summaryrefslogtreecommitdiffstats
path: root/src/plugins/platforms/xcb
Commit message (Collapse)AuthorAgeFilesLines
* Fix XCB on endian mismatched client and server with SHM offAllan Sandfeld Jensen2018-09-121-0/+2
| | | | | | | If SHM is disabled, that code path already does its own bswaping. Change-Id: I6c17f6c5c5502c8f89098d38d931b6b8f50b2640 Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
* Silence warning about comparing signed and unsignedShawn Rutledge2018-08-141-2/+2
| | | | | | | | qxcbimage.cpp:72:26: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare] Task-number: QTBUG-69923 Change-Id: Icdb4ce8cb7ce5b48d7ee3839166eb1c7c9520c78 Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
* xcb: make sure we have a valid m_qimage in backing storeGatis Paeglis2018-08-131-50/+61
| | | | | | | | | | | | | | | | | | | | | | | | This patch amends a62f1d03560937a306c7586669a46cd9575e9464. If the initial backing store resize request is called with QSize(0, 0), we end up with QXcbBackingStoreImage holding a default contructed QImage / m_qimage. This happens because of the logic in QXcbBackingStoreImage::create(), where if we detect that the requested segmentSize == 0, we do not allocate any memory, and thus don't create a valid image in m_qimage. On subsequent call to QXcbBackingStore::resize() we would only check if QXcbBackingStoreImage object has been created, but not if it is in a valid state. This obviously would cause problems. This patch re-factors the logic to handle better resize to QSize(0, 0). And make the code cleaner by: - merging ::create and ::resize as semantically it is always resize(). - dropping unnecessary argument passing. Task-number: QTBUG-69581 Change-Id: Ied337beb449dea8259fcf6b7d29f0a5bd553019d Reviewed-by: Błażej Szczygieł <spaz16@wp.pl> Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
* xcb: partly revert 3bc0f1724ae49c2fd7e6d7bcb650350d20d12246Gatis Paeglis2018-08-014-101/+3
| | | | | | | | | | | | | | | After trying to fix (work around) system resize/move issues in various ways from within the platform plugin, it has been concluded that it is a bug at widget layer and should be fixed there instead: QTBUG-69716. This patch reverts parts of 3bc0f1724a and disables system move / resize on XCB plugin. Meaning, QSizeGrip will use its own implementation for resizing a window. Task-number: QTBUG-68501 Task-number: QTBUG-69628 Change-Id: Ib4744a93fb3e3c20f690a8f43713103856cb7d1a Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* xcb: Fix build when xinput2 is not availableFriedemann Kleint2018-07-251-4/+10
| | | | | | | | | | | | Fix the #ifdefs. Amends 3bc0f1724ae49c2fd7e6d7bcb650350d20d12246. Task-number: QTBUG-68501 Task-number: QTBUG-51385 Task-number: QTBUG-32476 Change-Id: Icc6421fe2e91a3b29bcec8cb1a8a91cb71ae3172 Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* xcb: fix various bugs with _NET_WM_MOVERESIZEGatis Paeglis2018-07-235-9/+129
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | 1) After a37785ec7638e7485112b87dd7e767881fecc114 it become apparent that we don't get mouse release event from X server when system move/resize ends (because WM is grabbing the pointer). The old code (before a37785ec) would wrongly deduce mouse move as mouse release, which is why the issue was not seen before. The solution is to subscribe to slave device events. 2) This patch also amends 2488f34ecfd68702b5508c50cca3fb8e967ac8ea as that patch was solving the issue only for 1/3 of the supported DEs. It worked with KWin, but not with Unity and Gnome. Its worth noting that it also worked with two other WMs that I tested - openbox and awesomewm. The way forward is to detect when system move/resize was started as a result of touch event and let the QSizeGrip do the move/resize instead of WMs that are known to have bugs. With this patch we also need to adjust the event compression algorithm to not treat all XI_TouchUpdate events equally. For XI_Motion we don't care if the event that we process comes from a master or a slave device, so we can process them as equal. Task-number: QTBUG-68501 Task-number: QTBUG-51385 Task-number: QTBUG-32476 Change-Id: Iab4e79a289d7bc0fe26f7ae2cff7c562f51a3334 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* XCB/Xlib: make sure we don't get problems for sys headers using registerThiago Macieira2018-07-117-0/+14
