Cleaned up revert of remaining changes.
Subsystems (5)
Maarten Lankhorst
maintainercover Cleaned up revert of remaining changes.
From: Maarten Lankhorst <[email protected]>
All reverts are made simpler, and the header updates are squashed to a single commit.
This makes reviewing a lot easier.
Maarten Lankhorst (9):
drm/i915/display: Revert all before remaining changes to make xe
compile
drm/i915/display: Revert remaining changes to make xe compile drm/xe: Update headers to be more compatible with i915 drm/atomic: Add drm_atomic_helper_set_fence_deadline() function drm/i915: Implement fence->set_deadline.
FIXME drm/i915/display: Use fence->set_deadline to boost requests drm/i915/display: Use i915_gem_object_get_dma_address to get dma
address
FIXME drm/i915/display: Minimal changes to fbdev to make xe work FIXME drm/i915/display: Make intel_fb.c code compatible with xe
drivers/gpu/drm/drm_atomic_helper.c | 15 ++- .../gpu/drm/i915/display/intel_atomic_plane.c | 70 +++--------- drivers/gpu/drm/i915/display/intel_cursor.c | 12 +-- drivers/gpu/drm/i915/display/intel_display.c | 37 +------ .../drm/i915/display/intel_display_types.h | 4 - drivers/gpu/drm/i915/display/intel_fb.c | 89 ++++++---------- drivers/gpu/drm/i915/display/intel_fbc.c | 13 ++- drivers/gpu/drm/i915/display/intel_fbdev.c | 100 ++++++------------ .../gpu/drm/i915/display/intel_frontbuffer.c | 7 +- .../gpu/drm/i915/display/intel_frontbuffer.h | 4 - .../drm/i915/display/skl_universal_plane.c | 4 - drivers/gpu/drm/i915/i915_request.c | 9 ++ .../compat-i915-headers/gem/i915_gem_lmem.h | 2 + .../compat-i915-headers/gem/i915_gem_mman.h | 17 +++ .../compat-i915-headers/gem/i915_gem_object.h | 76 +++++++++++++ .../gem/i915_gem_object_frontbuffer.h | 4 +- .../gpu/drm/xe/compat-i915-headers/i915_drv.h | 1 + .../xe/compat-i915-headers/i915_gem_stolen.h | 13 ++- .../gpu/drm/xe/compat-i915-headers/i915_vma.h | 6 ++ include/drm/drm_atomic_helper.h | 3 +- 20 files changed, 234 insertions(+), 252 deletions(-) create mode 100644 drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_lmem.h create mode 100644 drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_mman.h create mode 100644 drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_object.h
Signature · 1 line
2.40.1
Maarten Lankhorst
maintainerPATCH 1/9 drm/i915/display: Revert all before remaining changes to make xe compile
From: Maarten Lankhorst <[email protected]>
drm/xe/display: Allow scanout of bos that can be migrated to lmem drm/i915/display: Use intel_bo_to_drm_bo instead of obj->base
Can probably be slightly ignored by squashing with next patch, at least for intel_bo_to_drm_bo part.
--- drivers/gpu/drm/i915/display/intel_fb.c | 9 ++-- drivers/gpu/drm/i915/display/intel_fbdev.c | 48 ++++++++-------------- 2 files changed, 22 insertions(+), 35 deletions(-)
drivers/gpu/drm/i915/display/intel_fb.c +4 -5
@@ -1964,10 +1964,10 @@ static int intel_user_framebuffer_dirty(struct drm_framebuffer *fb,if (!atomic_read (&front->bits))return 0 ; -if (dma_resv_test_signaled (intel_bo_to_drm_bo (obj)->resv,dma_resv_usage_rw (false))) +if (dma_resv_test_signaled (obj->base.resv,dma_resv_usage_rw (false))) goto flush; - ret =dma_resv_get_singleton (intel_bo_to_drm_bo (obj)->resv,dma_resv_usage_rw (false), + ret =dma_resv_get_singleton (obj->base.resv,dma_resv_usage_rw (false), &fence);if (ret || !fence) goto flush;
@@ -2227,9 +2227,8 @@ intel_user_framebuffer_create(struct drm_device *dev,return ERR_PTR (-ENOENT ); obj =gem_to_xe_bo (gem); -/* Require vram placement or dma-buf import */ -if (IS_DGFX (i915) && - !xe_bo_can_migrate (gem_to_xe_bo (gem),XE_PL_VRAM0 ) && +/* Require vram exclusive objects, but allow dma-buf imports */ +if (IS_DGFX (i915) && obj->flags &XE_BO_CREATE_SYSTEM_BIT && obj->ttm.type != ttm_bo_type_sg) {drm_gem_object_put (gem);return ERR_PTR (-EREMOTE );
drivers/gpu/drm/i915/display/intel_fbdev.c +18 -30
79 lines not on this page · raw message
Signature · 1 line
2.40.1
Maarten Lankhorst
maintainerPATCH 2/9 drm/i915/display: Revert remaining changes to make xe compile
In reply to Maarten LankhorstShould be split out instead. This should effectively revert the complete patch.
--- .../gpu/drm/i915/display/intel_atomic_plane.c | 25 +---- drivers/gpu/drm/i915/display/intel_cursor.c | 8 +- drivers/gpu/drm/i915/display/intel_display.c | 35 ------- .../drm/i915/display/intel_display_types.h | 4 - drivers/gpu/drm/i915/display/intel_fb.c | 81 +++------------- drivers/gpu/drm/i915/display/intel_fbc.c | 13 ++- drivers/gpu/drm/i915/display/intel_fbdev.c | 95 ++----------------- .../gpu/drm/i915/display/intel_frontbuffer.c | 7 +- .../gpu/drm/i915/display/intel_frontbuffer.h | 4 - .../drm/i915/display/skl_universal_plane.c | 4 - 10 files changed, 40 insertions(+), 236 deletions(-)
drivers/gpu/drm/i915/display/intel_atomic_plane.c +4 -21
@@ -1033,11 +1033,10 @@ static int intel_prepare_plane_fb(struct drm_plane *_plane, struct drm_plane_state *_new_plane_state) { - struct intel_plane_state *new_plane_state = -to_intel_plane_state (_new_plane_state); -#ifdef I915 struct i915_sched_attr attr = { .priority = I915_PRIORITY_DISPLAY };struct intel_plane *plane =to_intel_plane (_plane); +struct intel_plane_state *new_plane_state = +to_intel_plane_state (_new_plane_state);struct intel_atomic_state *state =to_intel_atomic_state (new_plane_state->uapi.state);struct drm_i915_private *dev_priv =to_i915 (plane->base.dev);
@@ -1133,22 +1132,6 @@ intel_prepare_plane_fb(struct drm_plane *_plane,intel_plane_unpin_fb (new_plane_state);return ret; -#else -int ret; - -if (!intel_fb_obj (new_plane_state->hw.fb)) -return 0 ; - - ret =intel_plane_pin_fb (new_plane_state); -if (ret) -return ret; - - ret =drm_gem_plane_helper_prepare_fb (_plane, _new_plane_state); -if (ret) -intel_plane_unpin_fb (new_plane_state); - -return ret; -#endif }/**
