kwin/plugins/platforms/drm/drm_output.h

187 lines
4.9 KiB
C
Raw Normal View History

2020-08-02 22:22:19 +00:00
/*
KWin - the KDE window manager
This file is part of the KDE project.
2020-08-02 22:22:19 +00:00
SPDX-FileCopyrightText: 2015 Martin Gräßlin <mgraesslin@kde.org>
2020-08-02 22:22:19 +00:00
SPDX-License-Identifier: GPL-2.0-or-later
*/
#ifndef KWIN_DRM_OUTPUT_H
#define KWIN_DRM_OUTPUT_H
#include "abstract_wayland_output.h"
#include "drm_pointer.h"
#include "drm_object.h"
#include "drm_object_plane.h"
#include "edid.h"
#include <QObject>
#include <QPoint>
#include <QSize>
#include <QVector>
#include <xf86drmMode.h>
namespace KWin
{
class DrmBackend;
class DrmBuffer;
class DrmDumbBuffer;
class DrmPlane;
class DrmConnector;
class DrmCrtc;
class Cursor;
class DrmGpu;
class KWIN_EXPORT DrmOutput : public AbstractWaylandOutput
{
Q_OBJECT
public:
///deletes the output, calling this whilst a page flip is pending will result in an error
~DrmOutput() override;
///queues deleting the output after a page flip has completed.
void teardown();
[DRM plugin] Remember static kernel objects, amplify use of DrmCrtc To get an image from KWin to the screen in the DRM pipeline we combine a CRTC, an encoder and a connector. These objects are static in the sense, that they represent real hardware on the graphics card, which doesn't change in a session. See here for more details: https://01.org/linuxgraphics/gfx-docs/drm/gpu/drm-kms.html Until now we used DrmOutput as the main representation for such an active rendering pipeline. I.e. it gets created and destroyed on hot plug events of displays. On the other side we had no fixed representation of the static kernel objects throughout the lifetime of KWin. This has several disadvantages: * We always need to query all available static objects on an hot plug event. * We can't manipulate the frame buffer of a CRTC after an output has been disconnected * Adding functionality for driving multiple displays on a single CRTC (i.e. cloning) would be difficult * We can't destroy the last frame buffer on display disconnect because the CRTC still accesses it and have therefore a memory leak on every display disconnect This patch solves these issues by storing representations of all available CRTC and Connector objects in DrmBackend on init via DrmCrtc and DrmConnector instances. On an hotplug event these vectors are looped for a fitting CRTC and Connector combinations. Buffer handling is moved to the respective CRTC instance. All changes in overview: * Query all available CRTCs and Connectors and save for subsequent hotplug events * Fix logic errors in `queryResources()` * Move framebuffers, buffer flip and blank logic in DrmCrtc * Remove `restoreSaved()`. It isn't necessary and is dangerous if the old framebuffer was deleted in the meantime. Also could reveal sensitive user info from old session. Test Plan: Login, logout, VT switching, connect and disconnect external monitor, energy saving mode. Reviewers: #kwin Subscribers: kwin, #kwin Tags: #kwin Differential Revision: https://phabricator.kde.org/D5118
2017-05-09 18:02:49 +00:00
void releaseGbm();
bool showCursor(DrmDumbBuffer *buffer);
bool showCursor();
bool hideCursor();
void updateCursor();
void moveCursor(Cursor *cursor, const QPoint &globalPos);
bool init(drmModeConnector *connector);
bool present(DrmBuffer *buffer);
void pageFlipped();
// These values are defined by the kernel
enum class DpmsMode {
On = DRM_MODE_DPMS_ON,
Standby = DRM_MODE_DPMS_STANDBY,
Suspend = DRM_MODE_DPMS_SUSPEND,
Off = DRM_MODE_DPMS_OFF
};
Q_ENUM(DpmsMode);
bool isDpmsEnabled() const {
// We care for current as well as pending mode in order to allow first present in AMS.