| | | | | | | | | Found while compiling on FreeBSD 11.2 (clang 6 update has the warning): /usr/local/include/X11/Xlibint.h:675:7: error: ISO C++17 does not allow 'register' storage class specifier [-Wregister] Change-Id: I117816bf0f5e469b8d34fffd153e6482ccaed69f Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* xcb: Monitor XSETTINGS changeAlexander Volkov2018-07-101-0/+1
| | | | | | | | | | | | | Register QXcbXSettings object as a listener of events received by XSETTINGS window. It was forgotten to be done in 0f31a5d91f13bd6a574bed8db894a7ebb2813292 which introduced QXcbXSettings. XSettings are used by QXcbCursor to get cursor theme. Without this change QXcbCursor can't react on theme change. Change-Id: I0fdd4c913b1d7a482b507c5a054b7052cac61666 Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* xcb: Move handling of RRScreenChangeNotify from QXcbScreen to QXcbVirtualDesktopAlexander Volkov2018-07-093-100/+103
| | | | | | | | | | | | | | | This event relates to xcb_screen_t (virtual desktop), not to RandR crtcs, so move its processing to QXcbVirtualDesktop. Currently it triggers redundant calling of RRGetCrtcInfo (from QXcbScreen::updateGeometry()). It's called anyway after receiving RRCrtcChangeNotify, so just drop it. There is also a mess between the rotation of Screen and crtcs, obviously they should be processed separately. Task-number: QTBUG-65598 Change-Id: I124752ccbde03adb15e9ba592dd8b2d8d7fc35f4 Reviewed-by: Daniel Vrátil <dvratil@kde.org> Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* Do not update scroll coordinates on ignorable enter eventsAllan Sandfeld Jensen2018-06-221-4/+4
| | | | | | | | | | | | | | | This greatly reduces how often we reset the scroll evaluators, especially with non-focused windows in KWin which sends an enter for every wheel events in that case. The update of the evaluators also has race conditions with the normal events, and thus reducing them fixes odd scrolling behavior with rapid firing mouse wheels. Task-number: QTBUG-42415 Task-number: QTBUG-68734 Change-Id: I1c14ca3352bf9c6e57e47ad3aaee1712fe6ba30b Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* Recognize X11/XCB keysyms Undo, Redo, Find, CancelJonathan Marten2018-06-221-0/+7
| | | | | | | | | Convert these keysyms into the corresponding Qt::Key_ enum values, so that they can be part of a QKeySequence and used by applications. Task-number: QTBUG-69062 Change-Id: I6f2e28191dd8dacd63d4bf710e1714fc5dcce75f Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* xcb: add qt.qpa.input.events logging of wheel eventsShawn Rutledge2018-06-211-0/+2
| | | | | | | | | | This is comparable to what we do on macOS, except that the scroll phase and inverted state are missing. Task-number: QTBUG-38570 Task-number: QTBUG-56075 Change-Id: I27502e1e2667317ab701f30f1fc601ae1e0591d0 Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* xcb: add QT_XCB_NO_MITSHM envvar for disabling MIT-SHM extensionGatis Paeglis2018-06-151-1/+2
| | | | | | | | | | | We had something like this already in Qt4: QT_X11_NO_MITSHM The logic from 67227aeffdf94be8d177309d27291d5b3247586c not always works. There can still be cases that xcb_shm_attach_checked() returns with no errors on remote clients. Task-number: QTBUG-68783 Change-Id: Idd27ac66eb8f1114e3d1e1ddaaab2b00f235c561 Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* xcb: fix regression with remote X11 clientsGatis Paeglis2018-06-113-53/+88
