mesa/src/vulkan/wsi/wsi_common_display.c

2463 lines
76 KiB
C
Raw Normal View History

vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
/*
* Copyright © 2017 Keith Packard
*
* Permission to use, copy, modify, distribute, and sell this software and its
* documentation for any purpose is hereby granted without fee, provided that
* the above copyright notice appear in all copies and that both that copyright
* notice and this permission notice appear in supporting documentation, and
* that the name of the copyright holders not be used in advertising or
* publicity pertaining to distribution of the software without specific,
* written prior permission. The copyright holders make no representations
* about the suitability of this software for any purpose. It is provided "as
* is" without express or implied warranty.
*
* THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
* INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
* EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY SPECIAL, INDIRECT OR
* CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
* DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
* TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE
* OF THIS SOFTWARE.
*/
#include "util/macros.h"
#include <stdlib.h>
#include <stdio.h>
#include <unistd.h>
#include <errno.h>
#include <string.h>
#include <fcntl.h>
#include <poll.h>
#include <stdbool.h>
#include <math.h>
#include <xf86drm.h>
#include <xf86drmMode.h>
#include <drm_fourcc.h>
vulkan: Add EXT_acquire_xlib_display [v5] This extension adds the ability to borrow an X RandR output for temporary use directly by a Vulkan application. For DRM, we use the Linux resource leasing mechanism. v2: Clean up xlib_lease detection * Use separate temporary '_xlib_lease' variable to hold the option value to avoid changin the type of a variable. * Use boolean expressions instead of additional if statements to compute resulting with_xlib_lease value. * Simplify addition of VK_USE_PLATFORM_XLIB_XRANDR_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> Move mode list from wsi_display to wsi_display_connector Fix scope for wsi_display_mode and wsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v3: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Explicitly forbid multiple DRM leases. Making the code support this looks tricky and will require additional thought. Use xcb_randr_output_t throughout the internals of the implementation. Convert at the public API (wsi_get_randr_output_display). Clean up check for usable active_crtc (possible when only the desired output is connected to the crtc). Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v4: Move output resource fetching closer to use in wsi_display_get_output. This simplifies the error returns in earlier parts of the code a bit. Return VK_ERROR_INITIALIZATION_FAILED from wsi_acquire_xlib_display. Jason says this is the right error message. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v5: randr doesn't pass vscan over the wire, so we set vscan to 0 for randr-acquired modes, and test wsi modes for vscan <= 1 when comparing against randr modes. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
2018-02-09 15:45:58 +00:00
#ifdef VK_USE_PLATFORM_XLIB_XRANDR_EXT
#include <xcb/randr.h>
#include <X11/Xlib-xcb.h>
#endif
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
#include "util/hash_table.h"
#include "util/list.h"
#include "vk_util.h"
#include "wsi_common_private.h"
#include "wsi_common_display.h"
#include "wsi_common_queue.h"
#if 0
#define wsi_display_debug(...) fprintf(stderr, __VA_ARGS__)
#define wsi_display_debug_code(...) __VA_ARGS__
#else
#define wsi_display_debug(...)
#define wsi_display_debug_code(...)
#endif
/* These have lifetime equal to the instance, so they effectively
* never go away. This means we must keep track of them separately
* from all other resources.
*/
typedef struct wsi_display_mode {
struct list_head list;
struct wsi_display_connector *connector;
bool valid; /* was found in most recent poll */
bool preferred;
uint32_t clock; /* in kHz */
uint16_t hdisplay, hsync_start, hsync_end, htotal, hskew;
uint16_t vdisplay, vsync_start, vsync_end, vtotal, vscan;
uint32_t flags;
} wsi_display_mode;
typedef struct wsi_display_connector {
struct list_head list;
struct wsi_display *wsi;
uint32_t id;
uint32_t crtc_id;
char *name;
bool connected;
bool active;
struct list_head display_modes;
wsi_display_mode *current_mode;
drmModeModeInfo current_drm_mode;
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
uint32_t dpms_property;
vulkan: Add EXT_acquire_xlib_display [v5] This extension adds the ability to borrow an X RandR output for temporary use directly by a Vulkan application. For DRM, we use the Linux resource leasing mechanism. v2: Clean up xlib_lease detection * Use separate temporary '_xlib_lease' variable to hold the option value to avoid changin the type of a variable. * Use boolean expressions instead of additional if statements to compute resulting with_xlib_lease value. * Simplify addition of VK_USE_PLATFORM_XLIB_XRANDR_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> Move mode list from wsi_display to wsi_display_connector Fix scope for wsi_display_mode and wsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v3: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Explicitly forbid multiple DRM leases. Making the code support this looks tricky and will require additional thought. Use xcb_randr_output_t throughout the internals of the implementation. Convert at the public API (wsi_get_randr_output_display). Clean up check for usable active_crtc (possible when only the desired output is connected to the crtc). Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v4: Move output resource fetching closer to use in wsi_display_get_output. This simplifies the error returns in earlier parts of the code a bit. Return VK_ERROR_INITIALIZATION_FAILED from wsi_acquire_xlib_display. Jason says this is the right error message. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v5: randr doesn't pass vscan over the wire, so we set vscan to 0 for randr-acquired modes, and test wsi modes for vscan <= 1 when comparing against randr modes. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
2018-02-09 15:45:58 +00:00
#ifdef VK_USE_PLATFORM_XLIB_XRANDR_EXT
xcb_randr_output_t output;
#endif
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
} wsi_display_connector;
struct wsi_display {
struct wsi_interface base;
const VkAllocationCallbacks *alloc;
int fd;
pthread_mutex_t wait_mutex;
pthread_cond_t wait_cond;
pthread_t wait_thread;
struct list_head connectors;
};
#define wsi_for_each_display_mode(_mode, _conn) \
list_for_each_entry_safe(struct wsi_display_mode, _mode, \
&(_conn)->display_modes, list)
#define wsi_for_each_connector(_conn, _dev) \
list_for_each_entry_safe(struct wsi_display_connector, _conn, \
&(_dev)->connectors, list)
enum wsi_image_state {
WSI_IMAGE_IDLE,
WSI_IMAGE_DRAWING,
WSI_IMAGE_QUEUED,
WSI_IMAGE_FLIPPING,
WSI_IMAGE_DISPLAYING
};
struct wsi_display_image {
struct wsi_image base;
struct wsi_display_swapchain *chain;
enum wsi_image_state state;
uint32_t fb_id;
uint32_t buffer[4];
uint64_t flip_sequence;
};
struct wsi_display_swapchain {
struct wsi_swapchain base;
struct wsi_display *wsi;
VkIcdSurfaceDisplay *surface;
uint64_t flip_sequence;
VkResult status;
struct wsi_display_image images[0];
};
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
struct wsi_display_fence {
struct wsi_fence base;
bool event_received;
bool destroyed;
uint64_t sequence;
};
static uint64_t fence_sequence;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
ICD_DEFINE_NONDISP_HANDLE_CASTS(wsi_display_mode, VkDisplayModeKHR)
ICD_DEFINE_NONDISP_HANDLE_CASTS(wsi_display_connector, VkDisplayKHR)
static bool
wsi_display_mode_matches_drm(wsi_display_mode *wsi,
drmModeModeInfoPtr drm)
{
return wsi->clock == drm->clock &&
wsi->hdisplay == drm->hdisplay &&
wsi->hsync_start == drm->hsync_start &&
wsi->hsync_end == drm->hsync_end &&
wsi->htotal == drm->htotal &&
wsi->hskew == drm->hskew &&
wsi->vdisplay == drm->vdisplay &&
wsi->vsync_start == drm->vsync_start &&
wsi->vsync_end == drm->vsync_end &&
wsi->vtotal == drm->vtotal &&
MAX2(wsi->vscan, 1) == MAX2(drm->vscan, 1) &&
wsi->flags == drm->flags;
}
static double
wsi_display_mode_refresh(struct wsi_display_mode *wsi)
{
return (double) wsi->clock * 1000.0 / ((double) wsi->htotal *
(double) wsi->vtotal *
(double) MAX2(wsi->vscan, 1));
}
static uint64_t wsi_get_current_monotonic(void)
{
struct timespec tv;
clock_gettime(CLOCK_MONOTONIC, &tv);
return tv.tv_nsec + tv.tv_sec*1000000000ull;
}
static uint64_t wsi_rel_to_abs_time(uint64_t rel_time)
{
uint64_t current_time = wsi_get_current_monotonic();
/* check for overflow */
if (rel_time > UINT64_MAX - current_time)
return UINT64_MAX;
return current_time + rel_time;
}
static struct wsi_display_mode *
wsi_display_find_drm_mode(struct wsi_device *wsi_device,
struct wsi_display_connector *connector,
drmModeModeInfoPtr mode)
{
wsi_for_each_display_mode(display_mode, connector) {
if (wsi_display_mode_matches_drm(display_mode, mode))
return display_mode;
}
return NULL;
}
static void
wsi_display_invalidate_connector_modes(struct wsi_device *wsi_device,
struct wsi_display_connector *connector)
{
wsi_for_each_display_mode(display_mode, connector) {
display_mode->valid = false;
}
}
static VkResult
wsi_display_register_drm_mode(struct wsi_device *wsi_device,
struct wsi_display_connector *connector,
drmModeModeInfoPtr drm_mode)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
struct wsi_display_mode *display_mode =
wsi_display_find_drm_mode(wsi_device, connector, drm_mode);
if (display_mode) {
display_mode->valid = true;
return VK_SUCCESS;
}
display_mode = vk_zalloc(wsi->alloc, sizeof (struct wsi_display_mode),
8, VK_SYSTEM_ALLOCATION_SCOPE_INSTANCE);
if (!display_mode)
return VK_ERROR_OUT_OF_HOST_MEMORY;
display_mode->connector = connector;
display_mode->valid = true;
display_mode->preferred = (drm_mode->type & DRM_MODE_TYPE_PREFERRED) != 0;
display_mode->clock = drm_mode->clock; /* kHz */
display_mode->hdisplay = drm_mode->hdisplay;
display_mode->hsync_start = drm_mode->hsync_start;
display_mode->hsync_end = drm_mode->hsync_end;
display_mode->htotal = drm_mode->htotal;
display_mode->hskew = drm_mode->hskew;
display_mode->vdisplay = drm_mode->vdisplay;
display_mode->vsync_start = drm_mode->vsync_start;
display_mode->vsync_end = drm_mode->vsync_end;
display_mode->vtotal = drm_mode->vtotal;
display_mode->vscan = drm_mode->vscan;
display_mode->flags = drm_mode->flags;
list_addtail(&display_mode->list, &connector->display_modes);
return VK_SUCCESS;
}
/*
* Update our information about a specific connector
*/
static struct wsi_display_connector *
wsi_display_find_connector(struct wsi_device *wsi_device,
uint32_t connector_id)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
wsi_for_each_connector(connector, wsi) {
if (connector->id == connector_id)
return connector;
}
return NULL;
}
static struct wsi_display_connector *
wsi_display_alloc_connector(struct wsi_display *wsi,
uint32_t connector_id)
{
struct wsi_display_connector *connector =
vk_zalloc(wsi->alloc, sizeof (struct wsi_display_connector),
8, VK_SYSTEM_ALLOCATION_SCOPE_INSTANCE);
connector->id = connector_id;
connector->wsi = wsi;
connector->active = false;
/* XXX use EDID name */
connector->name = "monitor";
list_inithead(&connector->display_modes);
return connector;
}
static struct wsi_display_connector *