@@ -1164,9 +1147,9 @@ intel_cleanup_plane_fb(struct drm_plane *plane, {struct intel_plane_state *old_plane_state =to_intel_plane_state (_old_plane_state); -__maybe_unused struct intel_atomic_state *state = +struct intel_atomic_state *state =to_intel_atomic_state (old_plane_state->uapi.state); -__maybe_unused struct drm_i915_private *dev_priv =to_i915 (plane->dev); +struct drm_i915_private *dev_priv =to_i915 (plane->dev);struct drm_i915_gem_object *obj =intel_fb_obj (old_plane_state->hw.fb);if (!obj)
drivers/gpu/drm/i915/display/intel_cursor.c +3 -5
@@ -30,17 +30,15 @@ static const u32 intel_cursor_formats[] = {static u32 intel_cursor_base (const struct intel_plane_state *plane_state) { -__maybe_unused struct drm_i915_private *dev_priv = +struct drm_i915_private *dev_priv =to_i915 (plane_state->uapi.plane->dev); -__maybe_unused const struct drm_framebuffer *fb = plane_state->hw.fb; -__maybe_unused const struct drm_i915_gem_object *obj =intel_fb_obj (fb); +const struct drm_framebuffer *fb = plane_state->hw.fb; +const struct drm_i915_gem_object *obj =intel_fb_obj (fb);u32 base; -#ifdef I915 if (DISPLAY_INFO (dev_priv)->cursor_needs_physical) base =sg_dma_address (obj->mm.pages->sgl);else -#endif base =intel_plane_ggtt_offset (plane_state);return base + plane_state->view.color_plane[0 ].offset;
drivers/gpu/drm/i915/display/intel_display.c +0 -35
@@ -42,10 +42,8 @@#include <drm/drm_probe_helper.h> #include <drm/drm_rect.h> -#ifdef I915 #include "gem/i915_gem_lmem.h" #include "gem/i915_gem_object.h" -#endif #include "g4x_dp.h" #include "g4x_hdmi.h"
@@ -7119,39 +7117,6 @@ static void intel_atomic_cleanup_work(struct work_struct *work)intel_atomic_helper_free_state (i915); } -#ifndef I915 -static int i915_gem_object_read_from_page (struct xe_bo *bo, -u32 ofs,u64 *ptr,u32 size) -{ -struct ttm_bo_kmap_obj map; -void *virtual; -bool is_iomem; -int ret; - -XE_WARN_ON (size !=8 ); - - ret =xe_bo_lock (bo, true); -if (ret) -return ret; - - ret =ttm_bo_kmap (&bo->ttm, ofs >>PAGE_SHIFT ,1 , &map); -if (ret) - goto out_unlock; - - ofs &= ~PAGE_MASK ; - virtual =ttm_kmap_obj_virtual (&map, &is_iomem); -if (is_iomem) - *ptr =readq ((void __iomem *)(virtual + ofs)); -else - *ptr = *(u64 *)(virtual + ofs); - -ttm_bo_kunmap (&map); -out_unlock: -xe_bo_unlock (bo); -return ret; -} -#endif -static void intel_atomic_prepare_plane_clear_colors (struct intel_atomic_state *state) {struct drm_i915_private *i915 =to_i915 (state->base.dev);
drivers/gpu/drm/i915/display/intel_display_types.h +0 -4
@@ -2115,11 +2115,7 @@ intel_crtc_needs_color_update(const struct intel_crtc_state *crtc_state)static inline u32 intel_plane_ggtt_offset (const struct intel_plane_state *plane_state) { -#ifdef I915 return i915_ggtt_offset (plane_state->ggtt_vma); -#else -return plane_state->ggtt_vma->node.start; -#endif } static inline struct intel_frontbuffer *
drivers/gpu/drm/i915/display/intel_fb.c +12 -69
208 lines not on this page · raw message
drivers/gpu/drm/i915/display/intel_fbc.c +9 -4
@@ -453,7 +453,6 @@ static void g4x_fbc_program_cfb(struct intel_fbc *fbc)intel_de_write (i915,DPFC_CB_BASE ,i915_gem_stolen_node_offset (&fbc->compressed_fb)); - }static const struct intel_fbc_funcs g4x_fbc_funcs = {
@@ -612,8 +611,10 @@ static void ivb_fbc_activate(struct intel_fbc *fbc)else if (DISPLAY_VER (i915) ==9 )skl_fbc_program_cfb_stride (fbc); +#ifdef I915 if (intel_gt_support_legacy_fencing (to_gt (i915)))snb_fbc_program_fence (fbc); +#endif intel_de_write (i915,ILK_DPFC_CONTROL (fbc->id),DPFC_CTL_EN |ivb_dpfc_ctl (fbc));
@@ -800,7 +801,6 @@ static int intel_fbc_alloc_cfb(struct intel_fbc *fbc,drm_dbg_kms (&i915->drm,"reserved %llu bytes of contiguous stolen space for FBC, limit: %d\n" ,i915_gem_stolen_node_size (&fbc->compressed_fb), fbc->limit); -return 0 ; err_llb:
@@ -978,7 +978,7 @@ static void intel_fbc_update_state(struct intel_atomic_state *state,struct intel_crtc *crtc, struct intel_plane *plane) { - struct drm_i915_private *i915 = to_i915(state->base.dev); + __maybe_unused struct drm_i915_private *i915 = to_i915(state->base.dev);const struct intel_crtc_state *crtc_state =intel_atomic_get_new_crtc_state (state, crtc); const struct intel_plane_state *plane_state =
@@ -993,7 +993,7 @@ static void intel_fbc_update_state(struct intel_atomic_state *state,/* FBC1 compression interval: arbitrary choice of 1 second */ fbc_state->interval =drm_mode_vrefresh (&crtc_state->hw.adjusted_mode); - +#ifdef I915 fbc_state->fence_y_offset =intel_plane_fence_y_offset (plane_state); drm_WARN_ON(&i915->drm, plane_state->flags & PLANE_HAS_FENCE &&
@@ -1002,6 +1002,7 @@ static void intel_fbc_update_state(struct intel_atomic_state *state,if (plane_state->flags &PLANE_HAS_FENCE ) fbc_state->fence_id =i915_vma_fence_id (plane_state->ggtt_vma); else +#endif fbc_state->fence_id =-1 ; fbc_state->cfb_stride =intel_fbc_cfb_stride (plane_state);
@@ -1011,6 +1012,7 @@ static void intel_fbc_update_state(struct intel_atomic_state *state,static bool intel_fbc_is_fence_ok (const struct intel_plane_state *plane_state) { +#ifdef I915 struct drm_i915_private *i915 =to_i915 (plane_state->uapi.plane->dev);/*
@@ -1028,6 +1030,9 @@ static bool intel_fbc_is_fence_ok(const struct intel_plane_state *plane_state)return DISPLAY_VER (i915) >=9 || (plane_state->flags &PLANE_HAS_FENCE &&i915_vma_fence_id (plane_state->ggtt_vma) !=-1 ); +#else +return true; +#endif }static bool intel_fbc_is_cfb_ok (const struct intel_plane_state *plane_state)
drivers/gpu/drm/i915/display/intel_fbdev.c +8 -87
212 lines not on this page · raw message
drivers/gpu/drm/i915/display/intel_frontbuffer.c +4 -3