| | | | | | | | | | | | | | | | | | | | | | | | | There were several issues here: We were attempting to use MIT-SHM functions over SSH connection, which is not supported. X server should detect this and return with an appropriate error message. It does actually return BadAccess for non-fd code path, but Qt was stubbornly trying to repeat this action and always falling back to malloc (during window resizing). For fd code path we were hitting X server bug, which would result in window freeze [1]. During the initialization we check if xcb_shm_attach_checked() fails, and disable MIT-SHM if it does. We use this logic to detect if we are running remotely, as there are no public APIs for it. This way we can avoid X server bug and avoid needless calling of code path which will _always_ fail on a remote X11 connection. [1] https://lists.x.org/archives/xorg-devel/2018-June/057011.html Task-number: QTBUG-68449 Task-number: QTBUG-68783 Change-Id: I7ab3dcf0f323fd53001b9f7b88c2cb10809af509 Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* Revert "xcb: round down the scale factor for values < 0.8"Mitch Curtis2018-06-091-6/+1
| | | | | | | | | | | This reverts commit 23b139038a1dc9a769a358ab112453abcdd39290. The commit causes a regression on Ubuntu where the DPI setting of the display is ignored. Task-number: QTBUG-68620 Change-Id: Id176f8cda0daef1755abb3aa9382476cc4ed7d71 Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
* xcb: fix mouse event compression with certain configurationsGatis Paeglis2018-06-091-2/+4
| | | | | | | | | | | | | | | The bug was that we are accessing memory beyond 32 bytes. It is not safe to cast xcb_generic_event_t to Xlib's XI2 structs before we have memmoved bits to the expected layout (for details see QXcbConnection::xi2PrepareXIGenericDeviceEvent). We do this memmove later in the stack, when processing the XI2 events. Here at the compression step we can simply extract the necessary sourceId by reading the sourceId offset in the data. Task-number: QTBUG-68033 Change-Id: I6962bbb8f8b0834d6f780f62017fefa2de7f47df Reviewed-by: Mikhail Svetkin <mikhail.svetkin@qt.io> Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* Make sure we can build with -no-feature-draganddropJoerg Bornemann2018-05-2910-30/+24
| | | | | | | | | We move QInternalMimeData to a separate file, because this class is used, even if draganddrop is disabled. From now on, include qinternalmimedata_p.h instead of qdnd_p.h for QInternalMimeData. Change-Id: I594e08e2e90d574dc445119091686b4b69e4731b Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* Fix build without features.cursorTasuku Suzuki2018-05-151-0/+2
| | | | | Change-Id: I450bd9f160c64f718c49e87d274c1ccc4a657aca Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
* Merge remote-tracking branch 'origin/5.11.0' into 5.11Qt Forward Merge Bot2018-05-091-15/+25
|\ | | | | | | Change-Id: Id6e2acd5e31c1ac858ddf1d8873a6f10694141de
| * Also support reset notification on non-nvidiaAllan Sandfeld Jensen2018-04-191-15/+25
| | | | | | | | | | | | | | | | In 5.11 robustness was added for nvidia drivers, this patch extends that to other opengl drivers, and fixes the reported format. Change-Id: Ia81934c3bcf13e7300fb4e75674ea63317039870 Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* | xcb: round down the scale factor for values < 0.8Mathieu Velten2018-05-021-1/+6
|/ | | | | | | | | | This solves huge UI on some (most ?) HD 13 inches that have a calculated scale factor around 1.75. Task-number: QTBUG-53022 QTBUG-57211 Change-Id: I6cc22d3e47d38cb1d5a0daf5646a936256664e48 Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io> Reviewed-by: Frederik Gladhorn <frederik.gladhorn@qt.io>
* xcb: prevent crash with pixmap cursors on XRender-less X serversGatis Paeglis2018-04-143-8/+28
| | | | | | | | | | | | | | | We were using xcb_render_* APIs without checking if the server even supports this extension. Attempting to use an extension which is not present will always result in a crash. This patch adds the required guards and refactors how we detect presence of XRender extension. Also instead of falling back to some odd-looking bitmapped version just leave the current cursor unchanged. That is how we did it in Qt4 AFAICT. Task-number: QTBUG-66935 Change-Id: I4f27f1d65a77563ec34f3e0e94492c9236d7f9a6 Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* xcb: fix bitmap cursor loading performance regressionGatis Paeglis2018-04-144-45/+38