return m_dpmsModePending == DpmsMode::On;
}
DpmsMode dpmsMode() const {
return m_dpmsMode;
}
DpmsMode dpmsModePending() const {
return m_dpmsModePending;
}
[platforms/drm] EGLStream DRM Backend Initial Implementation Summary: This is the initial implementation of a DRM backend based on the EGLDevice, EGLOutput, and EGLStream extensions, supporting NVIDIA graphics hardware using their proprietary driver. The new backend will be used if the environment variable KWIN_DRM_USE_EGL_STREAMS is set. On initialization, it will attempt to create an EGLDevice based on the DRM device currently in use and create EGLOutputs and EGLStreams for any attached displays. These are used to control presentation of the final composited frame. Additionally, it will register the wl_eglstream_controller Wayland interface so that native EGL windows created by clients can be attached to an EGLStream allowing buffer contents to be shared with the compositor as a GL texture. At this time there are two known bugs in the NVIDIA driver's EGL implementation affecting desktop functionality. The first can result in tooltip windows drawn by plasmashell to contain incorrect contents. The second prevents KWayland from being able to query the format of EGLStream-backed buffers which interferes with the blur effect. Fixes for both of these are currently in development and should appear in an upcoming NVIDIA driver release. Additionally, hardware cursors are currently not supported with this backend. Enabling them causes the desktop to intermittently hang for several seconds. This is also likely a bug in the NVIDIA DRM-KMS implementation but the root cause is still under investigation. Test Plan: On a system with an NVIDIA graphics card running a recent release of their proprietary driver * Ensure the nvidia_drm kernel module is loaded with the option "modeset=1" ("# cat /sys/module/nvidia_drm/parameters/modeset" should print "Y") * Ensure EGL external platform support is installed https://github.com/NVIDIA/eglexternalplatform * Ensure KWin was build with the CMake option KWIN_BUILD_EGL_STREAM_BACKEND=ON (this is the default) * Start a plasma wayland session with the environment variable KWIN_DRM_USE_EGL_STREAMS set * Ensure output from KWin OpenGL initialization indicates the NVIDIA EGL driver is in use (as opposed to Mesa / llvmpipe). * Desktop should be fully functional and perform smoothly. Reviewers: #kwin, romangg, davidedmundson Reviewed By: #kwin, romangg, davidedmundson Subscribers: kwin Tags: #kwin Differential Revision: https://phabricator.kde.org/D18570
2019-04-15 14:26:22 +00:00
const DrmCrtc *crtc() const {
return m_crtc;
}
const DrmConnector *connector() const {
return m_conn;
}
[platforms/drm] EGLStream DRM Backend Initial Implementation Summary: This is the initial implementation of a DRM backend based on the EGLDevice, EGLOutput, and EGLStream extensions, supporting NVIDIA graphics hardware using their proprietary driver. The new backend will be used if the environment variable KWIN_DRM_USE_EGL_STREAMS is set. On initialization, it will attempt to create an EGLDevice based on the DRM device currently in use and create EGLOutputs and EGLStreams for any attached displays. These are used to control presentation of the final composited frame. Additionally, it will register the wl_eglstream_controller Wayland interface so that native EGL windows created by clients can be attached to an EGLStream allowing buffer contents to be shared with the compositor as a GL texture. At this time there are two known bugs in the NVIDIA driver's EGL implementation affecting desktop functionality. The first can result in tooltip windows drawn by plasmashell to contain incorrect contents. The second prevents KWayland from being able to query the format of EGLStream-backed buffers which interferes with the blur effect. Fixes for both of these are currently in development and should appear in an upcoming NVIDIA