167 Commits (a6122c3ec58b9f03fdf40be33d55bcb203c496ac)

Author	SHA1	Message	Date
Simon Ser	b4ad4671dc	render/vulkan: check for barrier array alloc failure	2 years ago
Simon Ser	bc7f8de842	render/vulkan: improve message on format prop error ... Use a more appropriate message on vkGetPhysicalDeviceImageFormatProperties2() error.	2 years ago
Simon Ser	338a9616bd	render/vulkan: simplify vulkan_format_props_query()	2 years ago
Simon Ser	a3874cac6c	render/vulkan: extract DMA-BUF format query to separate function	2 years ago
Simon Ser	21c4516838	render/vulkan: add 64-bit UNORM and SFLOAT formats	2 years ago
Simon Ser	9f938f7f2a	render/vulkan: add more packed formats	2 years ago
Simon Ser	171c9081d1	render/vulkan: add more 8 bits per channel formats	2 years ago
Simon Ser	ad165f7daf	render/vulkan: explain format mapping with DRM	2 years ago
Simon Ser	8b9a48c984	render/vulkan: drop "_format" in wlr_vk_format fields ... "format" is already in the name, no need to repeat ourselves. `format->vk_format` sounds a bit redundant.	2 years ago
Alexander Orzechowski	db0e962368	wlr_texture: Expose owning renderer	2 years ago
Simon Ser	d2238bf722	render/vulkan: import semaphore to DMA-BUF instead of blocking ... Right now the Vulkan renderer blocks until the frame is complete after rendering. This is necessary because Vulkan doesn't interoperate well with implicit sync we use everywhere else. Use the new kernel API to import a sync_file into a DMA-BUF to avoid blocking.	2 years ago
Simon Ser	10f543d579	render/vulkan: release stage buffers after command buffer completes ... We need to wait for the pending command buffer to complete before re-using stage buffers. Otherwise we'll overwrite the stage buffer with new contents before the texture is fully uploaded.	2 years ago
Simon Ser	2a414c896e	render/vulkan: destroy textures after command buffer completes ... We need to wait for any pending command buffer to complete before we're able to fully destroy a struct wlr_vk_texture: the Vulkan spec requires the VkDescriptorSet to be kept alive. So far we've done this in vulkan_end(), after blocking until the command buffer completes. We'll soon stop blocking, so move this logic in get_command_buffer(), where we check which commands buffers have completed in a non-blocking fashion.	2 years ago
Simon Ser	0730552e85	render/vulkan: drop unused fields from wlr_vk_format_modifier_props ... export_imported is never used, and dmabuf_flags is already checked in query_modifier_support().	2 years ago
Manuel Stoeckl	c02872e033	render/vulkan: align staging buffers for texture upload ... vkCmdCopyBufferToImage requires that the buffer offset be a multiple of the texel block size, which for single plane uncompressed formats is the same as the number of bytes per pixel. This commit adds an alignment parameter to vulkan_get_stage_span which ensures that the provided span (and the sequence of image copy operations derived which use it) have this alignment.	2 years ago
Manuel Stoeckl	3ed69b4946	render/vulkan: add support for RGB565 texture format ... Since this does not have a matching _SRGB-type vulkan format, add a new shader variant/pipeline to perform the sRGB->linear texture conversion.	2 years ago
Simon Ser	99134c26b8	render: stop pulling <wlr/backend.h> ... Let's just forward-declare struct wlr_backend instead. We need to fixup the Vulkan renderer: it needs makedev(), which got included by chance via <wlr/backend.h> → <wlr/backend/session.h> → <libudev.h>.	2 years ago
Manuel Stoeckl	b97ef59393	render/vulkan: update shm texture data in one batch	2 years ago
Simon Ser	d6e2ab5145	render/vulkan: detect device loss	2 years ago
Simon Ser	23540b5579	render: allow wlr_renderer_impl.begin to fail ... Make it return a bool to indicate success/failure. Adapt the various implementations to check errors.	2 years ago
zccrs	5b34fe5513	render/vulkan: add wlr_vk_texture_has_alpha ... Allow to get whether has alpha channel of the VkImage, it can help an optimization to disable blending when the texture doesn't have alpha. Because the VkFormat isn't enough because it's always set to VK_FORMAT_B8G8R8A8_SRGB or VK_FORMAT_R8G8B8A8_SRGB.	2 years ago
Simon Ser	6baf045590	render/texture: constify wlr_texture_update_from_buffer()	2 years ago
Simon Zeni	f3ba14e491	render/vulkan: remove hardcoded validation layers ... Users should use the VK_INSTANCE_LAYERS env var to set layers at runtime	2 years ago