| | | | | | | | | | | | | | | | ... introduced by 422838685c31d9b57133a8711bfd5db92095d96d. Instead of completely droping caching for bitmap cursors we can do the same what is done in libXcursor - have a fixed size cache, with oldest entries eventually being replaced with new bitmaps. This fixes the original issue, where the hash was growing indefinitely until running out of file descriptors and won't have the performance penalty as in 422838685c31d9b57133a8711bfd5db92095d96d. Task-number: QTBUG-66897 Change-Id: I14f80b46f97fd0e2c920e17a31ffbc0441cd9d22 Reviewed-by: Uli Schlachter <psychon@znc.in> Reviewed-by: Robin Burchell <robin.burchell@crimson.no> Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* xcb: cleanup the code for detecting when to create/destroy SHM segmentGatis Paeglis2018-04-141-6/+8
| | | | | | | | | | | - removed the check for "m_segmentSize > 0" as according to the code it will never be <= 0. - wrap the entire logic in connection()->hasShm() { .. } as that is when the logic becomes relevant. This makes the code more readable. Change-Id: I572420df8e29cc46593f8a13c250f8c05c6a9108 Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* glx: Fix loading OpenGL library when libGL.so symlink is missingDmitry Shachnev2018-04-101-0/+2
| | | | | | | | Task-number: QTBUG-67537 Change-Id: Ib59bc27582f44b0c6b04efa57c44d877e1685bbb Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io> Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io> Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
* xcb: various coding style cleanups in qxcbbackingstoreGatis Paeglis2018-04-092-37/+33
| | | | | | | | | | | - Pass QXcbBackingStoreImage a pointer to QXcbBackingStore. This allow for simpler QXcbBackingStoreImage ctor. - Use member initializers. Change-Id: Ia992390060bb30e1184813cd0d115a8bf0fbc237 Reviewed-by: Błażej Szczygieł <spaz16@wp.pl> Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* xcb: rename two classes in qxcbbackingstoreGatis Paeglis2018-04-092-23/+23
| | | | | | | | | | | | | | QXcbShmImage to QXcbBackingStoreImage as it is an image that might or might not have SHM capability. The current name implies that it always uses SHM. QXcbShmGraphicsBuffer to QXcbGraphicsBuffer as it has nothing to do with SHM. Change-Id: I57ced75891e8b10515142769278a7f3f40da91ef Reviewed-by: Alexander Volkov <a.volkov@rusbitech.ru> Reviewed-by: Błażej Szczygieł <spaz16@wp.pl> Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* Make nvidia resets opt-in based on QSurfaceFormat::ResetNotificationDavid Edmundson2018-04-051-1/+1
| | | | | | | | | | We can't be sure every user will check the return value from makeCurrent and reset appropriately. Even though after a reset a user will be left with the same garbage as before, it's safer than a potential infinite loop. Change-Id: I5b328c654ad2a89c5b8c4399e2eb38150f4f384b Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* Support graphics reset status for offscreen surfaces in XCB-GLXDavid Edmundson2018-04-051-0/+6
| | | | | | | | | | | | I originally didn't to match the ANGLE code, however it's important for all surfaces to be able to know to reset if their makeContext fails. This is espcially relevant on some Linux nvidia drivers which has a bug where while(glGetError()) won't ever clear whilst a reset is in progress. Change-Id: Iae3502168cda748c601a9aee1497e40c6d82cd83 Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* Support GLX_NV_robustness_video_memory_purge in XCB-GLXDavid Edmundson2018-03-222-8/+58
| | | | | | | | | | | | | Nvidia doesn't handle suspend very well in Linux and textures get corrupted. To handle this Nvidia has a bespoke extension to query when this has happened. This patch checks if graphics have been reset and invalidates the surface. Task-number: QTBUG-56610 Change-Id: I0b97d539ce6cc2b9cfe41c71bf6efd4f68496cd6 Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
* xcb: Prevent shared memory allocation on every window resizeBłażej Szczygieł2018-03-141-35/+83