driver release. Additionally, hardware cursors are currently not supported with this backend. Enabling them causes the desktop to intermittently hang for several seconds. This is also likely a bug in the NVIDIA DRM-KMS implementation but the root cause is still under investigation. Test Plan: On a system with an NVIDIA graphics card running a recent release of their proprietary driver * Ensure the nvidia_drm kernel module is loaded with the option "modeset=1" ("# cat /sys/module/nvidia_drm/parameters/modeset" should print "Y") * Ensure EGL external platform support is installed https://github.com/NVIDIA/eglexternalplatform * Ensure KWin was build with the CMake option KWIN_BUILD_EGL_STREAM_BACKEND=ON (this is the default) * Start a plasma wayland session with the environment variable KWIN_DRM_USE_EGL_STREAMS set * Ensure output from KWin OpenGL initialization indicates the NVIDIA EGL driver is in use (as opposed to Mesa / llvmpipe). * Desktop should be fully functional and perform smoothly. Reviewers: #kwin, romangg, davidedmundson Reviewed By: #kwin, romangg, davidedmundson Subscribers: kwin Tags: #kwin Differential Revision: https://phabricator.kde.org/D18570
2019-04-15 14:26:22 +00:00
const DrmPlane *primaryPlane() const {
return m_primaryPlane;
}
bool initCursor(const QSize &cursorSize);
bool supportsTransformations() const;
/**
* Drm planes might be capable of realizing the current output transform without usage
* of compositing. This is a getter to query the current state of that
*
* @return true if the hardware realizes the transform without further assistance
*/
bool hardwareTransforms() const;
DrmGpu *gpu() {
return m_gpu;
}
private:
friend class DrmGpu;
friend class DrmBackend;
[DRM plugin] Remember static kernel objects, amplify use of DrmCrtc To get an image from KWin to the screen in the DRM pipeline we combine a CRTC, an encoder and a connector. These objects are static in the sense, that they represent real hardware on the graphics card, which doesn't change in a session. See here for more details: https://01.org/linuxgraphics/gfx-docs/drm/gpu/drm-kms.html Until now we used DrmOutput as the main representation for such an active rendering pipeline. I.e. it gets created and destroyed on hot plug events of displays. On the other side we had no fixed representation of the static kernel objects throughout the lifetime of KWin. This has several disadvantages: * We always need to query all available static objects on an hot plug event. * We can't manipulate the frame buffer of a CRTC after an output has been disconnected * Adding functionality for driving multiple displays on a single CRTC (i.e. cloning) would be difficult * We can't destroy the last frame buffer on display disconnect because the CRTC still accesses it and have therefore a memory leak on every display disconnect This patch solves these issues by storing representations of all available CRTC and Connector objects in DrmBackend on init via DrmCrtc and DrmConnector instances. On an hotplug event these vectors are looped for a fitting CRTC and Connector combinations. Buffer handling is moved to the respective CRTC instance. All changes in overview: * Query all available CRTCs and Connectors and save for subsequent hotplug events * Fix logic errors in `queryResources()` * Move framebuffers, buffer flip and blank logic in DrmCrtc * Remove `restoreSaved()`. It isn't necessary and is dangerous if the old framebuffer was deleted in the meantime. Also could reveal sensitive user info from old session. Test Plan: Login, logout, VT switching, connect and disconnect external monitor, energy saving mode. Reviewers: #kwin Subscribers: kwin, #kwin Tags: #kwin Differential Revision: https://phabricator.kde.org/D5118
2017-05-09 18:02:49 +00:00
friend class DrmCrtc; // TODO: For use of setModeLegacy. Remove later when we allow multiple connectors per crtc
// and save the connector ids in the DrmCrtc instance.