wsi_display_get_connector(struct wsi_device *wsi_device,
uint32_t connector_id)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
if (wsi->fd < 0)
return NULL;
drmModeConnectorPtr drm_connector =
drmModeGetConnector(wsi->fd, connector_id);
if (!drm_connector)
return NULL;
struct wsi_display_connector *connector =
wsi_display_find_connector(wsi_device, connector_id);
if (!connector) {
connector = wsi_display_alloc_connector(wsi, connector_id);
if (!connector) {
drmModeFreeConnector(drm_connector);
return NULL;
}
list_addtail(&connector->list, &wsi->connectors);
}
connector->connected = drm_connector->connection != DRM_MODE_DISCONNECTED;
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
/* Look for a DPMS property if we haven't already found one */
for (int p = 0; connector->dpms_property == 0 &&
p < drm_connector->count_props; p++)
{
drmModePropertyPtr prop = drmModeGetProperty(wsi->fd,
drm_connector->props[p]);
if (!prop)
continue;
if (prop->flags & DRM_MODE_PROP_ENUM) {
if (!strcmp(prop->name, "DPMS"))
connector->dpms_property = drm_connector->props[p];
}
drmModeFreeProperty(prop);
}
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
/* Mark all connector modes as invalid */
wsi_display_invalidate_connector_modes(wsi_device, connector);
/*
* List current modes, adding new ones and marking existing ones as
* valid
*/
for (int m = 0; m < drm_connector->count_modes; m++) {
VkResult result = wsi_display_register_drm_mode(wsi_device,
connector,
&drm_connector->modes[m]);
if (result != VK_SUCCESS) {
drmModeFreeConnector(drm_connector);
return NULL;
}
}
drmModeFreeConnector(drm_connector);
return connector;
}
#define MM_PER_PIXEL (1.0/96.0 * 25.4)
static uint32_t
mode_size(struct wsi_display_mode *mode)
{
/* fortunately, these are both uint16_t, so this is easy */
return (uint32_t) mode->hdisplay * (uint32_t) mode->vdisplay;
}
static void
wsi_display_fill_in_display_properties(struct wsi_device *wsi_device,
struct wsi_display_connector *connector,
VkDisplayProperties2KHR *properties2)
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
{
assert(properties2->sType == VK_STRUCTURE_TYPE_DISPLAY_PROPERTIES_2_KHR);
VkDisplayPropertiesKHR *properties = &properties2->displayProperties;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
properties->display = wsi_display_connector_to_handle(connector);
properties->displayName = connector->name;
/* Find the first preferred mode and assume that's the physical
* resolution. If there isn't a preferred mode, find the largest mode and
* use that.
*/
struct wsi_display_mode *preferred_mode = NULL, *largest_mode = NULL;
wsi_for_each_display_mode(display_mode, connector) {
if (!display_mode->valid)
continue;
if (display_mode->preferred) {
preferred_mode = display_mode;
break;
}
if (largest_mode == NULL ||
mode_size(display_mode) > mode_size(largest_mode))
{
largest_mode = display_mode;
}
}
if (preferred_mode) {
properties->physicalResolution.width = preferred_mode->hdisplay;
properties->physicalResolution.height = preferred_mode->vdisplay;
} else if (largest_mode) {
properties->physicalResolution.width = largest_mode->hdisplay;
properties->physicalResolution.height = largest_mode->vdisplay;
} else {
properties->physicalResolution.width = 1024;
properties->physicalResolution.height = 768;
}
/* Make up physical size based on 96dpi */
properties->physicalDimensions.width =
floor(properties->physicalResolution.width * MM_PER_PIXEL + 0.5);
properties->physicalDimensions.height =
floor(properties->physicalResolution.height * MM_PER_PIXEL + 0.5);
properties->supportedTransforms = VK_SURFACE_TRANSFORM_IDENTITY_BIT_KHR;
properties->planeReorderPossible = VK_FALSE;
properties->persistentContent = VK_FALSE;
}
/*
* Implement vkGetPhysicalDeviceDisplayPropertiesKHR (VK_KHR_display)
*/
VkResult
wsi_display_get_physical_device_display_properties(
VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
uint32_t *property_count,
VkDisplayPropertiesKHR *properties)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
if (properties == NULL) {
return wsi_display_get_physical_device_display_properties2(
physical_device, wsi_device, property_count, NULL);
} else {
/* If we're actually returning properties, allocate a temporary array of
* VkDisplayProperties2KHR structs, call properties2 to fill them out,
* and then copy them to the client. This seems a bit expensive but
* wsi_display_get_physical_device_display_properties2() calls
* drmModeGetResources() which does an ioctl and then a bunch of
* allocations so this should get lost in the noise.
*/
VkDisplayProperties2KHR *props2 =
vk_zalloc(wsi->alloc, sizeof(*props2) * *property_count, 8,
VK_SYSTEM_ALLOCATION_SCOPE_OBJECT);
if (props2 == NULL)
return VK_ERROR_OUT_OF_HOST_MEMORY;
for (uint32_t i = 0; i < *property_count; i++)
props2[i].sType = VK_STRUCTURE_TYPE_DISPLAY_PROPERTIES_2_KHR;
VkResult result = wsi_display_get_physical_device_display_properties2(
physical_device, wsi_device, property_count, props2);
if (result == VK_SUCCESS || result == VK_INCOMPLETE) {
for (uint32_t i = 0; i < *property_count; i++)
properties[i] = props2[i].displayProperties;
}
vk_free(wsi->alloc, props2);
return result;
}
}
VkResult
wsi_display_get_physical_device_display_properties2(
VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
uint32_t *property_count,
VkDisplayProperties2KHR *properties)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
if (wsi->fd < 0)
goto bail;
drmModeResPtr mode_res = drmModeGetResources(wsi->fd);
if (!mode_res)
goto bail;
VK_OUTARRAY_MAKE(conn, properties, property_count);
/* Get current information */
for (int c = 0; c < mode_res->count_connectors; c++) {
struct wsi_display_connector *connector =
wsi_display_get_connector(wsi_device, mode_res->connectors[c]);
if (!connector) {
drmModeFreeResources(mode_res);
return VK_ERROR_OUT_OF_HOST_MEMORY;
}
if (connector->connected) {
vk_outarray_append(&conn, prop) {
wsi_display_fill_in_display_properties(wsi_device,
connector,
prop);
}
}
}
drmModeFreeResources(mode_res);
return vk_outarray_status(&conn);
bail:
*property_count = 0;
return VK_SUCCESS;
}
/*
* Implement vkGetPhysicalDeviceDisplayPlanePropertiesKHR (VK_KHR_display
*/
static void
wsi_display_fill_in_display_plane_properties(
struct wsi_device *wsi_device,
struct wsi_display_connector *connector,
VkDisplayPlaneProperties2KHR *properties)
{
assert(properties->sType == VK_STRUCTURE_TYPE_DISPLAY_PLANE_PROPERTIES_2_KHR);
VkDisplayPlanePropertiesKHR *prop = &properties->displayPlaneProperties;
if (connector && connector->active) {
prop->currentDisplay = wsi_display_connector_to_handle(connector);
prop->currentStackIndex = 0;
} else {
prop->currentDisplay = VK_NULL_HANDLE;
prop->currentStackIndex = 0;
}
}
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
VkResult
wsi_display_get_physical_device_display_plane_properties(
VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
uint32_t *property_count,
VkDisplayPlanePropertiesKHR *properties)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
VK_OUTARRAY_MAKE(conn, properties, property_count);
wsi_for_each_connector(connector, wsi) {
vk_outarray_append(&conn, prop) {
VkDisplayPlaneProperties2KHR prop2 = {
.sType = VK_STRUCTURE_TYPE_DISPLAY_PLANE_PROPERTIES_2_KHR,
};
wsi_display_fill_in_display_plane_properties(wsi_device, connector,
&prop2);
*prop = prop2.displayPlaneProperties;
}
}
return vk_outarray_status(&conn);
}
VkResult
wsi_display_get_physical_device_display_plane_properties2(
VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
uint32_t *property_count,
VkDisplayPlaneProperties2KHR *properties)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
VK_OUTARRAY_MAKE(conn, properties, property_count);
wsi_for_each_connector(connector, wsi) {
vk_outarray_append(&conn, prop) {
wsi_display_fill_in_display_plane_properties(wsi_device, connector,
prop);
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
}
}
return vk_outarray_status(&conn);
}
/*
* Implement vkGetDisplayPlaneSupportedDisplaysKHR (VK_KHR_display)
*/
VkResult
wsi_display_get_display_plane_supported_displays(
VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
uint32_t plane_index,
uint32_t *display_count,
VkDisplayKHR *displays)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
VK_OUTARRAY_MAKE(conn, displays, display_count);
int c = 0;
wsi_for_each_connector(connector, wsi) {
if (c == plane_index && connector->connected) {
vk_outarray_append(&conn, display) {
*display = wsi_display_connector_to_handle(connector);
}
}
c++;
}
return vk_outarray_status(&conn);
}
/*
* Implement vkGetDisplayModePropertiesKHR (VK_KHR_display)
*/
static void
wsi_display_fill_in_display_mode_properties(
struct wsi_device *wsi_device,
struct wsi_display_mode *display_mode,
VkDisplayModeProperties2KHR *properties)
{
assert(properties->sType == VK_STRUCTURE_TYPE_DISPLAY_MODE_PROPERTIES_2_KHR);
VkDisplayModePropertiesKHR *prop = &properties->displayModeProperties;
prop->displayMode = wsi_display_mode_to_handle(display_mode);
prop->parameters.visibleRegion.width = display_mode->hdisplay;
prop->parameters.visibleRegion.height = display_mode->vdisplay;
prop->parameters.refreshRate =
(uint32_t) (wsi_display_mode_refresh(display_mode) * 1000 + 0.5);
}
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
VkResult
wsi_display_get_display_mode_properties(VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
VkDisplayKHR display,
uint32_t *property_count,
VkDisplayModePropertiesKHR *properties)
{
struct wsi_display_connector *connector =
wsi_display_connector_from_handle(display);
VK_OUTARRAY_MAKE(conn, properties, property_count);
wsi_for_each_display_mode(display_mode, connector) {
if (!display_mode->valid)
continue;
vk_outarray_append(&conn, prop) {
VkDisplayModeProperties2KHR prop2 = {
.sType = VK_STRUCTURE_TYPE_DISPLAY_MODE_PROPERTIES_2_KHR,
};
wsi_display_fill_in_display_mode_properties(wsi_device,
display_mode, &prop2);
*prop = prop2.displayModeProperties;
}
}
return vk_outarray_status(&conn);
}
VkResult
wsi_display_get_display_mode_properties2(VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
VkDisplayKHR display,
uint32_t *property_count,
VkDisplayModeProperties2KHR *properties)
{
struct wsi_display_connector *connector =
wsi_display_connector_from_handle(display);
VK_OUTARRAY_MAKE(conn, properties, property_count);
wsi_for_each_display_mode(display_mode, connector) {
if (!display_mode->valid)
continue;
vk_outarray_append(&conn, prop) {
wsi_display_fill_in_display_mode_properties(wsi_device,
display_mode, prop);
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
}
}
return vk_outarray_status(&conn);
}
static bool
wsi_display_mode_matches_vk(wsi_display_mode *wsi,
const VkDisplayModeParametersKHR *vk)
{
return (vk->visibleRegion.width == wsi->hdisplay &&
vk->visibleRegion.height == wsi->vdisplay &&
fabs(wsi_display_mode_refresh(wsi) * 1000.0 - vk->refreshRate) < 10);
}
/*
* Implement vkCreateDisplayModeKHR (VK_KHR_display)
*/
VkResult
wsi_display_create_display_mode(VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
VkDisplayKHR display,
const VkDisplayModeCreateInfoKHR *create_info,
const VkAllocationCallbacks *allocator,
VkDisplayModeKHR *mode)
{
struct wsi_display_connector *connector =
wsi_display_connector_from_handle(display);
if (create_info->flags != 0)
return VK_ERROR_INITIALIZATION_FAILED;
/* Check and see if the requested mode happens to match an existing one and
* return that. This makes the conformance suite happy. Doing more than
* this would involve embedding the CVT function into the driver, which seems
* excessive.
