mutter/src/backends/native/meta-gpu-kms.c

630 lines
16 KiB
C
Raw Normal View History

/* -*- mode: C; c-file-style: "gnu"; indent-tabs-mode: nil; -*- */
/*
* Copyright (C) 2017 Red Hat
* Copyright (c) 2018 DisplayLink (UK) Ltd.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation; either version 2 of the
* License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
* 02111-1307, USA.
*/
#include "config.h"
#include "backends/native/meta-gpu-kms.h"
#include <drm.h>
#include <drm_fourcc.h>
#include <errno.h>
#include <poll.h>
#include <string.h>
#include <time.h>
#include <xf86drm.h>
#include <xf86drmMode.h>
#include "backends/meta-crtc.h"
#include "backends/meta-monitor-manager-private.h"
#include "backends/meta-output.h"
#include "backends/native/meta-backend-native.h"
#include "backends/native/meta-crtc-kms.h"
#include "backends/native/meta-kms-connector.h"
backends/native: Add basic KMS abstraction building blocks The intention with KMS abstraction is to hide away accessing the drm functions behind an API that allows us to have different kind of KMS implementations, including legacy non-atomic and atomic. The intention is also that the code interacting with the drm device should be able to be run in a different thread than the main thread. This means that we need to make sure that all drm*() API usage must only occur from within tasks that eventually can be run in the dedicated thread. The idea here is that MetaKms provides a outward facing API other places of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is an internal implementation that only gets interacted with via "tasks" posted via the MetaKms object. These tasks will in the future potentially be run on the dedicated KMS thread. Initially, we don't create any new threads. Likewise, MetaKmsDevice is a outward facing representation of a KMS device, while MetaKmsImplDevice is the corresponding implementation, which only runs from within the MetaKmsImpl tasks. This commit only moves opening and closing the device to this new API, while leaking the fd outside of the impl enclosure, effectively making the isolation for drm*() calls pointless. This, however, is necessary to allow gradual porting of drm interaction, and eventually the file descriptor in MetaGpuKms will be removed. For now, it's harmless, since everything still run in the main thread. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-01-29 04:24:44 -05:00
#include "backends/native/meta-kms-device.h"
backend/native: Add and use transactional KMS API This commit introduces, and makes use of, a transactional API used for setting up KMS state, later to be applied, potentially atomically. From an API point of view, so is always the case, but in the current implementation, it still uses legacy drmMode* API to apply the state non-atomically. The API consists of various buliding blocks: * MetaKmsUpdate - a set of configuration changes, the higher level handle for handing over configuration to the impl backend. It's used to set mode, assign framebuffers to planes, queue page flips and set connector properties. * MetaKmsPlaneAssignment - the assignment of a framebuffer to a plane. Currently used to map a framebuffer to the primary plane of a CRTC. In the legacy KMS implementation, the plane assignment is used to derive the framebuffer used for mode setting and page flipping. This also means various high level changes: State, excluding configuring the cursor plane and creating/destroying DRM framebuffer handles, are applied in the end of a clutter frame, in one go. From an API point of view, this is done atomically, but as mentioned, only the non-atomic implementation exists so far. From MetaRendererNative's point of view, a page flip now initially always succeeds; the handling of EBUSY errors are done asynchronously in the MetaKmsImpl backend (still by retrying at refresh rate, but postponing flip callbacks instead of manipulating the frame clock). Handling of falling back to mode setting instead of page flipping is notified after the fact by a more precise page flip feedback API. EGLStream based page flipping relies on the impl backend not being atomic, as the page flipping is done in the EGLStream backend (e.g. nvidia driver). It uses a 'custom' page flip queueing method, keeping the EGLStream logic inside meta-renderer-native.c. Page flip handling is moved to meta-kms-impl-device.c from meta-gpu-kms.c. It goes via an extra idle callback before reaching meta-renderer-native.c to make sure callbacks are invoked outside of the impl context. While dummy power save page flipping is kept in meta-renderer-native.c, the EBUSY handling is moved to meta-kms-impl-simple.c. Instead of freezing the frame clock, actual page flip callbacks are postponed until all EBUSY retries have either succeeded or failed due to some other error than EBUSY. This effectively inhibits new frames to be drawn, meaning we won't stall waiting on the file descriptor for pending page flips. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-04-04 16:36:41 -04:00
#include "backends/native/meta-kms-update.h"
#include "backends/native/meta-kms-utils.h"