@@ -190,7 +190,7 @@ void __intel_fb_flush(struct intel_frontbuffer *front,enum fb_op_origin origin, unsigned int frontbuffer_bits) { - struct drm_i915_private *i915 = intel_bo_to_i915((front->obj)); + struct drm_i915_private *i915 = intel_bo_to_i915(front->obj);if (origin ==ORIGIN_CS ) {spin_lock (&i915->display.fb_tracking.lock);
@@ -259,6 +259,7 @@ static void frontbuffer_release(struct kref *ref)drm_WARN_ON (&intel_bo_to_i915 (obj)->drm,atomic_read (&front->bits));i915_ggtt_clear_scanout (obj); + ret =i915_gem_object_set_frontbuffer (obj,NULL );drm_WARN_ON (&intel_bo_to_i915 (obj)->drm, ret);spin_unlock (&intel_bo_to_i915 (obj)->display.fb_tracking.lock);
@@ -289,12 +290,12 @@ intel_frontbuffer_get(struct drm_i915_gem_object *obj) frontbuffer_retire, I915_ACTIVE_RETIRE_SLEEPS);INIT_WORK (&front->flush_work, intel_frontbuffer_flush_work); +spin_lock (&i915->display.fb_tracking.lock); cur =i915_gem_object_set_frontbuffer (obj, front);spin_unlock (&i915->display.fb_tracking.lock);if (cur != front)kfree (front); -return cur; }
@@ -337,7 +338,7 @@ void intel_frontbuffer_track(struct intel_frontbuffer *old, }if (new) { -drm_WARN_ON (&intel_bo_to_i915 (old->obj)->drm, +drm_WARN_ON (&intel_bo_to_i915 (new->obj)->drm,atomic_read (&new->bits) & frontbuffer_bits);atomic_or (frontbuffer_bits, &new->bits); }
drivers/gpu/drm/i915/display/intel_frontbuffer.h +0 -4
@@ -28,10 +28,6 @@#include <linux/bits.h> #include <linux/kref.h> -#ifndef I915 -#include "xe_bo.h" -#endif -#include "i915_active_types.h" struct drm_i915_private ;
drivers/gpu/drm/i915/display/skl_universal_plane.c +0 -4
@@ -2453,15 +2453,11 @@ skl_get_initial_plane_config(struct intel_crtc *crtc, fb->modifier =DRM_FORMAT_MOD_LINEAR ;break ;case PLANE_CTL_TILED_X : -#ifdef I915 plane_config->tiling =I915_TILING_X ; -#endif fb->modifier =I915_FORMAT_MOD_X_TILED ;break ;case PLANE_CTL_TILED_Y : -#ifdef I915 plane_config->tiling =I915_TILING_Y ; -#endif if (val &PLANE_CTL_RENDER_DECOMPRESSION_ENABLE )if (DISPLAY_VER (dev_priv) >=14 ) fb->modifier =I915_FORMAT_MOD_4_TILED_MTL_RC_CCS ;
Signature · 1 line
2.40.1
Maarten Lankhorst
maintainerPATCH 3/9 drm/xe: Update headers to be more compatible with i915
In reply to Maarten LankhorstAdd more compatibility changes to decrease delta with i915.
--- .../compat-i915-headers/gem/i915_gem_lmem.h | 2 + .../compat-i915-headers/gem/i915_gem_mman.h | 17 +++++ .../compat-i915-headers/gem/i915_gem_object.h | 76 +++++++++++++++++++ .../gem/i915_gem_object_frontbuffer.h | 4 +- .../gpu/drm/xe/compat-i915-headers/i915_drv.h | 1 + .../xe/compat-i915-headers/i915_gem_stolen.h | 13 +++- .../gpu/drm/xe/compat-i915-headers/i915_vma.h | 6 ++ 7 files changed, 114 insertions(+), 5 deletions(-) create mode 100644 drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_lmem.h create mode 100644 drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_mman.h create mode 100644 drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_object.h
drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_lmem.h new file +2 -0
@@ -0,0 +1,2 @@ +/* Empty */ +
drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_mman.h new file +17 -0
@@ -0,0 +1,17 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2023 Intel Corporation +*/ + +#ifndef _I915_GEM_MMAN_H_ +#define _I915_GEM_MMAN_H_ + +#include "xe_bo_types.h" +#include <drm/drm_prime.h> + +static inline int i915_gem_fb_mmap (struct xe_bo *bo,struct vm_area_struct *vma) +{ +return drm_gem_prime_mmap (&bo->ttm.base, vma); +} + +#endif
drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_object.h new file +76 -0
76 lines not on this page · raw message
drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_object_frontbuffer.h +2 -2
@@ -3,8 +3,8 @@* Copyright © 2022 Intel Corporation */ -#ifndef _I915_GEM_OBJECT_H_ -#define _I915_GEM_OBJECT_H_ +#ifndef _I915_GEM_OBJECT_FB_H_ +#define _I915_GEM_OBJECT_FB_H_#define i915_gem_object_get_frontbuffer (obj) NULL#define i915_gem_object_set_frontbuffer (obj, front) (front)
drivers/gpu/drm/xe/compat-i915-headers/i915_drv.h +1 -0
@@ -202,6 +202,7 @@ static inline void intel_runtime_pm_put(struct xe_runtime_pm *pm, bool wakeref)#define intel_uncore_arm_unclaimed_mmio_detection (x) do { } while (0)#define i915_sw_fence_commit (x) do { } while (0) +#define i915_sw_fence_await_reservation (...) (0)#define with_intel_runtime_pm (rpm, wf) \ for ((wf) = intel_runtime_pm_get(rpm); (wf); \
drivers/gpu/drm/xe/compat-i915-headers/i915_gem_stolen.h +10 -3
@@ -1,5 +1,5 @@ -#ifndef _I915_GEM_OBJECT_H_ -#define _I915_GEM_OBJECT_H_ +#ifndef _I915_GEM_STOLEN_H_ +#define _I915_GEM_STOLEN_H_#include "xe_ttm_stolen_mgr.h" #include "xe_res_cursor.h"
@@ -70,7 +70,14 @@ static inline u32 i915_gem_stolen_node_offset(struct i915_stolen_fb *fb)/* Used for < gen4. These are not supported by Xe */ #define i915_gem_stolen_area_address (xe) (!WARN_ON(1))/* Used for gen9 specific WA. Gen9 is not supported by Xe */ -#define i915_gem_stolen_area_size (xe) (!WARN_ON(1)) +static inline u64 i915_gem_stolen_area_size (struct xe_device *xe) +{ +struct ttm_resource_manager *ttm_mgr =ttm_manager_type (&xe->ttm,XE_PL_STOLEN ); + +if (!ttm_mgr) +return 0 ; +return ttm_mgr->size; +}#define i915_gem_stolen_node_address (xe, fb) (xe_ttm_stolen_gpu_offset(xe) + \ i915_gem_stolen_node_offset(fb))
drivers/gpu/drm/xe/compat-i915-headers/i915_vma.h +6 -0
@@ -19,4 +19,10 @@ struct i915_vma {#define i915_ggtt_clear_scanout (bo) do { } while (0)#define i915_vma_fence_id (vma) -1 + +static inline u32 i915_ggtt_offset (const struct i915_vma *vma) +{ +return vma->node.start; +} +#endif
Signature · 1 line
2.40.1
Maarten Lankhorst
maintainerPATCH 4/9 drm/atomic: Add drm_atomic_helper_set_fence_deadline() function
In reply to Maarten Lankhorsti915 has an i915 specific boost. Use set_fence_deadline instead.