Simon Ser	e1ba64c4b7	render/vulkan: use initializer for VkBufferImageCopy ... Missed that one it seems.	2 years ago
Simon Ser	d112c2d922	render/vulkan: wait for device to become idle in vulkan_destroy() ... It's not safe to destroy any resources which might still be in-use by the GPU. Wait for any asynchronous tasks to complete before destroying everything.	2 years ago
Simon Ser	f4f3c15c1e	render/vulkan: always wait for last stage to complete before rendering ... When we have multiple command buffers in flight, we need to make sure we don't start rendering before the previous texture uploads are complete.	2 years ago
Simon Ser	7fdcb45e66	render/vulkan: use command buffer pool for stage	2 years ago
Simon Ser	f1eae0eeeb	render/vulkan: add a command buffer pool ... Before re-using a VkCommandBuffer, we need to wait for its operations to complete. Right now we unconditionally wait for rendering to complete in vulkan_end(), however we have plans to fix this [1]. To fully avoid blocking, we need to handle multiple command buffers in flight at the same time (e.g. for multi-output, or for rendering followed by texture uploads). Implement a pool of command buffers. When we need to render, we pick a command buffer from the pool which has completed its operations. If we don't find one, try to allocate a new command buffer. If we don't have slots in the pool anymore, block like we did before. [1]: https://gitlab.freedesktop.org/wlroots/wlroots/-/merge_requests/3574	2 years ago
Simon Ser	a8a194d695	render/vulkan: switch to timeline semaphores ... Up until now we were using a VkFence for GPU-to-CPU synchronization. This has some limitations that become a blocker when trying to have multiple command buffers in flight at once (e.g. for multi-output). It's desirable to implement a command buffer pool [1], but VkFence cannot be used to track command buffer completion for individual subpasses. Let's just switch to timeline semaphores [2], which fix this issue, make synchronization a lot more ergonomic and are a core Vulkan 1.2 feature. [1]: https://gitlab.freedesktop.org/wlroots/wlroots/-/merge_requests/3802 [2]: https://www.khronos.org/blog/vulkan-timeline-semaphores	2 years ago
Simon Ser	f92d1499cd	render/vulkan: add helper to load command function pointer ... If NULL is returned by vkGetDeviceProcAddr(), either the driver is buggy, either the wlroots code is buggy. For a valid device and command name, drivers are not allowed to return NULL per the spec. This mirrors what the GLES2 renderer does in load_gl_proc().	2 years ago
zccrs	270914d379	render/vulkan: add some interfaces to allow compositors to integrate ... Added wlr_vk_renderer_get_* functions to allow get the VkInstance, VkPhysicalDevice, VkDevice, queue family of a wlr_vk_renderer. Added wlr_vk_renderer_get_current_image_attribs function to allow get the VkImage of current renderer buffer to use on compositors. Added wlr_renderer_is_vk function, it's like the wlr_renderer_is_gles2, returns true if the wlr_renderer is a wlr_vk_renderer. Added wlr_vk_image_get_attribs function to get a VkImage and it's extras information (e.g. a VkImageLayout and VkImageFormat of the VkImage) from a wlr_texture.	2 years ago
Simon Ser	6c31f3b078	render/vulkan: use wl_array for wlr_vk_shared_buffer.allocs ... Avoids the need to open-code the realloc() logic.	2 years ago
Simon Ser	9a5cc36818	render/vulkan: simplify texture VkPipelineShaderStageCreateInfo ... I simplified the quad VkPipelineShaderStageCreateInfo in [1], but missed the one for texture. [1]: https://gitlab.freedesktop.org/wlroots/wlroots/-/merge_requests/3823	2 years ago
Simon Ser	cfb768aab9	render/vulkan: use struct initializers ... Instead of filling the fields one by one, use a struct initializer. This avoids repeating the name of the variable and is more consistent with the wlroots code style.	2 years ago
Simon Ser	7f5180af77	render/vulkan: simplify extension checks ... find_extensions() is clunky to use when checking only a single extension, and it's surprising that it returns NULL on success. Simplify by replacing it with a check_extension() function which just checks whether a single extension is in the list.	2 years ago
Simon Ser	2ac2835a1f	render/vulkan: remove exts arg from vulkan_instance_create()	2 years ago
David96	7a42392da2	render/vulkan: add caching to vulkan_read_pixels	2 years ago