backends/native: Add basic KMS abstraction building blocks The intention with KMS abstraction is to hide away accessing the drm functions behind an API that allows us to have different kind of KMS implementations, including legacy non-atomic and atomic. The intention is also that the code interacting with the drm device should be able to be run in a different thread than the main thread. This means that we need to make sure that all drm*() API usage must only occur from within tasks that eventually can be run in the dedicated thread. The idea here is that MetaKms provides a outward facing API other places of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is an internal implementation that only gets interacted with via "tasks" posted via the MetaKms object. These tasks will in the future potentially be run on the dedicated KMS thread. Initially, we don't create any new threads. Likewise, MetaKmsDevice is a outward facing representation of a KMS device, while MetaKmsImplDevice is the corresponding implementation, which only runs from within the MetaKmsImpl tasks. This commit only moves opening and closing the device to this new API, while leaking the fd outside of the impl enclosure, effectively making the isolation for drm*() calls pointless. This, however, is necessary to allow gradual porting of drm interaction, and eventually the file descriptor in MetaGpuKms will be removed. For now, it's harmless, since everything still run in the main thread. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-01-29 04:24:44 -05:00
#include "backends/native/meta-kms.h"
#include "backends/native/meta-launcher.h"
#include "backends/native/meta-output-kms.h"
#include "meta-default-modes.h"
struct _MetaGpuKms
{
MetaGpu parent;
backends/native: Add basic KMS abstraction building blocks The intention with KMS abstraction is to hide away accessing the drm functions behind an API that allows us to have different kind of KMS implementations, including legacy non-atomic and atomic. The intention is also that the code interacting with the drm device should be able to be run in a different thread than the main thread. This means that we need to make sure that all drm*() API usage must only occur from within tasks that eventually can be run in the dedicated thread. The idea here is that MetaKms provides a outward facing API other places of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is an internal implementation that only gets interacted with via "tasks" posted via the MetaKms object. These tasks will in the future potentially be run on the dedicated KMS thread. Initially, we don't create any new threads. Likewise, MetaKmsDevice is a outward facing representation of a KMS device, while MetaKmsImplDevice is the corresponding implementation, which only runs from within the MetaKmsImpl tasks. This commit only moves opening and closing the device to this new API, while leaking the fd outside of the impl enclosure, effectively making the isolation for drm*() calls pointless. This, however, is necessary to allow gradual porting of drm interaction, and eventually the file descriptor in MetaGpuKms will be removed. For now, it's harmless, since everything still run in the main thread. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-01-29 04:24:44 -05:00
MetaKmsDevice *kms_device;
uint32_t id;
int fd;
clockid_t clock_id;
gboolean resources_init_failed_before;
};
G_DEFINE_TYPE (MetaGpuKms, meta_gpu_kms, META_TYPE_GPU)
gboolean
meta_gpu_kms_add_fb (MetaGpuKms *gpu_kms,
gboolean use_modifiers,
const MetaGpuKmsFBArgs *args,
uint32_t *fb_id_out,
GError **error)
{
MetaDrmFormatBuf tmp;
uint32_t fb_id;
if (use_modifiers && args->modifiers[0] != DRM_FORMAT_MOD_INVALID)
{
if (drmModeAddFB2WithModifiers (gpu_kms->fd,
args->width,
args->height,
args->format,
args->handles,
args->strides,
args->offsets,
args->modifiers,
&fb_id,
DRM_MODE_FB_MODIFIERS))
{
g_set_error (error,
G_IO_ERROR,
g_io_error_from_errno (errno),
"drmModeAddFB2WithModifiers failed: %s",
g_strerror (errno));
return FALSE;
}
}
else if (drmModeAddFB2 (gpu_kms->fd,
args->width,
args->height,
args->format,
args->handles,
args->strides,
args->offsets,
&fb_id,
0))
{
if (args->format != DRM_FORMAT_XRGB8888)
{
g_set_error (error,
G_IO_ERROR,
G_IO_ERROR_FAILED,
"drmModeAddFB does not support format '%s' (0x%x)",
meta_drm_format_to_string (&tmp, args->format),
args->format);
return FALSE;
}
if (drmModeAddFB (gpu_kms->fd,
args->width,
args->height,
24,
32,
args->strides[0],
args->handles[0],
&fb_id))
{
g_set_error (error,
G_IO_ERROR,
g_io_error_from_errno (errno),
"drmModeAddFB failed: %s",
g_strerror (errno));
return FALSE;
}
}
*fb_id_out = fb_id;
return TRUE;
}
gboolean
meta_gpu_kms_is_crtc_active (MetaGpuKms *gpu_kms,
MetaCrtc *crtc)
{
MetaGpu *gpu = META_GPU (gpu_kms);
MetaBackend *backend = meta_gpu_get_backend (gpu);
MetaMonitorManager *monitor_manager =
meta_backend_get_monitor_manager (backend);
GList *l;
gboolean connected_crtc_found;
g_assert (meta_crtc_get_gpu (crtc) == META_GPU (gpu_kms));
if (meta_monitor_manager_get_power_save_mode (monitor_manager))
return FALSE;
connected_crtc_found = FALSE;
for (l = meta_gpu_get_outputs (gpu); l; l = l->next)
{
MetaOutput *output = l->data;
MetaCrtc *assigned_crtc;
assigned_crtc = meta_output_get_assigned_crtc (output);
if (assigned_crtc == crtc)
{
connected_crtc_found = TRUE;
break;
}
}
if (!connected_crtc_found)
return FALSE;
return TRUE;
}
static int64_t
timespec_to_nanoseconds (const struct timespec *ts)
{
const int64_t one_billion = 1000000000;
return ((int64_t) ts->tv_sec) * one_billion + ts->tv_nsec;
}
gboolean
meta_gpu_kms_wait_for_flip (MetaGpuKms *gpu_kms,
GError **error)
{