--- drivers/gpu/drm/drm_atomic_helper.c | 15 +++++++++++---- include/drm/drm_atomic_helper.h | 3 ++- 2 files changed, 13 insertions(+), 5 deletions(-)
drivers/gpu/drm/drm_atomic_helper.c +11 -4
@@ -1520,12 +1520,16 @@ void drm_atomic_helper_commit_modeset_enables(struct drm_device *dev, }EXPORT_SYMBOL (drm_atomic_helper_commit_modeset_enables); -/* +/** +* drm_atomic_helper_set_fence_deadline - update fence deadlines. +* @dev: DRM device +* @state: atomic state object with new state structures +* * For atomic updates which touch just a single CRTC, calculate the time of the * next vblank, and inform all the fences of the deadline. */ -static void set_fence_deadline (struct drm_device *dev, -struct drm_atomic_state *state) +void drm_atomic_helper_set_fence_deadline (struct drm_device *dev, +struct drm_atomic_state *state) {struct drm_crtc *crtc;struct drm_crtc_state *new_crtc_state;
@@ -1560,6 +1564,7 @@ static void set_fence_deadline(struct drm_device *dev,dma_fence_set_deadline (new_plane_state->fence, vbltime); } } +EXPORT_SYMBOL (drm_atomic_helper_set_fence_deadline);/** * drm_atomic_helper_wait_for_fences - wait for fences stashed in plane state
@@ -1580,6 +1585,8 @@ static void set_fence_deadline(struct drm_device *dev,* to wait **before** we do anything that can't be easily rolled back. That is * before we call drm_atomic_helper_swap_state(). * +* Calls drm_atomic_helper_set_fence_deadline(). +* * Returns zero if success or < 0 if dma_fence_wait() fails. */ int drm_atomic_helper_wait_for_fences(struct drm_device *dev,
@@ -1590,7 +1597,7 @@ int drm_atomic_helper_wait_for_fences(struct drm_device *dev,struct drm_plane_state *new_plane_state;int i, ret; -set_fence_deadline (dev, state); +drm_atomic_helper_set_fence_deadline (dev, state);for_each_new_plane_in_state (state, plane, new_plane_state, i) {if (!new_plane_state->fence)
include/drm/drm_atomic_helper.h +2 -1
@@ -72,7 +72,8 @@ int drm_atomic_helper_async_check(struct drm_device *dev, struct drm_atomic_state *state);void drm_atomic_helper_async_commit (struct drm_device *dev,struct drm_atomic_state *state); - +void drm_atomic_helper_set_fence_deadline (struct drm_device *dev, +struct drm_atomic_state *state);int drm_atomic_helper_wait_for_fences (struct drm_device *dev,struct drm_atomic_state *state,bool pre_swap);
Signature · 1 line
2.40.1
Maarten Lankhorst
maintainerPATCH 5/9 drm/i915: Implement fence->set_deadline.
In reply to Maarten LankhorstUseful when we want to start using it in display.
--- drivers/gpu/drm/i915/i915_request.c | 9 +++++++++ 1 file changed, 9 insertions(+)
drivers/gpu/drm/i915/i915_request.c +9 -0
@@ -105,6 +105,14 @@ static signed long i915_fence_wait(struct dma_fence *fence, timeout); } +static void i915_fence_set_deadline (struct dma_fence *fence, +ktime_t deadline) +{ +struct i915_sched_attr attr = { .priority =I915_PRIORITY_DISPLAY }; + +i915_gem_fence_wait_priority (fence, &attr); +} +struct kmem_cache *i915_request_slab_cache (void ) {return slab_requests;
@@ -177,6 +185,7 @@ const struct dma_fence_ops i915_fence_ops = { .enable_signaling = i915_fence_enable_signaling, .signaled = i915_fence_signaled, .wait = i915_fence_wait, + .set_deadline = i915_fence_set_deadline, .release = i915_fence_release, };
Signature · 1 line
2.40.1
Maarten Lankhorst
maintainerPATCH 6/9 FIXME drm/i915/display: Use fence->set_deadline to boost requests
In reply to Maarten LankhorstInstead of using a i915 specific method, use the generic helper to combine fences and set the deadline.
--- .../gpu/drm/i915/display/intel_atomic_plane.c | 47 ++++--------------- drivers/gpu/drm/i915/display/intel_display.c | 2 + 2 files changed, 11 insertions(+), 38 deletions(-)
drivers/gpu/drm/i915/display/intel_atomic_plane.c +9 -38
@@ -1033,7 +1033,6 @@ static int intel_prepare_plane_fb(struct drm_plane *_plane, struct drm_plane_state *_new_plane_state) { - struct i915_sched_attr attr = { .priority = I915_PRIORITY_DISPLAY }; struct intel_plane *plane = to_intel_plane(_plane);struct intel_plane_state *new_plane_state =to_intel_plane_state (_new_plane_state);
@@ -1072,49 +1071,20 @@ intel_prepare_plane_fb(struct drm_plane *_plane, } } -if (new_plane_state->uapi.fence) {/* explicit fencing */ -i915_gem_fence_wait_priority (new_plane_state->uapi.fence, - &attr); - ret =i915_sw_fence_await_dma_fence (&state->commit_ready, - new_plane_state->uapi.fence, -i915_fence_timeout (dev_priv), -GFP_KERNEL ); -if (ret <0 ) +if (obj) { + ret =intel_plane_pin_fb (new_plane_state); +if (ret)return ret; } -if (!obj) -return 0 ; - - - ret =intel_plane_pin_fb (new_plane_state); +/* prepare after pinning */ + ret =drm_gem_plane_helper_prepare_fb (_plane, _new_plane_state);if (ret) -return ret; + goto unpin_fb; -i915_gem_object_wait_priority (obj,0 , &attr); - -if (!new_plane_state->uapi.fence) {/* implicit fencing */ -struct dma_resv_iter cursor; -struct dma_fence *fence; - - ret =i915_sw_fence_await_reservation (&state->commit_ready, - obj->base.resv, false, -i915_fence_timeout (dev_priv), -GFP_KERNEL ); -if (ret <0 ) - goto unpin_fb; - -dma_resv_iter_begin (&cursor, obj->base.resv, -DMA_RESV_USAGE_WRITE ); -dma_resv_for_each_fence_unlocked (&cursor, fence) { -intel_display_rps_boost_after_vblank (new_plane_state->hw.crtc, - fence); - } -dma_resv_iter_end (&cursor); - }else { +if (new_plane_state->uapi.fence)intel_display_rps_boost_after_vblank (new_plane_state->hw.crtc, new_plane_state->uapi.fence); - }/* * We declare pageflips to be interactive and so merit a small bias
@@ -1124,7 +1094,8 @@ intel_prepare_plane_fb(struct drm_plane *_plane,* that are not quite steady state without resorting to forcing * maximum clocks following a vblank miss (see do_rps_boost()). */ -intel_display_rps_mark_interactive (dev_priv, state, true); +if (obj) +intel_display_rps_mark_interactive (dev_priv, state, true);return 0 ;
drivers/gpu/drm/i915/display/intel_display.c +2 -0
@@ -6597,6 +6597,8 @@ static int intel_atomic_prepare_commit(struct intel_atomic_state *state)if (ret <0 )return ret; +drm_atomic_helper_set_fence_deadline (state->base.dev, &state->base); +for_each_new_intel_crtc_in_state (state, crtc, crtc_state, i) {if (intel_crtc_needs_color_update (crtc_state))intel_color_prepare_commit (crtc_state);
Signature · 1 line
2.40.1
Maarten Lankhorst
maintainerPATCH 7/9 drm/i915/display: Use i915_gem_object_get_dma_address to get dma address
In reply to Maarten LankhorstWorks better for xe like that. obj is no longer const.