Simon Ser	9e5ad7a845	render/vulkan: add missing entries in vulkan_strerror() ... And update the sort order to follow Khronos' <vulkan/vulkan_core.h>.	2 years ago
Gentaiii	11192e6930	Fixed false allocation failed	2 years ago
Simon Ser	221ee83d44	render/vulkan: drop wlr_vk_instance.extensions ... This was unused. Even if it was, it'd be better to have bool fields instead.	2 years ago
John Lindgren	ab8341975e	render/vulkan: Fix type-punned pointer warning/error	2 years ago
David96	f885aa0fbe	render/vulkan: implement vulkan_preferred_read_format	2 years ago
David96	dce1372e35	render/vulkan: Implement vulkan_read_pixels	2 years ago
Simon Ser	6832ae14aa	render: drop wlr_renderer_read_pixels() flags ... These are unused.	2 years ago
Simon Ser	e4efb6b601	render/vulkan: use addon for wlr_vk_texture ... Avoids having to walk the list of all textures.	2 years ago
Simon Ser	7899b97d20	render/vulkan: use addon for wlr_vk_render_buffer ... Avoids having to walk the list containing all of the render buffers.	2 years ago
Simon Ser	8d6d2101bc	render/vulkan: add error handling for vulkan_find_mem_type() ... The function returns -1 if it doesn't find a suitable memory type.	2 years ago
Simon Ser	bca60c4eec	render/vulkan: improve format logging	2 years ago
Simon Ser	8026953df7	render/vulkan: remove unnecessary casts ... No need to go back to the generic wlr_renderer/wlr_texture when passing a Vulkan renderer/texture to an internal function.	2 years ago
Simon Ser	30769723c1	render/vulkan: fix format features check for shm textures ... We were checking whether any of the features was supported. We need to check if all of them are. This makes the check consistent with query_modifier_support() above.	2 years ago
Simon Ser	ed018a3ee2	render/vulkan: drop wlr_vk_device.extensions ... We don't need to store the list of enabled extensions. While at it, rename variables to be less confusing.	2 years ago
Simon Ser	e19007dede	render/vulkan: drop ext params from vulkan_device_create() ... These are unused.	2 years ago
Bernardo Meurer	eee0f5e71f	render/vulkan: fix memory type detection ... From a comment by emersion: > There is a logic error here: we pass 0xFFFFFFFF to vulkan_find_mem_type, which > returns an index, and then we logical-and that with mem_reqs.memoryTypeBits. > Instead we should pass mem_reqs.memoryTypeBits to vulkan_find_mem_type and use > the result for the memoryTypeIndex field directly. Ideally checking for errors > (-1 return value) in case no memory type is suitable. Closes: #3470	2 years ago
Simon Ser	98cf38601f	render: replace wlr_texture_write_pixels with update_from_buffer ... This lets the renderer handle the wlr_buffer directly, just like it does in texture_from_buffer. This also allows the renderer to batch the rectangle updates, and update more than the damage region if desirable (e.g. too many rects), so can be more efficient.	2 years ago
Simon Ser	29291cb47c	render/texture: drop wlr_texture_is_opaque ... Whether a texture is opaque or not doesn't depend on the renderer at all, it just depends on the source buffer. Instead of forcing all renderers to implement wlr_texture_impl.is_opaque, let's move this in common code and use the wlr_buffer format to know whether a texture will be opaque.	3 years ago
bi4k8	f295aff762	Add missing &	3 years ago
bi4k8	35d1bcff6f	render/vulkan/pixel_format: do not leak props->... ... `vulkan_format_props_query` calls `query_modifier_support` which initializes fields of `props` with allocated memory. this memory is leaked if `query_modifier_support` does not find a supported format and shmtex is not supported, as in this case `add_fmt_props` ends up being false and the allocated fields of `props` are never freed.	3 years ago
Simon Ser	96b594110d	matrix: remove wlr_matrix_projection() ... 69477051cc ("matrix: deprecate wlr_matrix_projection") marked it as deprecated. 1 year later, we can now remove it from our public API.	3 years ago
zccrs	48c811ffb6	render/vulkan: fix undefined reference to wlr_texture_is_vk ... Added the implement of wlr_texture_is_vk. Fixes: https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3419	3 years ago
Simon Ser	d8d30463ac	render/vulkan: log physical device driver name ... This can be useful to figure out why a required feature is missing, e.g. as in [1]. We check VK_EXT_physical_device_drm availability after printing the driver name. [1]: https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3358	3 years ago