DrmOutput(DrmBackend *backend, DrmGpu* gpu);
bool presentAtomically(DrmBuffer *buffer);
enum class AtomicCommitMode {
Test,
Real
};
bool doAtomicCommit(AtomicCommitMode mode);
bool presentLegacy(DrmBuffer *buffer);
bool setModeLegacy(DrmBuffer *buffer);
void initEdid(drmModeConnector *connector);
void initDpms(drmModeConnector *connector);
void initOutputDevice(drmModeConnector *connector);
bool isCurrentMode(const drmModeModeInfo *mode) const;
void initUuid();
bool initPrimaryPlane();
bool initCursorPlane();
void atomicEnable();
void atomicDisable();
void updateEnablement(bool enable) override;
bool dpmsAtomicOff();
bool dpmsLegacyApply();
void dpmsFinishOn();
void dpmsFinishOff();
bool atomicReqModesetPopulate(drmModeAtomicReq *req, bool enable);
void updateDpms(KWaylandServer::OutputInterface::DpmsMode mode) override;
void updateMode(int modeIndex) override;
void setWaylandMode();
void updateTransform(Transform transform) override;
int gammaRampSize() const override;
bool setGammaRamp(const GammaRamp &gamma) override;
DrmBackend *m_backend;
DrmGpu *m_gpu;
[DRM plugin] Remember static kernel objects, amplify use of DrmCrtc To get an image from KWin to the screen in the DRM pipeline we combine a CRTC, an encoder and a connector. These objects are static in the sense, that they represent real hardware on the graphics card, which doesn't change in a session. See here for more details: https://01.org/linuxgraphics/gfx-docs/drm/gpu/drm-kms.html Until now we used DrmOutput as the main representation for such an active rendering pipeline. I.e. it gets created and destroyed on hot plug events of displays. On the other side we had no fixed representation of the static kernel objects throughout the lifetime of KWin. This has several disadvantages: * We always need to query all available static objects on an hot plug event. * We can't manipulate the frame buffer of a CRTC after an output has been disconnected * Adding functionality for driving multiple displays on a single CRTC (i.e. cloning) would be difficult * We can't destroy the last frame buffer on display disconnect because the CRTC still accesses it and have therefore a memory leak on every display disconnect This patch solves these issues by storing representations of all available CRTC and Connector objects in DrmBackend on init via DrmCrtc and DrmConnector instances. On an hotplug event these vectors are looped for a fitting CRTC and Connector combinations. Buffer handling is moved to the respective CRTC instance. All changes in overview: * Query all available CRTCs and Connectors and save for subsequent hotplug events * Fix logic errors in `queryResources()` * Move framebuffers, buffer flip and blank logic in DrmCrtc * Remove `restoreSaved()`. It isn't necessary and is dangerous if the old framebuffer was deleted in the meantime. Also could reveal sensitive user info from old session. Test Plan: Login, logout, VT switching, connect and disconnect external monitor, energy saving mode. Reviewers: #kwin Subscribers: kwin, #kwin Tags: #kwin Differential Revision: https://phabricator.kde.org/D5118
2017-05-09 18:02:49 +00:00
DrmConnector *m_conn = nullptr;
DrmCrtc *m_crtc = nullptr;
bool m_lastGbm = false;
drmModeModeInfo m_mode;
Edid m_edid;
DrmScopedPointer<drmModePropertyRes> m_dpms;
DpmsMode m_dpmsMode = DpmsMode::On;
DpmsMode m_dpmsModePending = DpmsMode::On;
QByteArray m_uuid;
uint32_t m_blobId = 0;
DrmPlane *m_primaryPlane = nullptr;
DrmPlane *m_cursorPlane = nullptr;
QVector<DrmPlane*> m_nextPlanesFlipList;
bool m_pageFlipPending = false;
bool m_atomicOffPending = false;
bool m_modesetRequested = true;
struct {
Transform transform;
drmModeModeInfo mode;
DrmPlane::Transformations planeTransformations;
QPoint globalPos;
bool valid = false;
} m_lastWorkingState;
QScopedPointer<DrmDumbBuffer> m_cursor[2];
int m_cursorIndex = 0;
bool m_hasNewCursor = false;
bool m_deleted = false;
};
}
Q_DECLARE_METATYPE(KWin::DrmOutput*)
QDebug& operator<<(QDebug& stream, const KWin::DrmOutput *);
#endif