--- drivers/gpu/drm/i915/display/intel_cursor.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/display/intel_cursor.c +4 -2
@@ -23,6 +23,8 @@#include "intel_psr.h" #include "skl_watermark.h" +#include "gem/i915_gem_object.h" +/* Cursor formats */ static const u32 intel_cursor_formats[] = {DRM_FORMAT_ARGB8888 ,
@@ -33,11 +35,11 @@ static u32 intel_cursor_base(const struct intel_plane_state *plane_state)struct drm_i915_private *dev_priv =to_i915 (plane_state->uapi.plane->dev);const struct drm_framebuffer *fb = plane_state->hw.fb; -const struct drm_i915_gem_object *obj =intel_fb_obj (fb); +struct drm_i915_gem_object *obj =intel_fb_obj (fb);u32 base;if (DISPLAY_INFO (dev_priv)->cursor_needs_physical) - base =sg_dma_address (obj->mm.pages->sgl); + base =i915_gem_object_get_dma_address (obj,0 );else base =intel_plane_ggtt_offset (plane_state);
Signature · 1 line
2.40.1
Maarten Lankhorst
maintainerPATCH 8/9 FIXME drm/i915/display: Minimal changes to fbdev to make xe work
In reply to Maarten LankhorstBO creation, allocation and placement is slightly different from i915 vs xe.
This is one of the few times we cannot hide between macros, so make the most of it. :)
--- drivers/gpu/drm/i915/display/intel_fbdev.c | 79 +++++++++++++++++++--- 1 file changed, 70 insertions(+), 9 deletions(-)
drivers/gpu/drm/i915/display/intel_fbdev.c +70 -9
@@ -45,6 +45,7 @@#include "gem/i915_gem_lmem.h" #include "gem/i915_gem_mman.h" +#include "gem/i915_gem_object.h" #include "i915_drv.h" #include "intel_display_types.h"
@@ -173,6 +174,7 @@ static int intelfb_alloc(struct drm_fb_helper *helper, size =PAGE_ALIGN (size); obj =ERR_PTR (-ENODEV ); +#ifdef I915 if (HAS_LMEM (dev_priv)) { obj = i915_gem_object_create_lmem(dev_priv, size, I915_BO_ALLOC_CONTIGUOUS |
@@ -185,11 +187,35 @@ static int intelfb_alloc(struct drm_fb_helper *helper,* * Also skip stolen on MTL as Wa_22018444074 mitigation. */ -if (!(IS_METEORLAKE (dev_priv )) && size *2 < dev_priv->dsm.usable_size) +if (!(IS_METEORLAKE (dev_priv )) && + size *2 <i915_gem_stolen_area_size (dev_priv)) obj =i915_gem_object_create_stolen (dev_priv, size);if (IS_ERR (obj)) obj =i915_gem_object_create_shmem (dev_priv, size); } +#else +/* +* The semantics for xe_bo_create_pin_map are different from those +* of i915_gem_object_create_*(), and need to be destroyed with +* xe_bo_unpin_map_no_vm(). +*/ +if (!IS_DGFX (dev_priv) && !IS_METEORLAKE (dev_priv) && + size *2 <i915_gem_stolen_area_size (dev_priv)) { + obj =xe_bo_create_pin_map (dev_priv,xe_device_get_root_tile (dev_priv), +NULL , size, ttm_bo_type_kernel, +XE_BO_CREATE_STOLEN_BIT | +XE_BO_CREATE_PINNED_BIT |XE_BO_SCANOUT_BIT ); +if (!IS_ERR (obj)) +drm_info (&dev_priv->drm,"Allocated fbdev into stolen\n" ); +else +drm_info (&dev_priv->drm,"Allocated fbdev into stolen failed: %li\n" ,PTR_ERR (obj)); + } +if (IS_ERR (obj)) + obj =xe_bo_create_pin_map (dev_priv,xe_device_get_root_tile (dev_priv), +NULL , size, ttm_bo_type_kernel, +XE_BO_CREATE_VRAM_IF_DGFX (xe_device_get_root_tile (dev_priv)) | +XE_BO_CREATE_PINNED_BIT |XE_BO_SCANOUT_BIT ); +#endif if (IS_ERR (obj)) {drm_err (&dev_priv->drm,"failed to allocate framebuffer (%pe)\n" , obj);
@@ -197,10 +223,18 @@ static int intelfb_alloc(struct drm_fb_helper *helper, } fb =intel_framebuffer_create (obj, &mode_cmd); -i915_gem_object_put (obj); -if (IS_ERR (fb)) +if (IS_ERR (fb)) { +#ifdef I915 +i915_gem_object_put (obj); +#else +/* undo pin/map, or we get a WARN */ +xe_bo_unpin_map_no_vm (obj); +#endif return PTR_ERR (fb); + } +/* reference transferred to fb */ +drm_gem_object_put (intel_bo_to_drm_bo (obj)); ifbdev->fb =to_intel_framebuffer (fb);return 0 ; }
@@ -213,7 +247,6 @@ static int intelfb_create(struct drm_fb_helper *helper,struct drm_device *dev = helper->dev;struct drm_i915_private *dev_priv =to_i915 (dev);struct pci_dev *pdev =to_pci_dev (dev_priv->drm.dev); -struct i915_ggtt *ggtt =to_gt (dev_priv)->ggtt;const struct i915_gtt_view view = { .type =I915_GTT_VIEW_NORMAL , };
@@ -224,8 +257,10 @@ static int intelfb_create(struct drm_fb_helper *helper,bool prealloc = false;void __iomem *vaddr;struct drm_i915_gem_object *obj; -struct i915_gem_ww_ctx ww;int ret; +#ifdef I915 +struct i915_gem_ww_ctx ww; +#endif mutex_lock (&ifbdev->hpd_lock); ret = ifbdev->hpd_suspended ? -EAGAIN :0 ;
@@ -283,6 +318,7 @@ static int intelfb_create(struct drm_fb_helper *helper, info->fbops = &intelfb_ops; obj =intel_fb_obj (&intel_fb->base); +#ifdef I915 if (i915_gem_object_is_lmem (obj)) {struct intel_memory_region *mem = obj->mm.region;
@@ -292,6 +328,8 @@ static int intelfb_create(struct drm_fb_helper *helper, i915_gem_object_get_dma_address(obj, 0)); info->fix.smem_len = obj->base.size; }else { +struct i915_ggtt *ggtt =to_gt (dev_priv)->ggtt; +/* Our framebuffer is the entirety of fbdev's system memory */ info->fix.smem_start = (unsigned long )(ggtt->gmadr.start +i915_ggtt_offset (vma));