| | | | | | | | | | | Allocate new shared memory only when window size grows or when window size is 2 times smaller than allocated memory size. This improves window resizing performance and also allows to free some memory if window becames much smaller. Change-Id: I3454cd3c6023eede8242d6b29038f4dd6638f9f1 Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io> Reviewed-by: Alexander Volkov <a.volkov@rusbitech.ru>
* qxcbconnection_xi2.cpp: fix debug check to avoid confusing debug outputDavid Faure2018-03-091-1/+1
| | | | | | | | | The check on line 972 says lcQpaXInputEvents().isDebugEnabled() so that debug output was always showing "from 0,0" due to lastScrollPosition not being set. Change-Id: I345732e36db05108f70474261c47e78333b57d30 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* xcb: Fix FP1616 to int conversionAlexander Volkov2018-03-091-1/+1
| | | | | | | | | | Divide the whole FP1616 value by 0x10000 instead of dividing the integer part by 0x10000 and the fractional part by 0xFFFF. It also makes fixed1616ToInt() consistent with fixed1616ToReal(), see 7d3f353a5bd573dc0e72f7f55c70212a6b3837fa. Change-Id: Id76025028c926872b002ef0a1ca8a8bdc2de3e1e Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
* Fix QXcbScreen::format() when rgb-swap is neededAllan Sandfeld Jensen2018-03-011-1/+3
| | | | | | | | | QXcbScreen::format() was producing console warning and returning invalid format if rgb-swap was necessary. Let it match possibly reversed formats since the backing-store will swizzle it anyway. Change-Id: I6a53cb8ceda029e9d708d25ad635832f1163ffe7 Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
* Silence a GCC 8 warning in qxcbnativeinterfaceVille Voutilainen2018-02-271-7/+7
| | | | | | | | qtbase/src/plugins/platforms/xcb/qxcbnativeinterface.cpp:309:65: error: cast between incompatible function types from ‘void (*)(const char*)’ to ‘QPlatformNativeInterface::NativeResourceForIntegrationFunction’ {aka ‘void* (*)()’} [-Werror=cast-function-type] return NativeResourceForIntegrationFunction(setStartupId); Change-Id: I83409b3a6cc67ccb4c9e91e592e6a01bb6ce45ea Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
* xcb: simplify handling of keymap updatesGatis Paeglis2018-02-253-43/+41
| | | | | | | The old code was somewhat too scattered. Change-Id: Ib0445c66653f757ccac28778f34f4bcb5df49a70 Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
* xcb: cleanup TODOs for KeyboardLayoutChangeGatis Paeglis2018-02-252-18/+23
| | | | | Change-Id: I5d2207fdb21cbabb04460faf73c35275bc76b7b0 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* xcb: minor refactoring in QXcbKeyboard::checkForLatinLayout()Gatis Paeglis2018-02-251-6/+6
| | | | | | | We don't need xkb state APIs to check for keys on first level. Change-Id: I728e6bfe09bce127ad8eae78ecee7cefd620f52e Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* xcb: remove X11 deps from QXcbKeyboard::checkForLatinLayout/lookupLatinKeysym()Gatis Paeglis2018-02-251-4/+4
| | | | | | | | | | This will allow us to reuse these functions on systems that do not rely on X11, but use libxkbcommon for handling keyboard input. Task-number: QTBUG-65503 Change-Id: I78034238771be96fbb38e8187801fefbee1a5fed Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* xcb: have a proper detection of key events originating from SendEventGatis Paeglis2018-02-252-47/+58
| | | | | | | | | | | This is a more correct fix for QTBUG-48795. The original fix was unnecessarily using non-XKB code path for updating state for all incoming key events. This would result in losing some valuable bits from xkb state. Task-number: QTBUG-48795 Change-Id: Ic4fb28b2d834272f1db2cbf5888cafb209707847 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
* xcb: use keysym defs from xkbcommon, instead of X11 headersGatis Paeglis2018-02-251-590/+322
| | | | | | | | | | | | | | | | xkbcommon-keysyms.h is generated from X11 header files, so it contains all the same values. Removed all of #ifndef XK_* as those keysyms are present in xkbcommon-keysyms.h (checked the header from 0.4.1, the minimal required version). The same for XF86XK_* defines. This will allow to reuse some of this code on platforms that don't depend on X11, for details see QTBUG-65503. Task-number: QTBUG-65503 Change-Id: I68083e11cea1f29d775a6ed46503a06b04b9a05c Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* xcb: minor cleanup in QXcbKeyboard::handleKeyEvent()v5.11.0-beta1Gatis Paeglis2018-02-241-15/+23