backend/native: Add and use transactional KMS API This commit introduces, and makes use of, a transactional API used for setting up KMS state, later to be applied, potentially atomically. From an API point of view, so is always the case, but in the current implementation, it still uses legacy drmMode* API to apply the state non-atomically. The API consists of various buliding blocks: * MetaKmsUpdate - a set of configuration changes, the higher level handle for handing over configuration to the impl backend. It's used to set mode, assign framebuffers to planes, queue page flips and set connector properties. * MetaKmsPlaneAssignment - the assignment of a framebuffer to a plane. Currently used to map a framebuffer to the primary plane of a CRTC. In the legacy KMS implementation, the plane assignment is used to derive the framebuffer used for mode setting and page flipping. This also means various high level changes: State, excluding configuring the cursor plane and creating/destroying DRM framebuffer handles, are applied in the end of a clutter frame, in one go. From an API point of view, this is done atomically, but as mentioned, only the non-atomic implementation exists so far. From MetaRendererNative's point of view, a page flip now initially always succeeds; the handling of EBUSY errors are done asynchronously in the MetaKmsImpl backend (still by retrying at refresh rate, but postponing flip callbacks instead of manipulating the frame clock). Handling of falling back to mode setting instead of page flipping is notified after the fact by a more precise page flip feedback API. EGLStream based page flipping relies on the impl backend not being atomic, as the page flipping is done in the EGLStream backend (e.g. nvidia driver). It uses a 'custom' page flip queueing method, keeping the EGLStream logic inside meta-renderer-native.c. Page flip handling is moved to meta-kms-impl-device.c from meta-gpu-kms.c. It goes via an extra idle callback before reaching meta-renderer-native.c to make sure callbacks are invoked outside of the impl context. While dummy power save page flipping is kept in meta-renderer-native.c, the EBUSY handling is moved to meta-kms-impl-simple.c. Instead of freezing the frame clock, actual page flip callbacks are postponed until all EBUSY retries have either succeeded or failed due to some other error than EBUSY. This effectively inhibits new frames to be drawn, meaning we won't stall waiting on the file descriptor for pending page flips. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-04-04 16:36:41 -04:00
if (meta_kms_device_dispatch_sync (gpu_kms->kms_device, error) < 0)
return FALSE;
return TRUE;
}
MetaKmsDevice *
meta_gpu_kms_get_kms_device (MetaGpuKms *gpu_kms)
{
return gpu_kms->kms_device;
}
int
meta_gpu_kms_get_fd (MetaGpuKms *gpu_kms)
{
return gpu_kms->fd;
}
uint32_t
meta_gpu_kms_get_id (MetaGpuKms *gpu_kms)
{
return gpu_kms->id;
}
const char *
meta_gpu_kms_get_file_path (MetaGpuKms *gpu_kms)
{
backends/native: Add basic KMS abstraction building blocks The intention with KMS abstraction is to hide away accessing the drm functions behind an API that allows us to have different kind of KMS implementations, including legacy non-atomic and atomic. The intention is also that the code interacting with the drm device should be able to be run in a different thread than the main thread. This means that we need to make sure that all drm*() API usage must only occur from within tasks that eventually can be run in the dedicated thread. The idea here is that MetaKms provides a outward facing API other places of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is an internal implementation that only gets interacted with via "tasks" posted via the MetaKms object. These tasks will in the future potentially be run on the dedicated KMS thread. Initially, we don't create any new threads. Likewise, MetaKmsDevice is a outward facing representation of a KMS device, while MetaKmsImplDevice is the corresponding implementation, which only runs from within the MetaKmsImpl tasks. This commit only moves opening and closing the device to this new API, while leaking the fd outside of the impl enclosure, effectively making the isolation for drm*() calls pointless. This, however, is necessary to allow gradual porting of drm interaction, and eventually the file descriptor in MetaGpuKms will be removed. For now, it's harmless, since everything still run in the main thread. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-01-29 04:24:44 -05:00
return meta_kms_device_get_path (gpu_kms->kms_device);
}
int64_t
meta_gpu_kms_get_current_time_ns (MetaGpuKms *gpu_kms)
{
struct timespec ts;
if (clock_gettime (gpu_kms->clock_id, &ts))
return 0;
return timespec_to_nanoseconds (&ts);
}
void
backend/native: Add and use transactional KMS API This commit introduces, and makes use of, a transactional API used for setting up KMS state, later to be applied, potentially atomically. From an API point of view, so is always the case, but in the current implementation, it still uses legacy drmMode* API to apply the state non-atomically. The API consists of various buliding blocks: * MetaKmsUpdate - a set of configuration changes, the higher level handle for handing over configuration to the impl backend. It's used to set mode, assign framebuffers to planes, queue page flips and set connector properties. * MetaKmsPlaneAssignment - the assignment of a framebuffer to a plane. Currently used to map a framebuffer to the primary plane of a CRTC. In the legacy KMS implementation, the plane assignment is used to derive the framebuffer used for mode setting and page flipping. This also means various high level changes: State, excluding configuring the cursor plane and creating/destroying DRM framebuffer handles, are applied in the end of a clutter frame, in one go. From an API point of view, this is