@@ -312,12 +350,35 @@ static int intelfb_create(struct drm_fb_helper *helper, continue; } } +#else +if (obj->flags &XE_BO_CREATE_STOLEN_BIT ) { + info->fix.smem_start =xe_ttm_stolen_io_offset (obj,0 ); + }else { +struct pci_dev *pdev =to_pci_dev (dev_priv->drm.dev); + + info->fix.smem_start =pci_resource_start (pdev,2 ); +if (obj->flags &XE_BO_CREATE_VRAM0_BIT ) + info->fix.smem_start +=xe_bo_addr (obj,0 ,XE_PAGE_SIZE ); +else + info->fix.smem_start +=xe_bo_ggtt_addr (obj); + } + info->fix.smem_len = obj->ttm.base.size; + +XE_WARN_ON (iosys_map_is_null (&obj->vmap)); + vaddr = obj->vmap.vaddr_iomem; +#endif +if (IS_ERR (vaddr)) { +drm_err (&dev_priv->drm, +"Failed to remap framebuffer into virtual memory (%pe)\n" , vaddr); + ret =PTR_ERR (vaddr); + goto out_unpin; + }if (ret) goto out_unpin; info->screen_base = vaddr; - info->screen_size = vma->size; + info->screen_size =intel_bo_to_drm_bo (obj)->size;drm_fb_helper_fill_info (info, &ifbdev->helper, sizes);
@@ -325,7 +386,7 @@ static int intelfb_create(struct drm_fb_helper *helper,* If the object is stolen however, it will be full of whatever * garbage was left in there. */ -if (!i915_gem_object_is_shmem (vma->obj) && !prealloc) +if (!i915_gem_object_is_shmem (obj) && !prealloc)memset_io (info->screen_base,0 , info->screen_size);/* Use default scratch pixmap (info->pixmap.flags = FB_PIXMAP_SYSTEM) */
@@ -424,12 +485,12 @@ static bool intel_fbdev_init_bios(struct drm_device *dev,continue ; } -if (obj->base.size > max_size) { +if (intel_bo_to_drm_bo (obj)->size > max_size) {drm_dbg_kms (&i915->drm,"found possible fb from [PLANE:%d:%s]\n" , plane->base.base.id, plane->base.name); fb =to_intel_framebuffer (plane_state->uapi.fb); - max_size = obj->base.size; + max_size =intel_bo_to_drm_bo (obj)->size; } }
Signature · 1 line
2.40.1
Maarten Lankhorst
maintainerPATCH 9/9 FIXME drm/i915/display: Make intel_fb.c code compatible with xe
In reply to Maarten LankhorstFrom: Maarten Lankhorst <[email protected]>
Again there are a few changes between Xe and i915, but in general code is similar for both
--- drivers/gpu/drm/i915/display/intel_fb.c | 85 +++++++++++++++++-------- 1 file changed, 57 insertions(+), 28 deletions(-)
drivers/gpu/drm/i915/display/intel_fb.c +57 -28
@@ -17,6 +17,8 @@#include "intel_fb.h" #include "intel_frontbuffer.h" +#include "gem/i915_gem_object.h" +#define check_array_bounds (i915, a, i) drm_WARN_ON(&(i915)->drm, (i) >= ARRAY_SIZE(a))/*
@@ -737,7 +739,7 @@ intel_fb_align_height(const struct drm_framebuffer *fb,return ALIGN (height, tile_height); } -static unsigned int intel_fb_modifier_to_tiling (u64 fb_modifier) +__maybe_unused static unsigned int intel_fb_modifier_to_tiling (u64 fb_modifier) {u8 tiling_caps =lookup_modifier (fb_modifier)->plane_caps &INTEL_PLANE_CAP_TILING_MASK ;
@@ -1657,10 +1659,10 @@ int intel_fill_fb_info(struct drm_i915_private *i915, struct intel_framebuffer * max_size =max (max_size, offset + size); } -if (mul_u32_u32 (max_size, tile_size) > obj->base.size) { +if (mul_u32_u32 (max_size, tile_size) >intel_bo_to_drm_bo (obj)->size) {drm_dbg_kms (&i915->drm,"fb too big for bo (need %llu bytes, have %zu bytes)\n" , -mul_u32_u32 (max_size, tile_size), obj->base.size); +mul_u32_u32 (max_size, tile_size),intel_bo_to_drm_bo (obj)->size);return -EINVAL ; }
@@ -1907,7 +1909,7 @@ static int intel_user_framebuffer_create_handle(struct drm_framebuffer *fb, unsigned int *handle) {struct drm_i915_gem_object *obj =intel_fb_obj (fb); -struct drm_i915_private *i915 =to_i915 (obj->base.dev); +struct drm_i915_private *i915 =to_i915 (fb->dev);if (i915_gem_object_is_userptr (obj)) { drm_dbg(&i915->drm,
@@ -1915,7 +1917,7 @@ static int intel_user_framebuffer_create_handle(struct drm_framebuffer *fb,return -EINVAL ; } -return drm_gem_handle_create (file, &obj->base, handle); +return drm_gem_handle_create (file,intel_bo_to_drm_bo (obj), handle); }struct frontbuffer_fence_cb {
@@ -1948,10 +1950,10 @@ static int intel_user_framebuffer_dirty(struct drm_framebuffer *fb,if (!atomic_read (&front->bits))return 0 ; -if (dma_resv_test_signaled (obj->base.resv,dma_resv_usage_rw (false))) +if (dma_resv_test_signaled (intel_bo_to_drm_bo (obj)->resv,dma_resv_usage_rw (false))) goto flush; - ret =dma_resv_get_singleton (obj->base.resv,dma_resv_usage_rw (false), + ret =dma_resv_get_singleton (intel_bo_to_drm_bo (obj)->resv,dma_resv_usage_rw (false), &fence);if (ret || !fence) goto flush;
@@ -1993,10 +1995,10 @@ int intel_framebuffer_init(struct intel_framebuffer *intel_fb,struct drm_i915_gem_object *obj, struct drm_mode_fb_cmd2 *mode_cmd) { - struct drm_i915_private *dev_priv = to_i915(obj->base.dev); + struct drm_i915_private *dev_priv = intel_bo_to_i915(obj);struct drm_framebuffer *fb = &intel_fb->base;u32 max_stride; -unsigned int tiling, stride; +unsigned int tiling =I915_TILING_NONE , stride =0 ;int ret = -EINVAL ;int i;
@@ -2004,6 +2006,7 @@ int intel_framebuffer_init(struct intel_framebuffer *intel_fb,if (!intel_fb->frontbuffer)return -ENOMEM ; +#ifdef I915 i915_gem_object_lock (obj,NULL ); tiling =i915_gem_object_get_tiling (obj); stride =i915_gem_object_get_stride (obj);
@@ -2030,15 +2033,6 @@ int intel_framebuffer_init(struct intel_framebuffer *intel_fb, } } -if (!drm_any_plane_has_format (&dev_priv->drm, - mode_cmd->pixel_format, - mode_cmd->modifier[0 ])) { -drm_dbg_kms (&dev_priv->drm, -"unsupported pixel format %p4cc / modifier 0x%llx\n" , - &mode_cmd->pixel_format, mode_cmd->modifier[0 ]); - goto err; - } -/* * gen2/3 display engine uses the fence if present, * so the tiling mode must match the fb modifier exactly.