nyorain	9988eb3378	vulkan: Fix imported image layout	3 years ago
Joshua Ashton	f132d66816	render/vulkan: Optimize vertex shader ... This ends up being a horrible global load: s_getpc_b64 s[4:5] // 000000000000: BE841C80 v_add_u32 v0, s2, v0 // 000000000004: 68000002 v_sub_co_u32 v1, vcc, 0, v0 // 000000000008: 34020080 v_max_i32 v1, v0, v1 // 00000000000C: 1A020300 v_and_b32 v1, 3, v1 // 000000000010: 26020283 v_cmp_lt_i32 s[0:1], v0, 0 // 000000000014: D0C10000 00010100 v_sub_co_u32 v0, vcc, 0, v1 // 00000000001C: 34000280 v_cndmask_b32 v0, v1, v0, s[0:1] // 000000000020: D1000000 00020101 v_lshlrev_b32 v1, 3, v0 // 000000000028: 24020083 v_mad_u32_u24 v0, v0, 8, 4 // 00000000002C: D1C30000 02111100 v_min_u32 v1, 32, v1 // 000000000034: 1C0202A0 v_min_u32 v0, 32, v0 // 000000000038: 1C0000A0 s_getpc_b64 s[0:1] // 00000000003C: BE801C00 s_add_u32 s0, s0, 0x0000003c // 000000000040: 8000FF00 0000003C s_addc_u32 s1, s1, 0 // 000000000048: 82018001 global_load_dword v1, v[1:2], s[0:1] // 00000000004C: DC508000 01000001 global_load_dword v0, v[0:1], s[0:1] // 000000000054: DC508000 00000000 v_mov_b32 v2, 0 // 00000000005C: 7E040280 v_mov_b32 v3, 1.0 // 000000000060: 7E0602F2 s_waitcnt vmcnt(0) // 000000000064: BF8C0F70 exp pos0, v1, v0, v2, v3 done // 000000000068: C40008CF 03020001 exp param0, off, off, off, off // 000000000070: C4000200 00000000 s_endpgm // 000000000078: BF810000 v_cndmask_b32 v0, s0, v0, vcc // 00000000007C: 00000000 v_cndmask_b32 v0, s0, v0, vcc // 000000000080: 00000000 v_add_f16 v192, s0, v0 // 000000000084: 3F800000 v_cndmask_b32 v0, s0, v0, vcc // 000000000088: 00000000 v_add_f16 v192, s0, v0 // 00000000008C: 3F800000 v_add_f16 v192, s0, v0 // 000000000090: 3F800000 v_cndmask_b32 v0, s0, v0, vcc // 000000000094: 00000000 v_add_f16 v192, s0, v0 // 000000000098: 3F800000 v_cndmask_b32 v0, s0, v0, vcc // 00000000009C: 00000000 With some bit magic, we can get something much nicer: v_add_u32 v0, s2, v0 // 000000000000: 68000002 v_add_u32 v1, 1, v0 // 000000000004: 68020081 v_and_b32 v1, 2, v1 // 000000000008: 26020282 v_cvt_f32_i32 v1, v1 // 00000000000C: 7E020B01 v_mul_f32 v1, 0.5, v1 // 000000000010: 0A0202F0 v_and_b32 v0, 2, v0 // 000000000014: 26000082 v_cvt_f32_i32 v0, v0 // 000000000018: 7E000B00 v_mul_f32 v0, 0.5, v0 // 00000000001C: 0A0000F0 v_mov_b32 v2, 0 // 000000000020: 7E040280 v_mov_b32 v3, 1.0 // 000000000024: 7E0602F2 exp pos0, v1, v0, v2, v3 done // 000000000028: C40008CF 03020001 exp param0, off, off, off, off // 000000000030: C4000200 00000000 s_endpgm // 000000000038: BF810000 The above output was based on just shoving it in ShaderPlayground -- I was not able to use pipeline feedback as I was unable to get RenderDoc working due to the EXT_physical_device_drm requirement. I additionally considered using >> 1 instead of * 0.5, but AMD has dedicated modifiers to merge a * 0.5, * 2.0, etc in a single instruction. (Albeit, not taken advantage of in the code above, but might with ACO) Signed-off-by: Joshua Ashton <joshua@froggi.es>	3 years ago
Simon Ser	5332935afc	render/vulkan: quiet glslangValidator ... This suppresses the output filename printed to stdout. Errors and warnings should still be printed to stderr as usual.	3 years ago
Simon Ser	a04cfca4da	Remove support for DMA-BUF flags ... They are never used in practice, which makes all of our flag handling effectively dead code. Also, APIs such as KMS don't provide a good way to deal with the flags. Let's just fail the DMA-BUF import when clients provide flags.	3 years ago
Simon Ser	36cf387427	render/vulkan: check vulkan-headers dependency ... There's no pkg-config file we can check for sadly, so check vulkan/vulkan.h as a fallback. Closes: https://github.com/swaywm/wlroots/issues/3272	3 years ago
Joshua Ashton	b62ce3c3c8	render/vulkan: Use image view swizzles instead of shader hack ... Signed-off-by: Joshua Ashton <joshua@froggi.es>	3 years ago
nyorain	8e34692250	render/vulkan: add Vulkan renderer ... This new renderer is implemented with the existing wlr_renderer API (which is known to be sub-optimal for some operations). It's not used by default, but users can opt-in by setting WLR_RENDERER=vulkan. The renderer depends on VK_EXT_image_drm_format_modifier and VK_EXT_physical_device_drm. Co-authored-by: Simon Ser <contact@emersion.fr> Co-authored-by: Jan Beich <jbeich@FreeBSD.org>	3 years ago