| | | | | | | | | | - Use smart pointer for handling xkb state. - Make it more clear (in the code and the comment) when a latin keysym is used. Change-Id: Iee8106c72177c22b1a8fe875027b1dda82196b36 Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
* xcb: refactor QXcbKeyboard::keysymToQtKey() and fix bugGatis Paeglis2018-02-243-61/+102
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Now also digits from other alphabets e.g ۲ (arabic two) are mapped to Qt::Key_* digit keys. Re-factored logic: - All known dead keys have direct mappings since 1d86e5f84abac6db0b7b1503a6f52c72b272a897. Don't special treat them in "unicode mapping" code path. - Removed the ISO8859-1 legacy logic, which is leftover from Qt4 where keysym to Qt decoding was done from raw data. In Qt5 we always get a utf8 string from xkb_state_key_get_utf8(). Furthermore, ISO8859-1 and utf8 encode ASCII exactly the same way. - Set Qt::KeypadModifier from key input handler methods. This logic does not belong in keysymToQtKey(). Note: KeyTbl[] and keysymToQtKey() have been duplicated in several places in Qt. That stuff will be cleaned up as part of QTBUG-65503. This change will make those cleanups easier. Task-number: QTBUG-58865 Task-number: QTBUG-65503 Change-Id: Iaf10205a26804f7fc03eb8a16a0879f1bd7bf332 Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
* xcb: re-factor QXcbKeyboard::updateKeymap() to remove various fallbacksGatis Paeglis2018-02-242-176/+69
| | | | | | | | | | | | | | | | | | | | | 3edcd9420e3ad661cad89420e18dbb70e7ad450b added more robust support for keyboard input on XKeyboard-less X servers. The various fallbacks that we had did not work that well in practice. We can remove them now. The xkb_keymap_new_from_names() function relies on reading XKB config files from a file system. Since we don't use this function anymore, we can also simplify xkb context creation (see XKB_CONTEXT_NO_DEFAULT_INCLUDES), as we don't care about DFLT_XKB_CONFIG_ROOT (which we previously set via -xkb-config-root for the bundled libxkbcommon). This patch also changes the code to use smart pointers for managing the global xkb context, keymap and state. [ChangeLog][X11] The -xkb-config-root command line switch has been removed as it it no longer needed when configuring with -qt-xkbcommon-x11. Change-Id: I80eecf83adae90af5cd20df434c1fba0358a12fd Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
* xcb: remove xlib dependency from core keymap assemblingGatis Paeglis2018-02-244-24/+238
| | | | | | | | | | | | | | | Remove the Xlib dependency by extracting XConvertCase from libxkbcommon sources (xkbcommon/src/keysym.c). libxkbcommon >= 0.8.0 exposes case conversion APIs, but we should prefer using the slightly adjusted version (see the patch for more details). This change also is necessary for follow-up cleanups. Change-Id: Icf1716e0ad26f46a7aefb23722cfc57957754d5e Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io> Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
* xcb: remove fragile and unnecessary missing-latin-keymap workaroundGatis Paeglis2018-02-224-49/+32
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | ... which was trying to fix a rarely occurring situation where system settings from a desktop environment does not set any latin keymap on X. This is a DE bug and it has a simple workaround (details in the patch). Ubuntu has fixed this issue sometime between 12.10 -> 14.04. Gnome 3 always appends 'us' layout, even if you have only e.g. 