done atomically, but as mentioned, only the non-atomic implementation exists so far. From MetaRendererNative's point of view, a page flip now initially always succeeds; the handling of EBUSY errors are done asynchronously in the MetaKmsImpl backend (still by retrying at refresh rate, but postponing flip callbacks instead of manipulating the frame clock). Handling of falling back to mode setting instead of page flipping is notified after the fact by a more precise page flip feedback API. EGLStream based page flipping relies on the impl backend not being atomic, as the page flipping is done in the EGLStream backend (e.g. nvidia driver). It uses a 'custom' page flip queueing method, keeping the EGLStream logic inside meta-renderer-native.c. Page flip handling is moved to meta-kms-impl-device.c from meta-gpu-kms.c. It goes via an extra idle callback before reaching meta-renderer-native.c to make sure callbacks are invoked outside of the impl context. While dummy power save page flipping is kept in meta-renderer-native.c, the EBUSY handling is moved to meta-kms-impl-simple.c. Instead of freezing the frame clock, actual page flip callbacks are postponed until all EBUSY retries have either succeeded or failed due to some other error than EBUSY. This effectively inhibits new frames to be drawn, meaning we won't stall waiting on the file descriptor for pending page flips. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-04-04 16:36:41 -04:00
meta_gpu_kms_set_power_save_mode (MetaGpuKms *gpu_kms,
uint64_t state,
MetaKmsUpdate *kms_update)
{
GList *l;
for (l = meta_gpu_get_outputs (META_GPU (gpu_kms)); l; l = l->next)
{
MetaOutput *output = l->data;
backend/native: Add and use transactional KMS API This commit introduces, and makes use of, a transactional API used for setting up KMS state, later to be applied, potentially atomically. From an API point of view, so is always the case, but in the current implementation, it still uses legacy drmMode* API to apply the state non-atomically. The API consists of various buliding blocks: * MetaKmsUpdate - a set of configuration changes, the higher level handle for handing over configuration to the impl backend. It's used to set mode, assign framebuffers to planes, queue page flips and set connector properties. * MetaKmsPlaneAssignment - the assignment of a framebuffer to a plane. Currently used to map a framebuffer to the primary plane of a CRTC. In the legacy KMS implementation, the plane assignment is used to derive the framebuffer used for mode setting and page flipping. This also means various high level changes: State, excluding configuring the cursor plane and creating/destroying DRM framebuffer handles, are applied in the end of a clutter frame, in one go. From an API point of view, this is done atomically, but as mentioned, only the non-atomic implementation exists so far. From MetaRendererNative's point of view, a page flip now initially always succeeds; the handling of EBUSY errors are done asynchronously in the MetaKmsImpl backend (still by retrying at refresh rate, but postponing flip callbacks instead of manipulating the frame clock). Handling of falling back to mode setting instead of page flipping is notified after the fact by a more precise page flip feedback API. EGLStream based page flipping relies on the impl backend not being atomic, as the page flipping is done in the EGLStream backend (e.g. nvidia driver). It uses a 'custom' page flip queueing method, keeping the EGLStream logic inside meta-renderer-native.c. Page flip handling is moved to meta-kms-impl-device.c from meta-gpu-kms.c. It goes via an extra idle callback before reaching meta-renderer-native.c to make sure callbacks are invoked outside of the impl context. While dummy power save page flipping is kept in meta-renderer-native.c, the EBUSY handling is moved to meta-kms-impl-simple.c. Instead of freezing the frame clock, actual page flip callbacks are postponed until all EBUSY retries have either succeeded or failed due to some other error than EBUSY. This effectively inhibits new frames to be drawn, meaning we won't stall waiting on the file descriptor for pending page flips. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-04-04 16:36:41 -04:00
meta_output_kms_set_power_save_mode (output, state, kms_update);
}
}
gboolean
meta_gpu_kms_is_boot_vga (MetaGpuKms *gpu_kms)
{
backends/native: Add basic KMS abstraction building blocks The intention with KMS abstraction is to hide away accessing the drm functions behind an API that allows us to have different kind of KMS implementations, including legacy non-atomic and atomic. The intention is also that the code interacting with the drm device should be able to be run in a different thread than the main thread. This means that we need to make sure that all drm*() API usage must only occur from within tasks that eventually can be run in the dedicated thread. The idea here is that MetaKms provides a outward facing API other places of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is an internal implementation that only gets interacted with via "tasks" posted via the MetaKms object. These tasks will in the future potentially be run on the dedicated KMS thread. Initially, we don't create any new threads. Likewise, MetaKmsDevice is a outward facing representation of a KMS device, while MetaKmsImplDevice is the corresponding implementation, which only runs from within the MetaKmsImpl tasks. This commit only moves opening and closing the device to this new API, while leaking the fd outside of the impl enclosure, effectively making the isolation for drm*() calls pointless. This, however, is necessary to allow gradual porting of drm interaction, and eventually the file descriptor in MetaGpuKms will be removed. For now, it's harmless, since everything still run in the main thread. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-01-29 04:24:44 -05:00