*/
wsi_for_each_display_mode(display_mode, connector) {
if (display_mode->valid) {
if (wsi_display_mode_matches_vk(display_mode, &create_info->parameters)) {
*mode = wsi_display_mode_to_handle(display_mode);
return VK_SUCCESS;
}
}
}
return VK_ERROR_INITIALIZATION_FAILED;
}
/*
* Implement vkGetDisplayPlaneCapabilities
*/
VkResult
wsi_get_display_plane_capabilities(VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
VkDisplayModeKHR mode_khr,
uint32_t plane_index,
VkDisplayPlaneCapabilitiesKHR *capabilities)
{
struct wsi_display_mode *mode = wsi_display_mode_from_handle(mode_khr);
/* XXX use actual values */
capabilities->supportedAlpha = VK_DISPLAY_PLANE_ALPHA_OPAQUE_BIT_KHR;
capabilities->minSrcPosition.x = 0;
capabilities->minSrcPosition.y = 0;
capabilities->maxSrcPosition.x = 0;
capabilities->maxSrcPosition.y = 0;
capabilities->minSrcExtent.width = mode->hdisplay;
capabilities->minSrcExtent.height = mode->vdisplay;
capabilities->maxSrcExtent.width = mode->hdisplay;
capabilities->maxSrcExtent.height = mode->vdisplay;
capabilities->minDstPosition.x = 0;
capabilities->minDstPosition.y = 0;
capabilities->maxDstPosition.x = 0;
capabilities->maxDstPosition.y = 0;
capabilities->minDstExtent.width = mode->hdisplay;
capabilities->minDstExtent.height = mode->vdisplay;
capabilities->maxDstExtent.width = mode->hdisplay;
capabilities->maxDstExtent.height = mode->vdisplay;
return VK_SUCCESS;
}
VkResult
wsi_get_display_plane_capabilities2(
VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
const VkDisplayPlaneInfo2KHR *pDisplayPlaneInfo,
VkDisplayPlaneCapabilities2KHR *capabilities)
{
assert(capabilities->sType ==
VK_STRUCTURE_TYPE_DISPLAY_PLANE_CAPABILITIES_2_KHR);
return wsi_get_display_plane_capabilities(physical_device, wsi_device,
pDisplayPlaneInfo->mode,
pDisplayPlaneInfo->planeIndex,
&capabilities->capabilities);
}
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
VkResult
wsi_create_display_surface(VkInstance instance,
const VkAllocationCallbacks *allocator,
const VkDisplaySurfaceCreateInfoKHR *create_info,
VkSurfaceKHR *surface_khr)
{
VkIcdSurfaceDisplay *surface = vk_zalloc(allocator, sizeof *surface, 8,
VK_SYSTEM_ALLOCATION_SCOPE_OBJECT);
if (surface == NULL)
return VK_ERROR_OUT_OF_HOST_MEMORY;
surface->base.platform = VK_ICD_WSI_PLATFORM_DISPLAY;
surface->displayMode = create_info->displayMode;
surface->planeIndex = create_info->planeIndex;
surface->planeStackIndex = create_info->planeStackIndex;
surface->transform = create_info->transform;
surface->globalAlpha = create_info->globalAlpha;
surface->alphaMode = create_info->alphaMode;
surface->imageExtent = create_info->imageExtent;
*surface_khr = VkIcdSurfaceBase_to_handle(&surface->base);
return VK_SUCCESS;
}
static VkResult
wsi_display_surface_get_support(VkIcdSurfaceBase *surface,
struct wsi_device *wsi_device,
const VkAllocationCallbacks *allocator,
uint32_t queueFamilyIndex,
int local_fd,
VkBool32* pSupported)
{
*pSupported = VK_TRUE;
return VK_SUCCESS;
}
static VkResult
wsi_display_surface_get_capabilities(VkIcdSurfaceBase *surface_base,
VkSurfaceCapabilitiesKHR* caps)
{
VkIcdSurfaceDisplay *surface = (VkIcdSurfaceDisplay *) surface_base;
wsi_display_mode *mode = wsi_display_mode_from_handle(surface->displayMode);
caps->currentExtent.width = mode->hdisplay;
caps->currentExtent.height = mode->vdisplay;
/* XXX Figure out extents based on driver capabilities */
caps->maxImageExtent = caps->minImageExtent = caps->currentExtent;
caps->supportedCompositeAlpha = VK_COMPOSITE_ALPHA_OPAQUE_BIT_KHR;
caps->minImageCount = 2;
caps->maxImageCount = 0;
caps->supportedTransforms = VK_SURFACE_TRANSFORM_IDENTITY_BIT_KHR;
caps->currentTransform = VK_SURFACE_TRANSFORM_IDENTITY_BIT_KHR;
caps->maxImageArrayLayers = 1;
caps->supportedUsageFlags =
VK_IMAGE_USAGE_TRANSFER_SRC_BIT |
VK_IMAGE_USAGE_SAMPLED_BIT |
VK_IMAGE_USAGE_TRANSFER_DST_BIT |
VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT;
return VK_SUCCESS;
}
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
static VkResult
wsi_display_surface_get_surface_counters(
VkIcdSurfaceBase *surface_base,
VkSurfaceCounterFlagsEXT *counters)
{
*counters = VK_SURFACE_COUNTER_VBLANK_EXT;
return VK_SUCCESS;
}
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
static VkResult
wsi_display_surface_get_capabilities2(VkIcdSurfaceBase *icd_surface,
const void *info_next,
VkSurfaceCapabilities2KHR *caps)
{
assert(caps->sType == VK_STRUCTURE_TYPE_SURFACE_CAPABILITIES_2_KHR);
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
VkResult result;
result = wsi_display_surface_get_capabilities(icd_surface,
&caps->surfaceCapabilities);
if (result != VK_SUCCESS)
return result;
struct wsi_surface_supported_counters *counters =
vk_find_struct( caps->pNext, WSI_SURFACE_SUPPORTED_COUNTERS_MESA);
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
if (counters) {
result = wsi_display_surface_get_surface_counters(
icd_surface,
&counters->supported_surface_counters);
}
return result;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
}
static const struct {
VkFormat format;
uint32_t drm_format;
} available_surface_formats[] = {
{ .format = VK_FORMAT_B8G8R8A8_SRGB, .drm_format = DRM_FORMAT_XRGB8888 },
{ .format = VK_FORMAT_B8G8R8A8_UNORM, .drm_format = DRM_FORMAT_XRGB8888 },
};
static VkResult
wsi_display_surface_get_formats(VkIcdSurfaceBase *icd_surface,
struct wsi_device *wsi_device,
uint32_t *surface_format_count,
VkSurfaceFormatKHR *surface_formats)
{
VK_OUTARRAY_MAKE(out, surface_formats, surface_format_count);
for (unsigned i = 0; i < ARRAY_SIZE(available_surface_formats); i++) {
vk_outarray_append(&out, f) {
f->format = available_surface_formats[i].format;
f->colorSpace = VK_COLORSPACE_SRGB_NONLINEAR_KHR;
}
}
return vk_outarray_status(&out);
}
static VkResult
wsi_display_surface_get_formats2(VkIcdSurfaceBase *surface,
struct wsi_device *wsi_device,
const void *info_next,
uint32_t *surface_format_count,
VkSurfaceFormat2KHR *surface_formats)
{
VK_OUTARRAY_MAKE(out, surface_formats, surface_format_count);
for (unsigned i = 0; i < ARRAY_SIZE(available_surface_formats); i++) {
vk_outarray_append(&out, f) {
assert(f->sType == VK_STRUCTURE_TYPE_SURFACE_FORMAT_2_KHR);
f->surfaceFormat.format = available_surface_formats[i].format;
f->surfaceFormat.colorSpace = VK_COLORSPACE_SRGB_NONLINEAR_KHR;
}
}
return vk_outarray_status(&out);
}
static VkResult
wsi_display_surface_get_present_modes(VkIcdSurfaceBase *surface,
uint32_t *present_mode_count,
VkPresentModeKHR *present_modes)
{
VK_OUTARRAY_MAKE(conn, present_modes, present_mode_count);
vk_outarray_append(&conn, present) {
*present = VK_PRESENT_MODE_FIFO_KHR;
}
return vk_outarray_status(&conn);
}
static void
wsi_display_destroy_buffer(struct wsi_display *wsi,
uint32_t buffer)
{
(void) drmIoctl(wsi->fd, DRM_IOCTL_MODE_DESTROY_DUMB,
&((struct drm_mode_destroy_dumb) { .handle = buffer }));
}
static VkResult
wsi_display_image_init(VkDevice device_h,
struct wsi_swapchain *drv_chain,
const VkSwapchainCreateInfoKHR *create_info,
const VkAllocationCallbacks *allocator,
struct wsi_display_image *image)
{
struct wsi_display_swapchain *chain =
(struct wsi_display_swapchain *) drv_chain;
struct wsi_display *wsi = chain->wsi;
uint32_t drm_format = 0;
for (unsigned i = 0; i < ARRAY_SIZE(available_surface_formats); i++) {
if (create_info->imageFormat == available_surface_formats[i].format) {
drm_format = available_surface_formats[i].drm_format;
break;
}
}
/* the application provided an invalid format, bail */
if (drm_format == 0)
return VK_ERROR_DEVICE_LOST;
VkResult result = wsi_create_native_image(&chain->base, create_info,
0, NULL, NULL,
&image->base);
if (result != VK_SUCCESS)
return result;
memset(image->buffer, 0, sizeof (image->buffer));
for (unsigned int i = 0; i < image->base.num_planes; i++) {
int ret = drmPrimeFDToHandle(wsi->fd, image->base.fds[i],