@@ -2049,6 +2043,36 @@ int intel_framebuffer_init(struct intel_framebuffer *intel_fb, "tiling_mode must match fb modifier exactly on gen2/3\n"); goto err; } +#else + ret =ttm_bo_reserve (&obj->ttm, true, false,NULL ); +if (ret) + goto err; + ret = -EINVAL ; + +if (!(obj->flags &XE_BO_SCANOUT_BIT )) { +/* +* XE_BO_SCANOUT_BIT should ideally be set at creation, or is +* automatically set when creating FB. We cannot change caching +* mode when the object is VM_BINDed, so we can only set +* coherency with display when unbound. +*/ +if (XE_IOCTL_DBG (dev_priv, !list_empty (&obj->ttm.base.gpuva.list))) { +ttm_bo_unreserve (&obj->ttm); + goto err; + } + obj->flags |=XE_BO_SCANOUT_BIT ; + } +ttm_bo_unreserve (&obj->ttm); +#endif + +if (!drm_any_plane_has_format (&dev_priv->drm, + mode_cmd->pixel_format, + mode_cmd->modifier[0 ])) { +drm_dbg_kms (&dev_priv->drm, +"unsupported pixel format %p4cc / modifier 0x%llx\n" , + &mode_cmd->pixel_format, mode_cmd->modifier[0 ]); + goto err; + } max_stride =intel_fb_max_stride (dev_priv, mode_cmd->pixel_format, mode_cmd->modifier[0 ]);
@@ -2111,11 +2135,7 @@ int intel_framebuffer_init(struct intel_framebuffer *intel_fb, } } -#ifdef I915 - fb->obj[i] = &obj->base; -#else - fb->obj[i] = &obj->ttm.base; -#endif + fb->obj[i] =intel_bo_to_drm_bo (obj); } ret =intel_fill_fb_info (dev_priv, intel_fb);
@@ -2163,23 +2183,32 @@ intel_user_framebuffer_create(struct drm_device *dev,struct drm_framebuffer *fb;struct drm_i915_gem_object *obj;struct drm_mode_fb_cmd2 mode_cmd = *user_mode_cmd; -struct drm_i915_private *i915; +struct drm_i915_private *i915 =to_i915 (dev); +struct drm_gem_object *gem =drm_gem_object_lookup (filp, mode_cmd.handles[0 ]); - obj =i915_gem_object_lookup (filp, mode_cmd.handles[0 ]); -if (!obj) +if (!gem)return ERR_PTR (-ENOENT ); + obj =to_intel_bo (gem); +#ifdef I915 /* object is backed with LMEM for discrete */ - i915 =to_i915 (obj->base.dev);if (HAS_LMEM (i915) && !i915_gem_object_can_migrate (obj,INTEL_REGION_LMEM_0 )) {/* object is "remote", not in local memory */ i915_gem_object_put (obj);drm_dbg_kms (&i915->drm,"framebuffer must reside in local memory\n" );return ERR_PTR (-EREMOTE ); } +#else +/* Require vram placement or dma-buf import */ +if (XE_IOCTL_DBG (i915,IS_DGFX (i915) && !xe_bo_can_migrate (obj,XE_PL_VRAM0 ) && + obj->ttm.type != ttm_bo_type_sg)) { +drm_gem_object_put (gem); +return ERR_PTR (-EREMOTE ); + } +#endif fb =intel_framebuffer_create (obj, &mode_cmd); -i915_gem_object_put (obj); +drm_gem_object_put (gem);return fb; }
Signature · 1 line
2.40.1
Quoted 7 lines · Maarten Lankhorst
From: Maarten Lankhorst <[email protected]>
All reverts are made simpler, and the header updates are squashed to a
single commit.This makes reviewing a lot easier.
I took overview of patches 4, 5 and 6. It seems your approach will take care of i915_gem_fence_wait_priority. These three patches should be sent to [email protected] (at least patch 4) and [email protected]. I think we could use them in drm-xe- next while waiting for review to complete. I will soon start backporting this from drm-intel-next:
https://patchwork.freedesktop.org/series/125160/
Still waiting for CI results there before merging into drm-intel-next. I was originally planning to use i915_compat headers for i915_gem_fence_wait_priority. I think your approach is better. So if it's ok to you I could sent these three patches together with my backport (https://patchwork.freedesktop.org/series/125160/) into intel- xe mailing list. what do you think?
Generally speaking I found this patch set being hard to follow and review. I would expect
1. "fixup! FIXME: drm/i915/display: Remaining changes to make xe compile"
2. FIXME: <name for changes split out>
where:
1. removing changes from FIXME: drm/i915/display: Remaining changes to make xe compile" and 2. introducing same changes as a separate FIXME patch.
BR,
Jouni Högander
Quoted 42 lines
Maarten Lankhorst (9):
drm/i915/display: Revert all before remaining changes to make xe compile
drm/i915/display: Revert remaining changes to make xe compile drm/xe: Update headers to be more compatible with i915 drm/atomic: Adddrm_atomic_helper_set_fence_deadline()function drm/i915: Implementfence->set_deadline.
FIXME drm/i915/display: Usefence->set_deadlineto boost requests drm/i915/display: Usei915_gem_object_get_dma_addressto get dma address
FIXME drm/i915/display: Minimal changes to fbdev to make xe work FIXME drm/i915/display: Makeintel_fb.ccode compatible with xedrivers/gpu/drm/drm_atomic_helper.c | 15 ++-
.../gpu/drm/i915/display/intel_atomic_plane.c| 70 +++--------- drivers/gpu/drm/i915/display/intel_cursor.c | 12 +-- drivers/gpu/drm/i915/display/intel_display.c | 37 +------ .../drm/i915/display/intel_display_types.h | 4 - drivers/gpu/drm/i915/display/intel_fb.c | 89 ++++++---------- drivers/gpu/drm/i915/display/intel_fbc.c | 13 ++- drivers/gpu/drm/i915/display/intel_fbdev.c | 100 ++++++---------- --
.../gpu/drm/i915/display/intel_frontbuffer.c | 7 +- .../gpu/drm/i915/display/intel_frontbuffer.h | 4 - .../drm/i915/display/skl_universal_plane.c | 4 - drivers/gpu/drm/i915/i915_request.c | 9 ++ .../compat-i915-headers/gem/i915_gem_lmem.h | 2 + .../compat-i915-headers/gem/i915_gem_mman.h | 17 +++.../compat-i915-headers/gem/i915_gem_object.h| 76 +++++++++++++ .../gem/i915_gem_object_frontbuffer.h | 4 +-.../gpu/drm/xe/compat-i915-headers/i915_drv.h| 1 + .../xe/compat-i915-headers/i915_gem_stolen.h | 13 ++-.../gpu/drm/xe/compat-i915-headers/i915_vma.h| 6 ++ include/drm/drm_atomic_helper.h | 3 +-20 files changed, 234 insertions(+), 252 deletions(-)create mode 100644
drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_lmem.h
create mode 100644drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_mman.h
create mode 100644drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_object.h
Hey,
Quoted 9 lines · Jouni
Quoted 7 lines · Maarten Lankhorst
From: Maarten Lankhorst <[email protected]>
All reverts are made simpler, and the header updates are squashed to a
single commit.This makes reviewing a lot easier.
Quoted 8 lines
I took overview of patches 4, 5 and 6. It seems your approach will take care of
i915_gem_fence_wait_priority. These three patches should be sent to [email protected] (at least patch 4) and [email protected]. I think we could use them in drm-xe- next while waiting for review to complete. I will soon start backporting this from drm-intel-next:
Noticed this patch. You really only want to wait for USAGE_KERNEL if state->fence is set.
See drm_gem_plane_helper_prepare_fb.
Can you fix this up? Else I send a patch on monday.