MetaKmsDeviceFlag flags;
flags = meta_kms_device_get_flags (gpu_kms->kms_device);
return !!(flags & META_KMS_DEVICE_FLAG_BOOT_VGA);
}
gboolean
meta_gpu_kms_is_platform_device (MetaGpuKms *gpu_kms)
{
backends/native: Add basic KMS abstraction building blocks The intention with KMS abstraction is to hide away accessing the drm functions behind an API that allows us to have different kind of KMS implementations, including legacy non-atomic and atomic. The intention is also that the code interacting with the drm device should be able to be run in a different thread than the main thread. This means that we need to make sure that all drm*() API usage must only occur from within tasks that eventually can be run in the dedicated thread. The idea here is that MetaKms provides a outward facing API other places of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is an internal implementation that only gets interacted with via "tasks" posted via the MetaKms object. These tasks will in the future potentially be run on the dedicated KMS thread. Initially, we don't create any new threads. Likewise, MetaKmsDevice is a outward facing representation of a KMS device, while MetaKmsImplDevice is the corresponding implementation, which only runs from within the MetaKmsImpl tasks. This commit only moves opening and closing the device to this new API, while leaking the fd outside of the impl enclosure, effectively making the isolation for drm*() calls pointless. This, however, is necessary to allow gradual porting of drm interaction, and eventually the file descriptor in MetaGpuKms will be removed. For now, it's harmless, since everything still run in the main thread. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-01-29 04:24:44 -05:00
MetaKmsDeviceFlag flags;
flags = meta_kms_device_get_flags (gpu_kms->kms_device);
return !!(flags & META_KMS_DEVICE_FLAG_PLATFORM_DEVICE);
}
static int
compare_outputs (gconstpointer one,
gconstpointer two)
{
const MetaOutput *o_one = one, *o_two = two;
return strcmp (o_one->name, o_two->name);
}
static void
meta_crtc_mode_destroy_notify (MetaCrtcMode *mode)
{
g_slice_free (drmModeModeInfo, mode->driver_private);
}
gboolean
meta_drm_mode_equal (const drmModeModeInfo *one,
const drmModeModeInfo *two)
{
return (one->clock == two->clock &&
one->hdisplay == two->hdisplay &&
one->hsync_start == two->hsync_start &&
one->hsync_end == two->hsync_end &&
one->htotal == two->htotal &&
one->hskew == two->hskew &&
one->vdisplay == two->vdisplay &&
one->vsync_start == two->vsync_start &&
one->vsync_end == two->vsync_end &&
one->vtotal == two->vtotal &&
one->vscan == two->vscan &&
one->vrefresh == two->vrefresh &&
one->flags == two->flags &&
one->type == two->type &&
strncmp (one->name, two->name, DRM_DISPLAY_MODE_LEN) == 0);
}
static guint
drm_mode_hash (gconstpointer ptr)
{
const drmModeModeInfo *mode = ptr;
guint hash = 0;
/*
* We don't include the name in the hash because it's generally
* derived from the other fields (hdisplay, vdisplay and flags)
*/
hash ^= mode->clock;
hash ^= mode->hdisplay ^ mode->hsync_start ^ mode->hsync_end;
hash ^= mode->vdisplay ^ mode->vsync_start ^ mode->vsync_end;
hash ^= mode->vrefresh;
hash ^= mode->flags ^ mode->type;
return hash;
}
MetaCrtcMode *
meta_gpu_kms_get_mode_from_drm_mode (MetaGpuKms *gpu_kms,
const drmModeModeInfo *drm_mode)
{
MetaGpu *gpu = META_GPU (gpu_kms);
GList *l;
for (l = meta_gpu_get_modes (gpu); l; l = l->next)
{
MetaCrtcMode *mode = l->data;
if (meta_drm_mode_equal (drm_mode, mode->driver_private))
return mode;
}
g_assert_not_reached ();
return NULL;
}
static MetaCrtcMode *
create_mode (const drmModeModeInfo *drm_mode,
long mode_id)
{
MetaCrtcMode *mode;
mode = g_object_new (META_TYPE_CRTC_MODE, NULL);
mode->mode_id = mode_id;
mode->name = g_strndup (drm_mode->name, DRM_DISPLAY_MODE_LEN);
mode->width = drm_mode->hdisplay;
mode->height = drm_mode->vdisplay;
mode->flags = drm_mode->flags;
mode->refresh_rate = meta_calculate_drm_mode_refresh_rate (drm_mode);
mode->driver_private = g_slice_dup (drmModeModeInfo, drm_mode);
mode->driver_notify = (GDestroyNotify) meta_crtc_mode_destroy_notify;
return mode;
}
static MetaOutput *
find_output_by_connector_id (GList *outputs,
uint32_t connector_id)
{
GList *l;
for (l = outputs; l; l = l->next)
{
MetaOutput *output = l->data;
if (meta_output_kms_get_connector_id (output) == connector_id)
return output;
}
return NULL;
}
static void
setup_output_clones (MetaGpu *gpu)
{
GList *l;
for (l = meta_gpu_get_outputs (gpu); l; l = l->next)
{
MetaOutput *output = l->data;
GList *k;
for (k = meta_gpu_get_outputs (gpu); k; k = k->next)
{
MetaOutput *other_output = k->data;
if (other_output == output)
continue;
if (meta_output_kms_can_clone (output, other_output))
{
output->n_possible_clones++;
output->possible_clones = g_renew (MetaOutput *,
output->possible_clones,
output->n_possible_clones);
output->possible_clones[output->n_possible_clones - 1] =
other_output;
}
}
}
}
static void
init_modes (MetaGpuKms *gpu_kms)
{
MetaGpu *gpu = META_GPU (gpu_kms);
GHashTable *modes_table;
GList *l;
GList *modes;
GHashTableIter iter;
drmModeModeInfo *drm_mode;
int i;
long mode_id;
/*
* Gather all modes on all connected connectors.
