avoid integer overflow in XListInputDevices() [CVE-2013-1984 8/8]Alan Coopersmith
If the length of the reply as reported by the Xserver is too long, it could overflow the calculation for the size of the buffer to copy the reply into, causing memory corruption. Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23Avoid integer overflow in XGetDeviceProperties() [CVE-2013-1984 7/8]Alan Coopersmith
If the number of items as reported by the Xserver is too large, it could overflow the calculation for the size of the buffer to copy the reply into, causing memory corruption. Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23integer overflow in XIGetSelectedEvents() [CVE-2013-1984 6/8]Alan Coopersmith
If the number of events or masks reported by the server is large enough that it overflows when multiplied by the size of the appropriate struct, or the sizes overflow as they are totaled up, then memory corruption can occur when more bytes are copied from the X server reply than the size of the buffer we allocated to hold them. v2: check that reply size fits inside the data read from the server, so that we don't read out of bounds either Reported-by: Ilja Van Sprundel <ivansprundel@ioactive.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23integer overflow in XIGetProperty() [CVE-2013-1984 5/8]Alan Coopersmith
If the number of items reported by the server is large enough that it overflows when multiplied by the size of the appropriate item type, then memory corruption can occur when more bytes are copied from the X server reply than the size of the buffer we allocated to hold them. Reported-by: Ilja Van Sprundel <ivansprundel@ioactive.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23integer overflow in XGetDeviceMotionEvents() [CVE-2013-1984 4/8]Alan Coopersmith
If the number of events or axes reported by the server is large enough that it overflows when multiplied by the size of the appropriate struct, then memory corruption can occur when more bytes are copied from the X server reply than the size of the buffer we allocated to hold them. Reported-by: Ilja Van Sprundel <ivansprundel@ioactive.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23integer overflow in XGetDeviceDontPropagateList() [CVE-2013-1984 3/8]Alan Coopersmith
If the number of event classes reported by the server is large enough that it overflows when multiplied by the size of the appropriate struct, then memory corruption can occur when more bytes are copied from the X server reply than the size of the buffer we allocated to hold them. V2: EatData if count is 0 but length is > 0 to avoid XIOErrors Reported-by: Ilja Van Sprundel <ivansprundel@ioactive.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23integer overflow in XGetFeedbackControl() [CVE-2013-1984 2/8]Alan Coopersmith
If the number of feedbacks reported by the server is large enough that it overflows when multiplied by the size of the appropriate struct, or if the total size of all the feedback structures overflows when added together, then memory corruption can occur when more bytes are copied from the X server reply than the size of the buffer we allocated to hold them. v2: check that reply size fits inside the data read from the server, so we don't read out of bounds either Reported-by: Ilja Van Sprundel <ivansprundel@ioactive.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23integer overflow in XGetDeviceControl() [CVE-2013-1984 1/8]Alan Coopersmith
If the number of valuators reported by the server is large enough that it overflows when multiplied by the size of the appropriate struct, then memory corruption can occur when more bytes are copied from the X server reply than the size of the buffer we allocated to hold them. v2: check that reply size fits inside the data read from the server, so we don't read out of bounds either Reported-by: Ilja Van Sprundel <ivansprundel@ioactive.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23unvalidated lengths in XQueryDeviceState() [CVE-2013-1998 3/3]Alan Coopersmith
If the lengths given for each class state in the reply add up to more than the rep.length, we could read past the end of the buffer allocated to hold the data read from the server. Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23memory corruption in _XIPassiveGrabDevice() [CVE-2013-1998 2/3]Alan Coopersmith
If the server returned more modifiers than the caller asked for, we'd just keep copying past the end of the array provided by the caller, writing over who-knows-what happened to be there. Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23Stack buffer overflow in XGetDeviceButtonMapping() [CVE-2013-1998 1/3]Alan Coopersmith
We copy the entire reply sent by the server into the fixed size mapping[] array on the stack, even if the server says it's a larger size than the mapping array can hold. HULK SMASH STACK! Reported-by: Ilja Van Sprundel <ivansprundel@ioactive.com> Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-23Use _XEatDataWords to avoid overflow of rep.length bit shiftingAlan Coopersmith
rep.length is a CARD32, so rep.length << 2 could overflow in 32-bit builds Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-05-20Copy the sequence number into the target event too (#64687)Peter Hutterer
X.Org Bug 64687 <http://bugs.freedesktop.org/show_bug.cgi?id=64687> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Reviewed-by: Jasper St. Pierre <jstpierre@mecheye.net>
2013-04-30Expand comment on the memory vs. reply ordering in XIGetSelectedEvents()Alan Coopersmith
Unpacking from the wire involves un-interleaving the structs & masks, which wasn't obvious to me the first time I read it, so make notes before I forget again. Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2013-04-05Require XFixes for PointerBarrier, remove duplicate typedefPeter Hutterer
The PointerBarrier typedef is duplicate if a client includes both Xfixes.h and XInput2.h. gcc 4.6 won't complain about that, but earlier versions do: http://gcc.gnu.org/git/?p=gcc.git;a=commitdiff;h=ce3765bf44e49ef0568a1ad4a0b7f807591d6412 gcc 4.6 with -pedantic-errors shows: /opt/xorg/include/X11/extensions/XInput2.h:172:13: error: redefinition of typedef ‘PointerBarrier’ [-pedantic] In file included from test.c:1:0: /opt/xorg/include/X11/extensions/Xfixes.h:255:13: note: previous declaration of ‘PointerBarrier’ was here Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net> Reviewed-by: Julien Cristau <jcristau@debian.org>
2013-01-15Add missing XI_RawTouch* in XInputCopyCookieBenjamin Tissoires
Looks like XI_RawTouch* events are missing in the big switch in this function. When running XIT tests for multitouch devices, several following errors appears: XInputCopyCookie: Failed to copy evtype 22 XInputCopyCookie: Failed to copy evtype 23 XInputCopyCookie: Failed to copy evtype 24 Signed-off-by: Benjamin Tissoires <benjamin.tissoires@gmail.com> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
2012-12-09Add support for pointer barrier eventsJasper St. Pierre
Signed-off-by: Jasper St. Pierre <jstpierre@mecheye.net> Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-25 13:48:53 +0000