&image->buffer[i]);
close(image->base.fds[i]);
image->base.fds[i] = -1;
if (ret < 0)
goto fail_handle;
}
image->chain = chain;
image->state = WSI_IMAGE_IDLE;
image->fb_id = 0;
int ret = drmModeAddFB2(wsi->fd,
create_info->imageExtent.width,
create_info->imageExtent.height,
drm_format,
image->buffer,
image->base.row_pitches,
image->base.offsets,
&image->fb_id, 0);
if (ret)
goto fail_fb;
return VK_SUCCESS;
fail_fb:
fail_handle:
for (unsigned int i = 0; i < image->base.num_planes; i++) {
if (image->buffer[i])
wsi_display_destroy_buffer(wsi, image->buffer[i]);
if (image->base.fds[i] != -1) {
close(image->base.fds[i]);
image->base.fds[i] = -1;
}
}
wsi_destroy_image(&chain->base, &image->base);
return VK_ERROR_OUT_OF_HOST_MEMORY;
}
static void
wsi_display_image_finish(struct wsi_swapchain *drv_chain,
const VkAllocationCallbacks *allocator,
struct wsi_display_image *image)
{
struct wsi_display_swapchain *chain =
(struct wsi_display_swapchain *) drv_chain;
struct wsi_display *wsi = chain->wsi;
drmModeRmFB(wsi->fd, image->fb_id);
for (unsigned int i = 0; i < image->base.num_planes; i++)
wsi_display_destroy_buffer(wsi, image->buffer[i]);
wsi_destroy_image(&chain->base, &image->base);
}
static VkResult
wsi_display_swapchain_destroy(struct wsi_swapchain *drv_chain,
const VkAllocationCallbacks *allocator)
{
struct wsi_display_swapchain *chain =
(struct wsi_display_swapchain *) drv_chain;
for (uint32_t i = 0; i < chain->base.image_count; i++)
wsi_display_image_finish(drv_chain, allocator, &chain->images[i]);
vk_free(allocator, chain);
return VK_SUCCESS;
}
static struct wsi_image *
wsi_display_get_wsi_image(struct wsi_swapchain *drv_chain,
uint32_t image_index)
{
struct wsi_display_swapchain *chain =
(struct wsi_display_swapchain *) drv_chain;
return &chain->images[image_index].base;
}
static void
wsi_display_idle_old_displaying(struct wsi_display_image *active_image)
{
struct wsi_display_swapchain *chain = active_image->chain;
wsi_display_debug("idle everyone but %ld\n",
active_image - &(chain->images[0]));
for (uint32_t i = 0; i < chain->base.image_count; i++)
if (chain->images[i].state == WSI_IMAGE_DISPLAYING &&
&chain->images[i] != active_image)
{
wsi_display_debug("idle %d\n", i);
chain->images[i].state = WSI_IMAGE_IDLE;
}
}
static VkResult
_wsi_display_queue_next(struct wsi_swapchain *drv_chain);
static void
wsi_display_page_flip_handler2(int fd,
unsigned int frame,
unsigned int sec,
unsigned int usec,
uint32_t crtc_id,
void *data)
{
struct wsi_display_image *image = data;
struct wsi_display_swapchain *chain = image->chain;
wsi_display_debug("image %ld displayed at %d\n",
image - &(image->chain->images[0]), frame);
image->state = WSI_IMAGE_DISPLAYING;
wsi_display_idle_old_displaying(image);
VkResult result = _wsi_display_queue_next(&(chain->base));
if (result != VK_SUCCESS)
chain->status = result;
}
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
static void wsi_display_fence_event_handler(struct wsi_display_fence *fence);
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
static void wsi_display_page_flip_handler(int fd,
unsigned int frame,
unsigned int sec,
unsigned int usec,
void *data)
{
wsi_display_page_flip_handler2(fd, frame, sec, usec, 0, data);
}
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
static void wsi_display_vblank_handler(int fd, unsigned int frame,
unsigned int sec, unsigned int usec,
void *data)
{
struct wsi_display_fence *fence = data;
wsi_display_fence_event_handler(fence);
}
static void wsi_display_sequence_handler(int fd, uint64_t frame,
uint64_t nsec, uint64_t user_data)
{
struct wsi_display_fence *fence =
(struct wsi_display_fence *) (uintptr_t) user_data;
wsi_display_fence_event_handler(fence);
}
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
static drmEventContext event_context = {
.version = DRM_EVENT_CONTEXT_VERSION,
.page_flip_handler = wsi_display_page_flip_handler,
#if DRM_EVENT_CONTEXT_VERSION >= 3
.page_flip_handler2 = wsi_display_page_flip_handler2,
#endif
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
.vblank_handler = wsi_display_vblank_handler,
.sequence_handler = wsi_display_sequence_handler,
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
};
static void *
wsi_display_wait_thread(void *data)
{
struct wsi_display *wsi = data;
struct pollfd pollfd = {
.fd = wsi->fd,
.events = POLLIN
};
pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, NULL);
for (;;) {
int ret = poll(&pollfd, 1, -1);
if (ret > 0) {
pthread_mutex_lock(&wsi->wait_mutex);
(void) drmHandleEvent(wsi->fd, &event_context);
pthread_mutex_unlock(&wsi->wait_mutex);
pthread_cond_broadcast(&wsi->wait_cond);
}
}
return NULL;
}
static int
wsi_display_start_wait_thread(struct wsi_display *wsi)
{
if (!wsi->wait_thread) {
int ret = pthread_create(&wsi->wait_thread, NULL,
wsi_display_wait_thread, wsi);
if (ret)
return ret;
}
return 0;
}
/*
* Wait for at least one event from the kernel to be processed.
* Call with wait_mutex held
*/
static int
wsi_display_wait_for_event(struct wsi_display *wsi,
uint64_t timeout_ns)
{
int ret;
ret = wsi_display_start_wait_thread(wsi);
if (ret)
return ret;
struct timespec abs_timeout = {
.tv_sec = timeout_ns / 1000000000ULL,
.tv_nsec = timeout_ns % 1000000000ULL,
};
ret = pthread_cond_timedwait(&wsi->wait_cond, &wsi->wait_mutex,
&abs_timeout);
wsi_display_debug("%9ld done waiting for event %d\n", pthread_self(), ret);
return ret;
}
static VkResult
wsi_display_acquire_next_image(struct wsi_swapchain *drv_chain,
uint64_t timeout,
VkSemaphore semaphore,
uint32_t *image_index)
{
struct wsi_display_swapchain *chain =
(struct wsi_display_swapchain *)drv_chain;
struct wsi_display *wsi = chain->wsi;
int ret = 0;
VkResult result = VK_SUCCESS;
/* Bail early if the swapchain is broken */
if (chain->status != VK_SUCCESS)
return chain->status;
if (timeout != 0 && timeout != UINT64_MAX)
timeout = wsi_rel_to_abs_time(timeout);
pthread_mutex_lock(&wsi->wait_mutex);
for (;;) {
for (uint32_t i = 0; i < chain->base.image_count; i++) {
if (chain->images[i].state == WSI_IMAGE_IDLE) {
*image_index = i;
wsi_display_debug("image %d available\n", i);
chain->images[i].state = WSI_IMAGE_DRAWING;
result = VK_SUCCESS;
goto done;
}
wsi_display_debug("image %d state %d\n", i, chain->images[i].state);
}
if (ret == ETIMEDOUT) {
result = VK_TIMEOUT;
goto done;
}
ret = wsi_display_wait_for_event(wsi, timeout);
if (ret && ret != ETIMEDOUT) {
result = VK_ERROR_SURFACE_LOST_KHR;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
goto done;
}
}
done:
pthread_mutex_unlock(&wsi->wait_mutex);
if (result != VK_SUCCESS)
return result;
return chain->status;
}
/*
* Check whether there are any other connectors driven by this crtc
*/
static bool
wsi_display_crtc_solo(struct wsi_display *wsi,
drmModeResPtr mode_res,
drmModeConnectorPtr connector,
uint32_t crtc_id)
{
/* See if any other connectors share the same encoder */
for (int c = 0; c < mode_res->count_connectors; c++) {
if (mode_res->connectors[c] == connector->connector_id)
continue;
drmModeConnectorPtr other_connector =
drmModeGetConnector(wsi->fd, mode_res->connectors[c]);
if (other_connector) {
bool match = (other_connector->encoder_id == connector->encoder_id);
drmModeFreeConnector(other_connector);
if (match)
return false;
}
}
/* See if any other encoders share the same crtc */
for (int e = 0; e < mode_res->count_encoders; e++) {
if (mode_res->encoders[e] == connector->encoder_id)
continue;
drmModeEncoderPtr other_encoder =
drmModeGetEncoder(wsi->fd, mode_res->encoders[e]);
if (other_encoder) {
bool match = (other_encoder->crtc_id == crtc_id);
drmModeFreeEncoder(other_encoder);
if (match)
return false;
}
}
return true;
}
/*
* Pick a suitable CRTC to drive this connector. Prefer a CRTC which is
* currently driving this connector and not any others. Settle for a CRTC
* which is currently idle.