*/
modes_table = g_hash_table_new (drm_mode_hash, (GEqualFunc) meta_drm_mode_equal);
for (l = meta_kms_device_get_connectors (gpu_kms->kms_device); l; l = l->next)
{
MetaKmsConnector *kms_connector = l->data;
const MetaKmsConnectorState *state;
state = meta_kms_connector_get_current_state (kms_connector);
if (!state)
continue;
for (i = 0; i < state->n_modes; i++)
g_hash_table_add (modes_table, &state->modes[i]);
}
modes = NULL;
g_hash_table_iter_init (&iter, modes_table);
mode_id = 0;
while (g_hash_table_iter_next (&iter, NULL, (gpointer *) &drm_mode))
{
MetaCrtcMode *mode;
mode = create_mode (drm_mode, (long) mode_id);
modes = g_list_append (modes, mode);
mode_id++;
}
g_hash_table_destroy (modes_table);
for (i = 0; i < G_N_ELEMENTS (meta_default_landscape_drm_mode_infos); i++)
{
MetaCrtcMode *mode;
mode = create_mode (&meta_default_landscape_drm_mode_infos[i], mode_id);
modes = g_list_append (modes, mode);
mode_id++;
}
for (i = 0; i < G_N_ELEMENTS (meta_default_portrait_drm_mode_infos); i++)
{
MetaCrtcMode *mode;
mode = create_mode (&meta_default_portrait_drm_mode_infos[i], mode_id);
modes = g_list_append (modes, mode);
mode_id++;
}
meta_gpu_take_modes (gpu, modes);
}
static void
init_crtcs (MetaGpuKms *gpu_kms)
{
MetaGpu *gpu = META_GPU (gpu_kms);
MetaKmsDevice *kms_device = gpu_kms->kms_device;
GList *l;
GList *crtcs;
crtcs = NULL;
for (l = meta_kms_device_get_crtcs (kms_device); l; l = l->next)
{
MetaKmsCrtc *kms_crtc = l->data;
MetaCrtc *crtc;
crtc = meta_create_kms_crtc (gpu_kms, kms_crtc);
crtcs = g_list_append (crtcs, crtc);
}
meta_gpu_take_crtcs (gpu, crtcs);
}
static void
init_frame_clock (MetaGpuKms *gpu_kms)
{
uint64_t uses_monotonic;
if (drmGetCap (gpu_kms->fd, DRM_CAP_TIMESTAMP_MONOTONIC, &uses_monotonic) != 0)
uses_monotonic = 0;
gpu_kms->clock_id = uses_monotonic ? CLOCK_MONOTONIC : CLOCK_REALTIME;
}
static void
init_outputs (MetaGpuKms *gpu_kms)
{
MetaGpu *gpu = META_GPU (gpu_kms);
GList *old_outputs;
GList *outputs;
GList *l;
old_outputs = meta_gpu_get_outputs (gpu);
outputs = NULL;
for (l = meta_kms_device_get_connectors (gpu_kms->kms_device); l; l = l->next)
{
MetaKmsConnector *kms_connector = l->data;
backend/native: Add and use transactional KMS API This commit introduces, and makes use of, a transactional API used for setting up KMS state, later to be applied, potentially atomically. From an API point of view, so is always the case, but in the current implementation, it still uses legacy drmMode* API to apply the state non-atomically. The API consists of various buliding blocks: * MetaKmsUpdate - a set of configuration changes, the higher level handle for handing over configuration to the impl backend. It's used to set mode, assign framebuffers to planes, queue page flips and set connector properties. * MetaKmsPlaneAssignment - the assignment of a framebuffer to a plane. Currently used to map a framebuffer to the primary plane of a CRTC. In the legacy KMS implementation, the plane assignment is used to derive the framebuffer used for mode setting and page flipping. This also means various high level changes: State, excluding configuring the cursor plane and creating/destroying DRM framebuffer handles, are applied in the end of a clutter frame, in one go. From an API point of view, this is done atomically, but as mentioned, only the non-atomic implementation exists so far. From MetaRendererNative's point of view, a page flip now initially always succeeds; the handling of EBUSY errors are done asynchronously in the MetaKmsImpl backend (still by retrying at refresh rate, but postponing flip callbacks instead of manipulating the frame clock). Handling of falling back to mode setting instead of page flipping is notified after the fact by a more precise page flip feedback API. EGLStream based page flipping relies on the impl backend not being atomic, as the page flipping is done in the EGLStream backend (e.g. nvidia driver). It uses a 'custom' page flip queueing method, keeping the EGLStream logic inside meta-renderer-native.c. Page flip handling is moved to meta-kms-impl-device.c from meta-gpu-kms.c. It goes via an extra idle callback before reaching meta-renderer-native.c to make sure callbacks are invoked outside of the impl context. While dummy power save page flipping is kept in meta-renderer-native.c, the EBUSY handling is moved to meta-kms-impl-simple.c. Instead of freezing the frame clock, actual page flip callbacks are postponed until all EBUSY retries have either succeeded or failed due to some other error than EBUSY. This effectively inhibits new frames to be drawn, meaning we won't stall waiting on the file descriptor for pending page flips. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-04-04 16:36:41 -04:00