'gr' listed in keyboard layouts (can be checked via setxkbmap -query). In KDE, the global system shorcuts seem to stop working as soon as latin keymap is not the first in the list, which means that KDE users won't be affected as they will likely always have a latin keymap present in the list. This patch removes parts of 2b666d9576210aa98700e219dba6b1bd4f93d793, the parts that in the commit message I was referring to by this quote: "lookupLatinKeysym() also handles the cases that did not work in Qt4 with XLookupString". Since finding a latin key is not working by XLookupString() in this rare case, then it would not work pretty much across the whole desktop. And users would be more interested at finding a solution that works across the desktop. We should not workaround this issue. Desktops that are doing it wrong should learn about this and not repeat the same mistakes on Wayland systems, where XKB keymap is assembled by compositor and passed to clients. Clients should work with the provided keymap as is. The missing-latin-keymap workaround is considered fragile for several reasons - it might not work with legacy or enterprise X server key codes and it relies on global _XKB_RULES_NAMES (there might be several connected keyboards). And theoretical limitation: client might be running in a restricted environment where we don't have access to keymaps on the file system. Change-Id: Ib445b2ea46174248cfa0e5da0eb642cd2a5cf2f6 Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
* xcb: Fix developer buildFriedemann Kleint2018-02-191-0/+3
| | | | | | | | | | | | Add Q_UNUSED, fixing: qxcbbackingstore.cpp:344:45: error: unused parameter 'segmentSize' [-Werror=unused-parameter] on Kubuntu 17.10. Amends 24adaa9a742e6f95ff897d0eb9a2bce0527dd042. Task-number: QTBUG-46017 Change-Id: I64f21d8f1d1ac21340cfbba66b97768140ce23a8 Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* xcb: Fix access to shm for X server running from another userAlexander Volkov2018-02-153-35/+109
| | | | | | | | | | | | | | | Use ShmCreateSegment call, that was added in MIT-SHM 1.2, to create shared memory segments on the server side. It returns a POSIX shared memory object that is used to mmap memory. It's in effect a file descriptor that is passed through the X server socket and thus avoids permission checks. On the other hand this scheme is more secure, because the file descriptor, and thus the shared memory, are accessible only by the X server and the application. Task-number: QTBUG-46017 Change-Id: I202eead9d01aee2ab5b65f4f74f4c13da7cb2239 Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* xcb: Enhance SHM management codeAlexander Volkov2018-02-153-44/+91
| | | | | | | | | | - extract the creation of a shared memory segment into a separate function - do extra checks for errors - check that MIT-SHM extension is present once in QXcbConnection Change-Id: I956bdf76b879ec5c95a7ed219a59ae722dc5afba Reviewed-by: Gatis Paeglis <gatis.paeglis@qt.io>
* Fix crash when reading window titles with XCBSimon Hausmann2018-02-151-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | This is a regression introduced with commit cb142954c54b7a6e391950d9209b5cea9252092b that changed the code from using QString:fromUtf8(name, propertyLength) to QString::fromUtf8(name), assuming that the property name is a zero-terminated string. That however is not correct. ASAN trace: ==4039==ERROR: AddressSanitizer: heap-buffer-overflow on address 0x60400001e0b4 at pc 0x7f3383c7d66e bp 0x7ffdc8e3d9b0 sp 0x7ffdc8e3d158 READ of size 5 at 0x60400001e0b4 thread T0 #0 0x7f3383c7d66d (/usr/lib/x86_64-linux-gnu/libasan.so.4+0x5166d) #1 0x7f337602f32a in QString::fromUtf8(char const*, int) ../../../../include/QtCore/../../src/corelib/tools/qstring.h:569 #2 0x7f337602f32a in QXcbWindow::windowTitle(QXcbConnection const*, unsigned int) /home/simon/dev/qt-5.11/qtbase/src/plugins/platforms/xcb/qxcbwindow.cpp:2861 [...] 0x60400001e0b4 is located 0 bytes to the right of 36-byte region [0x60400001e090,0x60400001e0b4) allocated by thread T1 (QXcbEventReader) here: #0 0x7f3383d0ab50 in __interceptor_malloc (/usr/lib/x86_64-linux-gnu/libasan.so.4+0xdeb50) #1 0x7f337b397e2b (/usr/lib/x86_64-linux-gnu/libxcb.so.1+0xde2b) Change-Id: Ia5024602d3aacb924b5dcd3956672da2a8f10feb Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
generated by cgit v1.2.3 (git 2.25.1) at 2024-06-07 22:47:37 +0000