*/
static uint32_t
wsi_display_select_crtc(struct wsi_display_connector *connector,
drmModeResPtr mode_res,
drmModeConnectorPtr drm_connector)
{
struct wsi_display *wsi = connector->wsi;
/* See what CRTC is currently driving this connector */
if (drm_connector->encoder_id) {
drmModeEncoderPtr encoder =
drmModeGetEncoder(wsi->fd, drm_connector->encoder_id);
if (encoder) {
uint32_t crtc_id = encoder->crtc_id;
drmModeFreeEncoder(encoder);
if (crtc_id) {
if (wsi_display_crtc_solo(wsi, mode_res, drm_connector, crtc_id))
return crtc_id;
}
}
}
uint32_t crtc_id = 0;
for (int c = 0; crtc_id == 0 && c < mode_res->count_crtcs; c++) {
drmModeCrtcPtr crtc = drmModeGetCrtc(wsi->fd, mode_res->crtcs[c]);
if (crtc && crtc->buffer_id == 0)
crtc_id = crtc->crtc_id;
drmModeFreeCrtc(crtc);
}
return crtc_id;
}
static VkResult
wsi_display_setup_connector(wsi_display_connector *connector,
wsi_display_mode *display_mode)
{
struct wsi_display *wsi = connector->wsi;
if (connector->current_mode == display_mode && connector->crtc_id)
return VK_SUCCESS;
VkResult result = VK_SUCCESS;
drmModeResPtr mode_res = drmModeGetResources(wsi->fd);
if (!mode_res) {
if (errno == ENOMEM)
result = VK_ERROR_OUT_OF_HOST_MEMORY;
else
result = VK_ERROR_SURFACE_LOST_KHR;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
goto bail;
}
drmModeConnectorPtr drm_connector =
drmModeGetConnectorCurrent(wsi->fd, connector->id);
if (!drm_connector) {
if (errno == ENOMEM)
result = VK_ERROR_OUT_OF_HOST_MEMORY;
else
result = VK_ERROR_SURFACE_LOST_KHR;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
goto bail_mode_res;
}
/* Pick a CRTC if we don't have one */
if (!connector->crtc_id) {
connector->crtc_id = wsi_display_select_crtc(connector,
mode_res, drm_connector);
if (!connector->crtc_id) {
result = VK_ERROR_SURFACE_LOST_KHR;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
goto bail_connector;
}
}
if (connector->current_mode != display_mode) {
/* Find the drm mode corresponding to the requested VkDisplayMode */
drmModeModeInfoPtr drm_mode = NULL;
for (int m = 0; m < drm_connector->count_modes; m++) {
drm_mode = &drm_connector->modes[m];
if (wsi_display_mode_matches_drm(display_mode, drm_mode))
break;
drm_mode = NULL;
}
if (!drm_mode) {
result = VK_ERROR_SURFACE_LOST_KHR;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
goto bail_connector;
}
connector->current_mode = display_mode;
connector->current_drm_mode = *drm_mode;
}
bail_connector:
drmModeFreeConnector(drm_connector);
bail_mode_res:
drmModeFreeResources(mode_res);
bail:
return result;
}
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
static VkResult
wsi_display_fence_wait(struct wsi_fence *fence_wsi, uint64_t timeout)
{
const struct wsi_device *wsi_device = fence_wsi->wsi_device;
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
struct wsi_display_fence *fence = (struct wsi_display_fence *) fence_wsi;
wsi_display_debug("%9lu wait fence %lu %ld\n",
pthread_self(), fence->sequence,
(int64_t) (timeout - wsi_get_current_monotonic()));
wsi_display_debug_code(uint64_t start_ns = wsi_get_current_monotonic());
pthread_mutex_lock(&wsi->wait_mutex);
VkResult result;
int ret = 0;
for (;;) {
if (fence->event_received) {
wsi_display_debug("%9lu fence %lu passed\n",
pthread_self(), fence->sequence);
result = VK_SUCCESS;
break;
}
if (ret == ETIMEDOUT) {
wsi_display_debug("%9lu fence %lu timeout\n",
pthread_self(), fence->sequence);
result = VK_TIMEOUT;
break;
}
ret = wsi_display_wait_for_event(wsi, timeout);
if (ret && ret != ETIMEDOUT) {
wsi_display_debug("%9lu fence %lu error\n",
pthread_self(), fence->sequence);
result = VK_ERROR_DEVICE_LOST;
break;
}
}
pthread_mutex_unlock(&wsi->wait_mutex);
wsi_display_debug("%9lu fence wait %f ms\n",
pthread_self(),
((int64_t) (wsi_get_current_monotonic() - start_ns)) /
1.0e6);
return result;
}
static void
wsi_display_fence_check_free(struct wsi_display_fence *fence)
{
if (fence->event_received && fence->destroyed)
vk_free(fence->base.alloc, fence);
}
static void wsi_display_fence_event_handler(struct wsi_display_fence *fence)
{
fence->event_received = true;
wsi_display_fence_check_free(fence);
}
static void
wsi_display_fence_destroy(struct wsi_fence *fence_wsi)
{
struct wsi_display_fence *fence = (struct wsi_display_fence *) fence_wsi;
assert(!fence->destroyed);
fence->destroyed = true;
wsi_display_fence_check_free(fence);
}
static struct wsi_display_fence *
wsi_display_fence_alloc(VkDevice device,
const struct wsi_device *wsi_device,
VkDisplayKHR display,
const VkAllocationCallbacks *allocator)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
struct wsi_display_fence *fence =
vk_zalloc2(wsi->alloc, allocator, sizeof (*fence),
8, VK_SYSTEM_ALLOCATION_SCOPE_OBJECT);
if (!fence)
return NULL;
fence->base.device = device;
fence->base.display = display;
fence->base.wsi_device = wsi_device;
fence->base.alloc = allocator ? allocator : wsi->alloc;
fence->base.wait = wsi_display_fence_wait;
fence->base.destroy = wsi_display_fence_destroy;
fence->event_received = false;
fence->destroyed = false;
fence->sequence = ++fence_sequence;
return fence;
}
static VkResult
wsi_register_vblank_event(struct wsi_display_fence *fence,
const struct wsi_device *wsi_device,
VkDisplayKHR display,
uint32_t flags,
uint64_t frame_requested,
uint64_t *frame_queued)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
struct wsi_display_connector *connector =
wsi_display_connector_from_handle(display);
if (wsi->fd < 0)
return VK_ERROR_INITIALIZATION_FAILED;
for (;;) {
int ret = drmCrtcQueueSequence(wsi->fd, connector->crtc_id,
flags,
frame_requested,
frame_queued,
(uintptr_t) fence);
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
if (!ret)
return VK_SUCCESS;
if (errno != ENOMEM) {
/* Something unexpected happened. Pause for a moment so the
* application doesn't just spin and then return a failure indication
*/
wsi_display_debug("queue vblank event %lu failed\n", fence->sequence);
struct timespec delay = {
.tv_sec = 0,
.tv_nsec = 100000000ull,
};
nanosleep(&delay, NULL);
return VK_ERROR_OUT_OF_HOST_MEMORY;
}
/* The kernel event queue is full. Wait for some events to be
* processed and try again
*/
pthread_mutex_lock(&wsi->wait_mutex);
ret = wsi_display_wait_for_event(wsi, wsi_rel_to_abs_time(100000000ull));
pthread_mutex_unlock(&wsi->wait_mutex);
if (ret) {
wsi_display_debug("vblank queue full, event wait failed\n");
return VK_ERROR_OUT_OF_HOST_MEMORY;
}
}
}
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
/*
* Check to see if the kernel has no flip queued and if there's an image
* waiting to be displayed.
*/
static VkResult
_wsi_display_queue_next(struct wsi_swapchain *drv_chain)
{
struct wsi_display_swapchain *chain =
(struct wsi_display_swapchain *) drv_chain;
struct wsi_display *wsi = chain->wsi;
VkIcdSurfaceDisplay *surface = chain->surface;
wsi_display_mode *display_mode =
wsi_display_mode_from_handle(surface->displayMode);
wsi_display_connector *connector = display_mode->connector;
if (wsi->fd < 0)
return VK_ERROR_SURFACE_LOST_KHR;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
if (display_mode != connector->current_mode)
connector->active = false;
for (;;) {
/* Check to see if there is an image to display, or if some image is
* already queued */
struct wsi_display_image *image = NULL;
for (uint32_t i = 0; i < chain->base.image_count; i++) {
struct wsi_display_image *tmp_image = &chain->images[i];
switch (tmp_image->state) {
case WSI_IMAGE_FLIPPING:
/* already flipping, don't send another to the kernel yet */
return VK_SUCCESS;
case WSI_IMAGE_QUEUED:
/* find the oldest queued */
if (!image || tmp_image->flip_sequence < image->flip_sequence)
image = tmp_image;
break;
default:
break;
}
}
if (!image)
return VK_SUCCESS;
int ret;
if (connector->active) {
ret = drmModePageFlip(wsi->fd, connector->crtc_id, image->fb_id,
DRM_MODE_PAGE_FLIP_EVENT, image);
if (ret == 0) {
image->state = WSI_IMAGE_FLIPPING;
return VK_SUCCESS;
}
wsi_display_debug("page flip err %d %s\n", ret, strerror(-ret));
} else {
ret = -EINVAL;
}
if (ret == -EINVAL) {
VkResult result = wsi_display_setup_connector(connector, display_mode);
if (result != VK_SUCCESS) {
image->state = WSI_IMAGE_IDLE;
return result;
}
/* XXX allow setting of position */
ret = drmModeSetCrtc(wsi->fd, connector->crtc_id,
image->fb_id, 0, 0,
&connector->id, 1,
&connector->current_drm_mode);
if (ret == 0) {
/* Assume that the mode set is synchronous and that any
* previous image is now idle.