const MetaKmsConnectorState *connector_state;
MetaOutput *output;
MetaOutput *old_output;
GError *error = NULL;
backend/native: Add and use transactional KMS API This commit introduces, and makes use of, a transactional API used for setting up KMS state, later to be applied, potentially atomically. From an API point of view, so is always the case, but in the current implementation, it still uses legacy drmMode* API to apply the state non-atomically. The API consists of various buliding blocks: * MetaKmsUpdate - a set of configuration changes, the higher level handle for handing over configuration to the impl backend. It's used to set mode, assign framebuffers to planes, queue page flips and set connector properties. * MetaKmsPlaneAssignment - the assignment of a framebuffer to a plane. Currently used to map a framebuffer to the primary plane of a CRTC. In the legacy KMS implementation, the plane assignment is used to derive the framebuffer used for mode setting and page flipping. This also means various high level changes: State, excluding configuring the cursor plane and creating/destroying DRM framebuffer handles, are applied in the end of a clutter frame, in one go. From an API point of view, this is done atomically, but as mentioned, only the non-atomic implementation exists so far. From MetaRendererNative's point of view, a page flip now initially always succeeds; the handling of EBUSY errors are done asynchronously in the MetaKmsImpl backend (still by retrying at refresh rate, but postponing flip callbacks instead of manipulating the frame clock). Handling of falling back to mode setting instead of page flipping is notified after the fact by a more precise page flip feedback API. EGLStream based page flipping relies on the impl backend not being atomic, as the page flipping is done in the EGLStream backend (e.g. nvidia driver). It uses a 'custom' page flip queueing method, keeping the EGLStream logic inside meta-renderer-native.c. Page flip handling is moved to meta-kms-impl-device.c from meta-gpu-kms.c. It goes via an extra idle callback before reaching meta-renderer-native.c to make sure callbacks are invoked outside of the impl context. While dummy power save page flipping is kept in meta-renderer-native.c, the EBUSY handling is moved to meta-kms-impl-simple.c. Instead of freezing the frame clock, actual page flip callbacks are postponed until all EBUSY retries have either succeeded or failed due to some other error than EBUSY. This effectively inhibits new frames to be drawn, meaning we won't stall waiting on the file descriptor for pending page flips. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-04-04 16:36:41 -04:00
connector_state = meta_kms_connector_get_current_state (kms_connector);
if (!connector_state)
continue;
old_output =
find_output_by_connector_id (old_outputs,
meta_kms_connector_get_id (kms_connector));
output = meta_create_kms_output (gpu_kms,
kms_connector,
old_output,
&error);
if (!output)
{
g_warning ("Failed to create KMS output: %s", error->message);
g_error_free (error);
}
else
{
outputs = g_list_prepend (outputs, output);
}
}
/* Sort the outputs for easier handling in MetaMonitorConfig */
outputs = g_list_sort (outputs, compare_outputs);
meta_gpu_take_outputs (gpu, outputs);
setup_output_clones (gpu);
}
static gboolean
meta_gpu_kms_read_current (MetaGpu *gpu,
GError **error)
{
MetaGpuKms *gpu_kms = META_GPU_KMS (gpu);
/* Note: we must not free the public structures (output, crtc, monitor
mode and monitor info) here, they must be kept alive until the API
users are done with them after we emit monitors-changed, and thus
are freed by the platform-independent layer. */
init_modes (gpu_kms);
init_crtcs (gpu_kms);
init_outputs (gpu_kms);
init_frame_clock (gpu_kms);
return TRUE;
}
gboolean
meta_gpu_kms_can_have_outputs (MetaGpuKms *gpu_kms)
{
GList *l;
int n_connected_connectors = 0;
for (l = meta_kms_device_get_connectors (gpu_kms->kms_device); l; l = l->next)
{
MetaKmsConnector *kms_connector = l->data;
if (meta_kms_connector_get_current_state (kms_connector))
n_connected_connectors++;
}
return n_connected_connectors > 0;
}
MetaGpuKms *
meta_gpu_kms_new (MetaBackendNative *backend_native,