*/
image->state = WSI_IMAGE_DISPLAYING;
wsi_display_idle_old_displaying(image);
connector->active = true;
return VK_SUCCESS;
}
}
if (ret != -EACCES) {
connector->active = false;
image->state = WSI_IMAGE_IDLE;
return VK_ERROR_SURFACE_LOST_KHR;
vulkan: Add KHR_display extension using DRM [v10] This adds support for the KHR_display extension support to the vulkan WSI layer. Driver support will be added separately. v2: * fix double ;; in wsi_common_display.c * Move mode list from wsi_display to wsi_display_connector * Fix scope for wsi_display_mode andwsi_display_connector allocs * Switch all allocations to vk_zalloc instead of vk_alloc. * Fix DRM failure in wsi_display_get_physical_device_display_properties When DRM fails, or when we don't have a master fd (presumably due to application errors), just return 0 properties from this function, which is at least a valid response. * Use vk_outarray for all property queries This is a bit less error-prone than open-coding the same stuff. * Remove VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR from surface caps Until we have multi-plane support, we shouldn't pretend to have any multi-plane semantics, even if undefined. Suggested-by: Jason Ekstrand <jason@jlekstrand.net> * Simplify addition of VK_USE_PLATFORM_DISPLAY_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> v3: Add separate 'display_fd' and 'render_fd' arguments to wsi_device_init API. This allows drivers to use different FDs for the different aspects of the device. Use largest mode as display size when no preferred mode. If the display doesn't provide a preferred mode, we'll assume that the largest supported mode is the "physical size" of the device and report that. v4: Make wsi_image_state enumeration values uppercase. Follow more common mesa conventions. Remove 'render_fd' from wsi_device_init API. The wsi_common_display code doesn't use this fd at all, so stop passing it in. This avoids any potential confusion over which fd to use when creating display-relative object handles. Remove call to wsi_create_prime_image which would never have been reached as the necessary condition (use_prime_blit) is never set. whitespace cleanups in wsi_common_display.c Suggested-by: Jason Ekstrand <jason@jlekstrand.net> Add depth/bpp info to available surface formats. Instead of hard-coding depth 24 bpp 32 in the drmModeAddFB call, use the requested format to find suitable values. Destroy kernel buffers and FBs when swapchain is destroyed. We were leaking both of these kernel objects across swapchain destruction. Note that wsi_display_wait_for_event waits for anything to happen. wsi_display_wait_for_event is simply a yield so that the caller can then check to see if the desired state change has occurred. Record swapchain failures in chain for later return. If some asynchronous swapchain activity fails, we need to tell the application eventually. Record the failure in the swapchain and report it at the next acquire_next_image or queue_present call. Fix error returns from wsi_display_setup_connector. If a malloc failed, then the result should be VK_ERROR_OUT_OF_HOST_MEMORY. Otherwise, the associated ioctl failed and we're either VT switched away, or our lease has been revoked, in which case we should return VK_ERROR_OUT_OF_DATE_KHR. Make sure both sides of if/else brace use matches Note that we assume drmModeSetCrtc is synchronous. Add a comment explaining why we can idle any previous displayed image as soon as the mode set returns. Note that EACCES from drmModePageFlip means VT inactive. When vt switched away drmModePageFlip returns EACCES. Poll once a second waiting until we get some other return value back. Clean up after alloc failure in wsi_display_surface_create_swapchain. Destroy any created images, free the swapchain. Remove physical_device from wsi_display_init_wsi. We never need this value, so remove it from the API and from the internal wsi_display structure. Use drmModeAddFB2 in wsi_display_image_init. This takes a drm format instead of depth/bpp, which provides more control over the format of the data. v5: Set the 'currentStackIndex' member of the VkDisplayPlanePropertiesKHR record to zero, instead of indexing across all displays. This value is the stack depth of the plane within an individual display, and as the current code supports only a single plane per display, should be set to zero for all elements Discovered-by: David Mao <David.Mao@amd.com> v6: Remove 'platform_display' bits from the build and use the existing 'platform_drm' instead. v7: Ensure VK_ICD_WSI_PLATFORM_MAX is large enough by setting to VK_ICD_WSI_PLATFORM_DISPLAY + 1 v8: Simplify wsi_device_init failure from wsi_display_init_wsi by using the same pattern as the other wsi layers. Adopt Jason Ekstrand's white space and variable declaration suggestions. Declare variables at first use, eliminate extra whitespace between types and names, add list iterator helpers, switch to lower-case list_ macros. Respond to Jason's April 8 review: * Create a function to convert relative to absolute timeouts to catch overflow issues in one place * use VK_NULL_HANDLE to clear prop->currentDisplay * Get rid of available_present_modes array. * return OUT_OF_DATE_KHR when display_queue_next called after display has been released. * Make errors from mode setting fatal in display_queue_next * Remove duplicate pthread_mutex_init call * Add wsi_init_pthread_cond_monotonic helper function to isolate pthread error handling from wsi_display_init_wsi Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v9: Fix vscan handling by using MAX2(vscan, 1) everywhere. Vscan can be zero anywhere, which is treated the same as 1. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Respond to Vulkan CTS failures. 1. Initialize planeReorderPossible in display_properties code 2. Only report connected displays in get_display_plane_supported_displays 3. Return VK_ERROR_OUT_OF_HOST_MEMORY when pthread cond initialization fails. Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> 4. Add vkCreateDisplayModeKHR. This doesn't actually create new modes, it only looks to see if the requested parameters matches an existing mode and returns that. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Keith Packard <keithp@keithp.com>
2018-02-07 18:31:44 +00:00
}
/* Some other VT is currently active. Sit here waiting for
* our VT to become active again by polling once a second
*/
usleep(1000 * 1000);
connector->active = false;
}
}
static VkResult
wsi_display_queue_present(struct wsi_swapchain *drv_chain,
uint32_t image_index,
const VkPresentRegionKHR *damage)
{
struct wsi_display_swapchain *chain =
(struct wsi_display_swapchain *) drv_chain;
struct wsi_display *wsi = chain->wsi;
struct wsi_display_image *image = &chain->images[image_index];
VkResult result;
/* Bail early if the swapchain is broken */
if (chain->status != VK_SUCCESS)
return chain->status;
assert(image->state == WSI_IMAGE_DRAWING);
wsi_display_debug("present %d\n", image_index);
pthread_mutex_lock(&wsi->wait_mutex);
image->flip_sequence = ++chain->flip_sequence;
image->state = WSI_IMAGE_QUEUED;
result = _wsi_display_queue_next(drv_chain);
if (result != VK_SUCCESS)
chain->status = result;
pthread_mutex_unlock(&wsi->wait_mutex);
if (result != VK_SUCCESS)
return result;
return chain->status;
}
static VkResult
wsi_display_surface_create_swapchain(
VkIcdSurfaceBase *icd_surface,
VkDevice device,
struct wsi_device *wsi_device,
int local_fd,
const VkSwapchainCreateInfoKHR *create_info,
const VkAllocationCallbacks *allocator,
struct wsi_swapchain **swapchain_out)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
assert(create_info->sType == VK_STRUCTURE_TYPE_SWAPCHAIN_CREATE_INFO_KHR);
const unsigned num_images = create_info->minImageCount;
struct wsi_display_swapchain *chain =
vk_zalloc(allocator,
sizeof(*chain) + num_images * sizeof(chain->images[0]),
8, VK_SYSTEM_ALLOCATION_SCOPE_OBJECT);
if (chain == NULL)
return VK_ERROR_OUT_OF_HOST_MEMORY;
VkResult result = wsi_swapchain_init(wsi_device, &chain->base, device,
create_info, allocator);
chain->base.destroy = wsi_display_swapchain_destroy;
chain->base.get_wsi_image = wsi_display_get_wsi_image;
chain->base.acquire_next_image = wsi_display_acquire_next_image;
chain->base.queue_present = wsi_display_queue_present;
chain->base.present_mode = create_info->presentMode;
chain->base.image_count = num_images;
chain->wsi = wsi;
chain->status = VK_SUCCESS;
chain->surface = (VkIcdSurfaceDisplay *) icd_surface;
for (uint32_t image = 0; image < chain->base.image_count; image++) {
result = wsi_display_image_init(device, &chain->base,
create_info, allocator,
&chain->images[image]);
if (result != VK_SUCCESS) {
while (image > 0) {
--image;
wsi_display_image_finish(&chain->base, allocator,
&chain->images[image]);
}
vk_free(allocator, chain);
goto fail_init_images;
}
}
*swapchain_out = &chain->base;
return VK_SUCCESS;
fail_init_images:
return result;
}
static bool
wsi_init_pthread_cond_monotonic(pthread_cond_t *cond)
{
pthread_condattr_t condattr;
bool ret = false;
if (pthread_condattr_init(&condattr) != 0)
goto fail_attr_init;
if (pthread_condattr_setclock(&condattr, CLOCK_MONOTONIC) != 0)
goto fail_attr_set;
if (pthread_cond_init(cond, &condattr) != 0)
goto fail_cond_init;
ret = true;
fail_cond_init:
fail_attr_set:
pthread_condattr_destroy(&condattr);
fail_attr_init:
return ret;
}
VkResult
wsi_display_init_wsi(struct wsi_device *wsi_device,
const VkAllocationCallbacks *alloc,
int display_fd)
{
struct wsi_display *wsi = vk_zalloc(alloc, sizeof(*wsi), 8,
VK_SYSTEM_ALLOCATION_SCOPE_INSTANCE);
VkResult result;
if (!wsi) {
result = VK_ERROR_OUT_OF_HOST_MEMORY;
goto fail;
}
wsi->fd = display_fd;
wsi->alloc = alloc;
list_inithead(&wsi->connectors);
int ret = pthread_mutex_init(&wsi->wait_mutex, NULL);
if (ret) {
result = VK_ERROR_OUT_OF_HOST_MEMORY;
goto fail_mutex;
}
if (!wsi_init_pthread_cond_monotonic(&wsi->wait_cond)) {
result = VK_ERROR_OUT_OF_HOST_MEMORY;
goto fail_cond;
}
wsi->base.get_support = wsi_display_surface_get_support;
wsi->base.get_capabilities2 = wsi_display_surface_get_capabilities2;
wsi->base.get_formats = wsi_display_surface_get_formats;
wsi->base.get_formats2 = wsi_display_surface_get_formats2;
wsi->base.get_present_modes = wsi_display_surface_get_present_modes;
wsi->base.create_swapchain = wsi_display_surface_create_swapchain;
wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY] = &wsi->base;
return VK_SUCCESS;
fail_cond:
pthread_mutex_destroy(&wsi->wait_mutex);
fail_mutex:
vk_free(alloc, wsi);
fail:
return result;
}
void
wsi_display_finish_wsi(struct wsi_device *wsi_device,
const VkAllocationCallbacks *alloc)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
if (wsi) {
wsi_for_each_connector(connector, wsi) {
wsi_for_each_display_mode(mode, connector) {
vk_free(wsi->alloc, mode);
}
vk_free(wsi->alloc, connector);
}
pthread_mutex_lock(&wsi->wait_mutex);
if (wsi->wait_thread) {
pthread_cancel(wsi->wait_thread);
pthread_join(wsi->wait_thread, NULL);
}
pthread_mutex_unlock(&wsi->wait_mutex);
pthread_mutex_destroy(&wsi->wait_mutex);
pthread_cond_destroy(&wsi->wait_cond);
vk_free(alloc, wsi);
}
}
/*
* Implement vkReleaseDisplay
*/
VkResult
wsi_release_display(VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
VkDisplayKHR display)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
if (wsi->fd >= 0) {
close(wsi->fd);
wsi->fd = -1;
}
vulkan: Add EXT_acquire_xlib_display [v5] This extension adds the ability to borrow an X RandR output for temporary use directly by a Vulkan application. For DRM, we use the Linux resource leasing mechanism. v2: Clean up xlib_lease detection * Use separate temporary '_xlib_lease' variable to hold the option value to avoid changin the type of a variable. * Use boolean expressions instead of additional if statements to compute resulting with_xlib_lease value. * Simplify addition of VK_USE_PLATFORM_XLIB_XRANDR_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> Move mode list from wsi_display to wsi_display_connector Fix scope for wsi_display_mode and wsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v3: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Explicitly forbid multiple DRM leases. Making the code support this looks tricky and will require additional thought. Use xcb_randr_output_t throughout the internals of the implementation. Convert at the public API (wsi_get_randr_output_display). Clean up check for usable active_crtc (possible when only the desired output is connected to the crtc). Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v4: Move output resource fetching closer to use in wsi_display_get_output. This simplifies the error returns in earlier parts of the code a bit. Return VK_ERROR_INITIALIZATION_FAILED from wsi_acquire_xlib_display. Jason says this is the right error message. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v5: randr doesn't pass vscan over the wire, so we set vscan to 0 for randr-acquired modes, and test wsi modes for vscan <= 1 when comparing against randr modes. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
2018-02-09 15:45:58 +00:00
#ifdef VK_USE_PLATFORM_XLIB_XRANDR_EXT
wsi_display_connector_from_handle(display)->output = None;
#endif
return VK_SUCCESS;
}
#ifdef VK_USE_PLATFORM_XLIB_XRANDR_EXT
static struct wsi_display_connector *
wsi_display_find_output(struct wsi_device *wsi_device,
xcb_randr_output_t output)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
wsi_for_each_connector(connector, wsi) {
if (connector->output == output)
return connector;
}
return NULL;
}
/*
* Given a RandR output, find the associated kernel connector_id by
* looking at the CONNECTOR_ID property provided by the X server
*/
static uint32_t
wsi_display_output_to_connector_id(xcb_connection_t *connection,
xcb_atom_t *connector_id_atom_p,
xcb_randr_output_t output)
{
uint32_t connector_id = 0;
xcb_atom_t connector_id_atom = *connector_id_atom_p;
if (connector_id_atom == 0) {
/* Go dig out the CONNECTOR_ID property */
xcb_intern_atom_cookie_t ia_c = xcb_intern_atom(connection,
true,
12,
"CONNECTOR_ID");
xcb_intern_atom_reply_t *ia_r = xcb_intern_atom_reply(connection,
ia_c,
NULL);
if (ia_r) {
*connector_id_atom_p = connector_id_atom = ia_r->atom;
free(ia_r);
}
}
/* If there's an CONNECTOR_ID atom in the server, then there may be a
* CONNECTOR_ID property. Otherwise, there will not be and we don't even
* need to bother.