backends/native: Add basic KMS abstraction building blocks The intention with KMS abstraction is to hide away accessing the drm functions behind an API that allows us to have different kind of KMS implementations, including legacy non-atomic and atomic. The intention is also that the code interacting with the drm device should be able to be run in a different thread than the main thread. This means that we need to make sure that all drm*() API usage must only occur from within tasks that eventually can be run in the dedicated thread. The idea here is that MetaKms provides a outward facing API other places of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is an internal implementation that only gets interacted with via "tasks" posted via the MetaKms object. These tasks will in the future potentially be run on the dedicated KMS thread. Initially, we don't create any new threads. Likewise, MetaKmsDevice is a outward facing representation of a KMS device, while MetaKmsImplDevice is the corresponding implementation, which only runs from within the MetaKmsImpl tasks. This commit only moves opening and closing the device to this new API, while leaking the fd outside of the impl enclosure, effectively making the isolation for drm*() calls pointless. This, however, is necessary to allow gradual porting of drm interaction, and eventually the file descriptor in MetaGpuKms will be removed. For now, it's harmless, since everything still run in the main thread. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-01-29 04:24:44 -05:00
MetaKmsDevice *kms_device,
GError **error)
{
MetaGpuKms *gpu_kms;
int kms_fd;
backends/native: Add basic KMS abstraction building blocks The intention with KMS abstraction is to hide away accessing the drm functions behind an API that allows us to have different kind of KMS implementations, including legacy non-atomic and atomic. The intention is also that the code interacting with the drm device should be able to be run in a different thread than the main thread. This means that we need to make sure that all drm*() API usage must only occur from within tasks that eventually can be run in the dedicated thread. The idea here is that MetaKms provides a outward facing API other places of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is an internal implementation that only gets interacted with via "tasks" posted via the MetaKms object. These tasks will in the future potentially be run on the dedicated KMS thread. Initially, we don't create any new threads. Likewise, MetaKmsDevice is a outward facing representation of a KMS device, while MetaKmsImplDevice is the corresponding implementation, which only runs from within the MetaKmsImpl tasks. This commit only moves opening and closing the device to this new API, while leaking the fd outside of the impl enclosure, effectively making the isolation for drm*() calls pointless. This, however, is necessary to allow gradual porting of drm interaction, and eventually the file descriptor in MetaGpuKms will be removed. For now, it's harmless, since everything still run in the main thread. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-01-29 04:24:44 -05:00
kms_fd = meta_kms_device_leak_fd (kms_device);
gpu_kms = g_object_new (META_TYPE_GPU_KMS,
"backend", backend_native,
NULL);
backends/native: Add basic KMS abstraction building blocks The intention with KMS abstraction is to hide away accessing the drm functions behind an API that allows us to have different kind of KMS implementations, including legacy non-atomic and atomic. The intention is also that the code interacting with the drm device should be able to be run in a different thread than the main thread. This means that we need to make sure that all drm*() API usage must only occur from within tasks that eventually can be run in the dedicated thread. The idea here is that MetaKms provides a outward facing API other places of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is an internal implementation that only gets interacted with via "tasks" posted via the MetaKms object. These tasks will in the future potentially be run on the dedicated KMS thread. Initially, we don't create any new threads. Likewise, MetaKmsDevice is a outward facing representation of a KMS device, while MetaKmsImplDevice is the corresponding implementation, which only runs from within the MetaKmsImpl tasks. This commit only moves opening and closing the device to this new API, while leaking the fd outside of the impl enclosure, effectively making the isolation for drm*() calls pointless. This, however, is necessary to allow gradual porting of drm interaction, and eventually the file descriptor in MetaGpuKms will be removed. For now, it's harmless, since everything still run in the main thread. https://gitlab.gnome.org/GNOME/mutter/issues/548 https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
2019-01-29 04:24:44 -05:00
gpu_kms->kms_device = kms_device;
gpu_kms->fd = kms_fd;
meta_gpu_kms_read_current (META_GPU (gpu_kms), NULL);
return gpu_kms;
}
static void
meta_gpu_kms_init (MetaGpuKms *gpu_kms)
{
static uint32_t id = 0;
gpu_kms->fd = -1;
gpu_kms->id = ++id;
}
static void
meta_gpu_kms_class_init (MetaGpuKmsClass *klass)
{
MetaGpuClass *gpu_class = META_GPU_CLASS (klass);
gpu_class->read_current = meta_gpu_kms_read_current;
}