*/
if (connector_id_atom) {
xcb_randr_query_version_cookie_t qv_c =
xcb_randr_query_version(connection, 1, 6);
xcb_randr_get_output_property_cookie_t gop_c =
xcb_randr_get_output_property(connection,
output,
connector_id_atom,
0,
0,
0xffffffffUL,
0,
0);
xcb_randr_query_version_reply_t *qv_r =
xcb_randr_query_version_reply(connection, qv_c, NULL);
free(qv_r);
xcb_randr_get_output_property_reply_t *gop_r =
xcb_randr_get_output_property_reply(connection, gop_c, NULL);
if (gop_r) {
if (gop_r->num_items == 1 && gop_r->format == 32)
memcpy(&connector_id, xcb_randr_get_output_property_data(gop_r), 4);
free(gop_r);
}
}
return connector_id;
}
static bool
wsi_display_check_randr_version(xcb_connection_t *connection)
{
xcb_randr_query_version_cookie_t qv_c =
xcb_randr_query_version(connection, 1, 6);
xcb_randr_query_version_reply_t *qv_r =
xcb_randr_query_version_reply(connection, qv_c, NULL);
bool ret = false;
if (!qv_r)
return false;
/* Check for version 1.6 or newer */
ret = (qv_r->major_version > 1 ||
(qv_r->major_version == 1 && qv_r->minor_version >= 6));
free(qv_r);
return ret;
}
/*
* Given a kernel connector id, find the associated RandR output using the
* CONNECTOR_ID property
*/
static xcb_randr_output_t
wsi_display_connector_id_to_output(xcb_connection_t *connection,
uint32_t connector_id)
{
if (!wsi_display_check_randr_version(connection))
return 0;
const xcb_setup_t *setup = xcb_get_setup(connection);
xcb_atom_t connector_id_atom = 0;
xcb_randr_output_t output = 0;
/* Search all of the screens for the provided output */
xcb_screen_iterator_t iter;
for (iter = xcb_setup_roots_iterator(setup);
output == 0 && iter.rem;
xcb_screen_next(&iter))
{
xcb_randr_get_screen_resources_cookie_t gsr_c =
xcb_randr_get_screen_resources(connection, iter.data->root);
xcb_randr_get_screen_resources_reply_t *gsr_r =
xcb_randr_get_screen_resources_reply(connection, gsr_c, NULL);
if (!gsr_r)
return 0;
xcb_randr_output_t *ro = xcb_randr_get_screen_resources_outputs(gsr_r);
int o;
for (o = 0; o < gsr_r->num_outputs; o++) {
if (wsi_display_output_to_connector_id(connection,
&connector_id_atom, ro[o])
== connector_id)
{
output = ro[o];
break;
}
}
free(gsr_r);
}
return output;
}
/*
* Given a RandR output, find out which screen it's associated with
*/
static xcb_window_t
wsi_display_output_to_root(xcb_connection_t *connection,
xcb_randr_output_t output)
{
if (!wsi_display_check_randr_version(connection))
return 0;
const xcb_setup_t *setup = xcb_get_setup(connection);
xcb_window_t root = 0;
/* Search all of the screens for the provided output */
for (xcb_screen_iterator_t iter = xcb_setup_roots_iterator(setup);
root == 0 && iter.rem;
xcb_screen_next(&iter))
{
xcb_randr_get_screen_resources_cookie_t gsr_c =
xcb_randr_get_screen_resources(connection, iter.data->root);
xcb_randr_get_screen_resources_reply_t *gsr_r =
xcb_randr_get_screen_resources_reply(connection, gsr_c, NULL);
if (!gsr_r)
return 0;
xcb_randr_output_t *ro = xcb_randr_get_screen_resources_outputs(gsr_r);
for (int o = 0; o < gsr_r->num_outputs; o++) {
if (ro[o] == output) {
root = iter.data->root;
break;
}
}
free(gsr_r);
}
return root;
}
static bool
wsi_display_mode_matches_x(struct wsi_display_mode *wsi,
xcb_randr_mode_info_t *xcb)
{
return wsi->clock == (xcb->dot_clock + 500) / 1000 &&
wsi->hdisplay == xcb->width &&
wsi->hsync_start == xcb->hsync_start &&
wsi->hsync_end == xcb->hsync_end &&
wsi->htotal == xcb->htotal &&
wsi->hskew == xcb->hskew &&
wsi->vdisplay == xcb->height &&
wsi->vsync_start == xcb->vsync_start &&
wsi->vsync_end == xcb->vsync_end &&
wsi->vtotal == xcb->vtotal &&
wsi->vscan <= 1 &&
wsi->flags == xcb->mode_flags;
}
static struct wsi_display_mode *
wsi_display_find_x_mode(struct wsi_device *wsi_device,
struct wsi_display_connector *connector,
xcb_randr_mode_info_t *mode)
{
wsi_for_each_display_mode(display_mode, connector) {
if (wsi_display_mode_matches_x(display_mode, mode))
return display_mode;
}
return NULL;
}
static VkResult
wsi_display_register_x_mode(struct wsi_device *wsi_device,
struct wsi_display_connector *connector,
xcb_randr_mode_info_t *x_mode,
bool preferred)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
struct wsi_display_mode *display_mode =
wsi_display_find_x_mode(wsi_device, connector, x_mode);
if (display_mode) {
display_mode->valid = true;
return VK_SUCCESS;
}
display_mode = vk_zalloc(wsi->alloc, sizeof (struct wsi_display_mode),
8, VK_SYSTEM_ALLOCATION_SCOPE_INSTANCE);
if (!display_mode)
return VK_ERROR_OUT_OF_HOST_MEMORY;
display_mode->connector = connector;
display_mode->valid = true;
display_mode->preferred = preferred;
display_mode->clock = (x_mode->dot_clock + 500) / 1000; /* kHz */
display_mode->hdisplay = x_mode->width;
display_mode->hsync_start = x_mode->hsync_start;
display_mode->hsync_end = x_mode->hsync_end;
display_mode->htotal = x_mode->htotal;
display_mode->hskew = x_mode->hskew;
display_mode->vdisplay = x_mode->height;
display_mode->vsync_start = x_mode->vsync_start;
display_mode->vsync_end = x_mode->vsync_end;
display_mode->vtotal = x_mode->vtotal;
display_mode->vscan = 0;
display_mode->flags = x_mode->mode_flags;
list_addtail(&display_mode->list, &connector->display_modes);
return VK_SUCCESS;
}
static struct wsi_display_connector *
wsi_display_get_output(struct wsi_device *wsi_device,
xcb_connection_t *connection,
xcb_randr_output_t output)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
struct wsi_display_connector *connector;
uint32_t connector_id;
xcb_window_t root = wsi_display_output_to_root(connection, output);
if (!root)
return NULL;
/* See if we already have a connector for this output */
connector = wsi_display_find_output(wsi_device, output);
if (!connector) {
xcb_atom_t connector_id_atom = 0;
/*
* Go get the kernel connector ID for this X output
*/
connector_id = wsi_display_output_to_connector_id(connection,
&connector_id_atom,
output);
/* Any X server with lease support will have this atom */
if (!connector_id) {
return NULL;
}
/* See if we already have a connector for this id */
connector = wsi_display_find_connector(wsi_device, connector_id);
if (connector == NULL) {
connector = wsi_display_alloc_connector(wsi, connector_id);
if (!connector) {
return NULL;
}
list_addtail(&connector->list, &wsi->connectors);
}
connector->output = output;
}
xcb_randr_get_screen_resources_cookie_t src =
xcb_randr_get_screen_resources(connection, root);
xcb_randr_get_output_info_cookie_t oic =
xcb_randr_get_output_info(connection, output, XCB_CURRENT_TIME);
xcb_randr_get_screen_resources_reply_t *srr =
xcb_randr_get_screen_resources_reply(connection, src, NULL);
xcb_randr_get_output_info_reply_t *oir =
xcb_randr_get_output_info_reply(connection, oic, NULL);
if (oir && srr) {
/* Get X modes and add them */
connector->connected =
oir->connection != XCB_RANDR_CONNECTION_DISCONNECTED;
wsi_display_invalidate_connector_modes(wsi_device, connector);
xcb_randr_mode_t *x_modes = xcb_randr_get_output_info_modes(oir);
for (int m = 0; m < oir->num_modes; m++) {
xcb_randr_mode_info_iterator_t i =
xcb_randr_get_screen_resources_modes_iterator(srr);
while (i.rem) {
xcb_randr_mode_info_t *mi = i.data;
if (mi->id == x_modes[m]) {
VkResult result = wsi_display_register_x_mode(
wsi_device, connector, mi, m < oir->num_preferred);
if (result != VK_SUCCESS) {
free(oir);
free(srr);
return NULL;
}
break;
}
xcb_randr_mode_info_next(&i);
}
}
}
free(oir);
free(srr);
return connector;
}
static xcb_randr_crtc_t
wsi_display_find_crtc_for_output(xcb_connection_t *connection,
xcb_window_t root,
xcb_randr_output_t output)
{
xcb_randr_get_screen_resources_cookie_t gsr_c =
xcb_randr_get_screen_resources(connection, root);
xcb_randr_get_screen_resources_reply_t *gsr_r =
xcb_randr_get_screen_resources_reply(connection, gsr_c, NULL);
if (!gsr_r)
return 0;
xcb_randr_crtc_t *rc = xcb_randr_get_screen_resources_crtcs(gsr_r);
xcb_randr_crtc_t idle_crtc = 0;
xcb_randr_crtc_t active_crtc = 0;
/* Find either a crtc already connected to the desired output or idle */
for (int c = 0; active_crtc == 0 && c < gsr_r->num_crtcs; c++) {
xcb_randr_get_crtc_info_cookie_t gci_c =
xcb_randr_get_crtc_info(connection, rc[c], gsr_r->config_timestamp);
xcb_randr_get_crtc_info_reply_t *gci_r =
xcb_randr_get_crtc_info_reply(connection, gci_c, NULL);
if (gci_r) {
if (gci_r->mode) {
int num_outputs = xcb_randr_get_crtc_info_outputs_length(gci_r);
xcb_randr_output_t *outputs =
xcb_randr_get_crtc_info_outputs(gci_r);
if (num_outputs == 1 && outputs[0] == output)
active_crtc = rc[c];
} else if (idle_crtc == 0) {
int num_possible = xcb_randr_get_crtc_info_possible_length(gci_r);
xcb_randr_output_t *possible =
xcb_randr_get_crtc_info_possible(gci_r);
for (int p = 0; p < num_possible; p++)
if (possible[p] == output) {
idle_crtc = rc[c];
break;
}
}
free(gci_r);
}
}
free(gsr_r);
if (active_crtc)
return active_crtc;
return idle_crtc;
}
VkResult
wsi_acquire_xlib_display(VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
Display *dpy,
VkDisplayKHR display)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
xcb_connection_t *connection = XGetXCBConnection(dpy);
struct wsi_display_connector *connector =
wsi_display_connector_from_handle(display);
xcb_window_t root;
/* XXX no support for multiple leases yet */
if (wsi->fd >= 0)
return VK_ERROR_INITIALIZATION_FAILED;
if (!connector->output) {
connector->output = wsi_display_connector_id_to_output(connection,
connector->id);
/* Check and see if we found the output */
if (!connector->output)
return VK_ERROR_INITIALIZATION_FAILED;
}
root = wsi_display_output_to_root(connection, connector->output);
if (!root)
return VK_ERROR_INITIALIZATION_FAILED;
xcb_randr_crtc_t crtc = wsi_display_find_crtc_for_output(connection,
root,
connector->output);
if (!crtc)
return VK_ERROR_INITIALIZATION_FAILED;
xcb_randr_lease_t lease = xcb_generate_id(connection);
xcb_randr_create_lease_cookie_t cl_c =
xcb_randr_create_lease(connection, root, lease, 1, 1,
&crtc, &connector->output);
xcb_randr_create_lease_reply_t *cl_r =
xcb_randr_create_lease_reply(connection, cl_c, NULL);
if (!cl_r)
return VK_ERROR_INITIALIZATION_FAILED;
int fd = -1;
if (cl_r->nfd > 0) {
int *rcl_f = xcb_randr_create_lease_reply_fds(connection, cl_r);
fd = rcl_f[0];
}
free (cl_r);
if (fd < 0)
return VK_ERROR_INITIALIZATION_FAILED;
wsi->fd = fd;
return VK_SUCCESS;
}
VkResult
wsi_get_randr_output_display(VkPhysicalDevice physical_device,
struct wsi_device *wsi_device,
Display *dpy,
RROutput output,
VkDisplayKHR *display)
{
xcb_connection_t *connection = XGetXCBConnection(dpy);
struct wsi_display_connector *connector =
wsi_display_get_output(wsi_device, connection, (xcb_randr_output_t) output);
if (connector)
*display = wsi_display_connector_to_handle(connector);
else
*display = VK_NULL_HANDLE;
return VK_SUCCESS;
}
vulkan: Add EXT_acquire_xlib_display [v5] This extension adds the ability to borrow an X RandR output for temporary use directly by a Vulkan application. For DRM, we use the Linux resource leasing mechanism. v2: Clean up xlib_lease detection * Use separate temporary '_xlib_lease' variable to hold the option value to avoid changin the type of a variable. * Use boolean expressions instead of additional if statements to compute resulting with_xlib_lease value. * Simplify addition of VK_USE_PLATFORM_XLIB_XRANDR_KHR to vulkan_wsi_args Suggested-by: Eric Engestrom <eric.engestrom@imgtec.com> Move mode list from wsi_display to wsi_display_connector Fix scope for wsi_display_mode and wsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v3: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Explicitly forbid multiple DRM leases. Making the code support this looks tricky and will require additional thought. Use xcb_randr_output_t throughout the internals of the implementation. Convert at the public API (wsi_get_randr_output_display). Clean up check for usable active_crtc (possible when only the desired output is connected to the crtc). Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v4: Move output resource fetching closer to use in wsi_display_get_output. This simplifies the error returns in earlier parts of the code a bit. Return VK_ERROR_INITIALIZATION_FAILED from wsi_acquire_xlib_display. Jason says this is the right error message. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v5: randr doesn't pass vscan over the wire, so we set vscan to 0 for randr-acquired modes, and test wsi modes for vscan <= 1 when comparing against randr modes. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
2018-02-09 15:45:58 +00:00
#endif
vulkan: add VK_EXT_display_control [v10] This extension provides fences and frame count information to direct display contexts. It uses new kernel ioctls to provide 64-bits of vblank sequence and nanosecond resolution. v2: Remove DRM_CRTC_SEQUENCE_FIRST_PIXEL_OUT flag. This has been removed from the proposed kernel API. Add NULL parameter to drmCrtcQueueSequence ioctl as we don't care what sequence the event was actually queued to. v3: Adapt to pthread clock switch to MONOTONIC v4: Fix scope for wsi_display_mode andwsi_display_connector allocs Suggested-by: Jason Ekstrand <jason@jlekstrand.net> v5: Adopt Jason Ekstrand's coding conventions Declare variables at first use, eliminate extra whitespace between types and names. Wrap lines to 80 columns. Use wsi_rel_to_abs_time helper function to convert relative timeouts to absolute timeouts without causing overflow. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v6: Change WSI fence wait function to return VkResult instead of bool. This makes the meaning of the return value easier to understand, and allows for the indication of failure. Also change the WSI fence wait function to take only absolute timeouts and not provide an option for a relative timeout. No users wanted relative timeouts, and it's simpler if that option isn't available. Terminate the DPMS property loop once we've found the property. Assert that the fence hasn't already been destroyed in wsi_display_fence_destroy. Rearrange the event handler function order in the file to place routines in an easier to find order. Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v7: Adapt to API changes for surface_get_capabilities v8: Use wsi->alloc in register_display_event so that callers don't have to dig out an allocator for us. v9: Fix a few minor formatting issues Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com> v10: Use wsi->alloc if none provided in wsi_display_fence_alloc. Now that drivers are expected to pass the allocator argument straight through from the application, we need to check those for NULL everywhere. Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
2017-06-16 05:00:56 +01:00
/* VK_EXT_display_control */
VkResult
wsi_display_power_control(VkDevice device,
struct wsi_device *wsi_device,
VkDisplayKHR display,
const VkDisplayPowerInfoEXT *display_power_info)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
struct wsi_display_connector *connector =
wsi_display_connector_from_handle(display);
int mode;
if (wsi->fd < 0)
return VK_ERROR_INITIALIZATION_FAILED;
switch (display_power_info->powerState) {
case VK_DISPLAY_POWER_STATE_OFF_EXT:
mode = DRM_MODE_DPMS_OFF;
break;
case VK_DISPLAY_POWER_STATE_SUSPEND_EXT:
mode = DRM_MODE_DPMS_SUSPEND;
break;
default:
mode = DRM_MODE_DPMS_ON;
break;
}
drmModeConnectorSetProperty(wsi->fd,
connector->id,
connector->dpms_property,
mode);
return VK_SUCCESS;
}
VkResult
wsi_register_device_event(VkDevice device,
struct wsi_device *wsi_device,
const VkDeviceEventInfoEXT *device_event_info,
const VkAllocationCallbacks *allocator,
struct wsi_fence **fence_p)
{
return VK_ERROR_FEATURE_NOT_PRESENT;
}
VkResult
wsi_register_display_event(VkDevice device,
struct wsi_device *wsi_device,
VkDisplayKHR display,
const VkDisplayEventInfoEXT *display_event_info,
const VkAllocationCallbacks *allocator,
struct wsi_fence **fence_p)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
struct wsi_display_fence *fence;
VkResult ret;
switch (display_event_info->displayEvent) {
case VK_DISPLAY_EVENT_TYPE_FIRST_PIXEL_OUT_EXT:
fence = wsi_display_fence_alloc(device, wsi_device, display, allocator);
if (!fence)
return VK_ERROR_OUT_OF_HOST_MEMORY;
ret = wsi_register_vblank_event(fence, wsi_device, display,
DRM_CRTC_SEQUENCE_RELATIVE, 1, NULL);
if (ret == VK_SUCCESS)
*fence_p = &fence->base;
else if (fence != NULL)
vk_free2(wsi->alloc, allocator, fence);
break;
default:
ret = VK_ERROR_FEATURE_NOT_PRESENT;
break;
}
return ret;
}
VkResult
wsi_get_swapchain_counter(VkDevice device,
struct wsi_device *wsi_device,
VkSwapchainKHR _swapchain,
VkSurfaceCounterFlagBitsEXT flag_bits,
uint64_t *value)
{
struct wsi_display *wsi =
(struct wsi_display *) wsi_device->wsi[VK_ICD_WSI_PLATFORM_DISPLAY];
struct wsi_display_swapchain *swapchain =
(struct wsi_display_swapchain *) wsi_swapchain_from_handle(_swapchain);
struct wsi_display_connector *connector =
wsi_display_mode_from_handle(swapchain->surface->displayMode)->connector;
if (wsi->fd < 0)
return VK_ERROR_INITIALIZATION_FAILED;
if (!connector->active) {
*value = 0;
return VK_SUCCESS;
}
int ret = drmCrtcGetSequence(wsi->fd, connector->crtc_id, value, NULL);
if (ret)
*value = 0;
return VK_SUCCESS;
}