Cheers,
~Maarten
Quoted 69 lines
Still waiting for CI results there before merging into drm-intel-next. I was originally planning to use
i915_compatheaders fori915_gem_fence_wait_priority. I think your approach is better. So if it's ok to you I could sent these three patches together with my backport (https://patchwork.freedesktop.org/series/125160/) into intel- xe mailing list. what do you think?Generally speaking I found this patch set being hard to follow and review. I would expect
1. "fixup! FIXME: drm/i915/display: Remaining changes to make xe compile"
2. FIXME: <name for changes split out>
where:
1. removing changes from FIXME: drm/i915/display: Remaining changes to make xe compile" and 2. introducing same changes as a separate FIXME patch.
BR,
Jouni Högander
Quoted 42 lines
Maarten Lankhorst (9):
drm/i915/display: Revert all before remaining changes to make xe compile
drm/i915/display: Revert remaining changes to make xe compile drm/xe: Update headers to be more compatible with i915 drm/atomic: Adddrm_atomic_helper_set_fence_deadline()function drm/i915: Implementfence->set_deadline.
FIXME drm/i915/display: Usefence->set_deadlineto boost requests drm/i915/display: Usei915_gem_object_get_dma_addressto get dma address
FIXME drm/i915/display: Minimal changes to fbdev to make xe work FIXME drm/i915/display: Makeintel_fb.ccode compatible with xedrivers/gpu/drm/drm_atomic_helper.c | 15 ++- .../gpu/drm/i915/display/intel_atomic_plane.c | 70 +++--------- drivers/gpu/drm/i915/display/intel_cursor.c | 12 +-- drivers/gpu/drm/i915/display/intel_display.c | 37 +------ .../drm/i915/display/intel_display_types.h | 4 - drivers/gpu/drm/i915/display/intel_fb.c | 89 ++++++---------- drivers/gpu/drm/i915/display/intel_fbc.c | 13 ++- drivers/gpu/drm/i915/display/intel_fbdev.c | 100 ++++++---------- -- .../gpu/drm/i915/display/intel_frontbuffer.c | 7 +- .../gpu/drm/i915/display/intel_frontbuffer.h | 4 - .../drm/i915/display/skl_universal_plane.c | 4 - drivers/gpu/drm/i915/i915_request.c | 9 ++ .../compat-i915-headers/gem/i915_gem_lmem.h | 2 + .../compat-i915-headers/gem/i915_gem_mman.h | 17 +++ .../compat-i915-headers/gem/i915_gem_object.h | 76 +++++++++++++ .../gem/i915_gem_object_frontbuffer.h | 4 +- .../gpu/drm/xe/compat-i915-headers/i915_drv.h | 1 + .../xe/compat-i915-headers/i915_gem_stolen.h | 13 ++- .../gpu/drm/xe/compat-i915-headers/i915_vma.h | 6 ++ include/drm/drm_atomic_helper.h | 3 +- 20 files changed, 234 insertions(+), 252 deletions(-)create mode 100644
drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_lmem.h
create mode 100644drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_mman.h
create mode 100644drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_object.h
Quoted 22 lines · Maarten Lankhorst
Hey,
Quoted 19 lines · Jouni
Quoted 8 lines · Maarten Lankhorst
From: Maarten Lankhorst <[email protected]>
All reverts are made simpler, and the header updates are squashed to
a
single commit.This makes reviewing a lot easier.
I took overview of patches 4, 5 and 6. It seems your approach will take
care ofi915_gem_fence_wait_priority. These three patches should be sent to [email protected] (at least patch 4) and [email protected]. I think we could use them in drm- xe-
next while waiting for review to complete. I will soon start backporting this from drm-intel-next:
Quoted 7 lines
Quoted 1 line
Noticed this patch. You really only want to wait for
USAGE_KERNELifstate->fenceis set.See
drm_gem_plane_helper_prepare_fb.Can you fix this up? Else I send a patch on monday.
I already merged this. There was review comment from Ville asking vice versa:
https://patchwork.freedesktop.org/patch/564991/?series=125160&rev=3
BR,
Jouni Högander
Quoted 88 lines
Cheers,
~MaartenQuoted 84 lines
Still waiting for CI results there before merging into drm-intel- next.
I was originally planning to usei915_compatheaders fori915_gem_fence_wait_priority. I think your approach is better. So if
it's ok to you I could sent these three patches together with my backport (https://patchwork.freedesktop.org/series/125160/) into intel-
xe mailing list. what do you think?Generally speaking I found this patch set being hard to follow and review. I would expect
1. "fixup! FIXME: drm/i915/display: Remaining changes to make xe compile"
2. FIXME: <name for changes split out>
where:
1. removing changes from FIXME: drm/i915/display: Remaining changes to
make xe compile" and 2. introducing same changes as a separate FIXME
patch.BR,
Jouni Högander
Quoted 52 lines
Maarten Lankhorst (9):
drm/i915/display: Revert all before remaining changes to make xe
compile
drm/i915/display: Revert remaining changes to make xe compile drm/xe: Update headers to be more compatible with i915 drm/atomic: Adddrm_atomic_helper_set_fence_deadline()function
drm/i915: Implementfence->set_deadline.
FIXME drm/i915/display: Usefence->set_deadlineto boost requests
drm/i915/display: Usei915_gem_object_get_dma_addressto get dma
address
FIXME drm/i915/display: Minimal changes to fbdev to make xe work
FIXME drm/i915/display: Makeintel_fb.ccode compatible with xedrivers/gpu/drm/drm_atomic_helper.c | 15 ++-
.../gpu/drm/i915/display/intel_atomic_plane.c| 70 +++-------- -
drivers/gpu/drm/i915/display/intel_cursor.c | 12 +-- drivers/gpu/drm/i915/display/intel_display.c | 37 +------ .../drm/i915/display/intel_display_types.h | 4 - drivers/gpu/drm/i915/display/intel_fb.c | 89 ++++++-----
-----
drivers/gpu/drm/i915/display/intel_fbc.c | 13 ++- drivers/gpu/drm/i915/display/intel_fbdev.c | 100 ++++++-----
-----
--
.../gpu/drm/i915/display/intel_frontbuffer.c | 7 +- .../gpu/drm/i915/display/intel_frontbuffer.h | 4 - .../drm/i915/display/skl_universal_plane.c | 4 - drivers/gpu/drm/i915/i915_request.c | 9 ++ .../compat-i915-headers/gem/i915_gem_lmem.h | 2 + .../compat-i915-headers/gem/i915_gem_mman.h | 17 +++.../compat-i915-headers/gem/i915_gem_object.h| 76
+++++++++++++
.../gem/i915_gem_object_frontbuffer.h | 4 +-.../gpu/drm/xe/compat-i915-headers/i915_drv.h| 1 + .../xe/compat-i915-headers/i915_gem_stolen.h | 13 ++-.../gpu/drm/xe/compat-i915-headers/i915_vma.h| 6 ++ include/drm/drm_atomic_helper.h | 3 +-20 files changed, 234 insertions(+), 252 deletions(-)create mode 100644
drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_lmem.h
create mode 100644drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_mman.h
create mode 100644drivers/gpu/drm/xe/compat-i915-headers/gem/i915_gem_object.h
Patchwork
CI[Intel-xe] ✗ CI.Patch_applied: failure for Cleaned up revert of remaining changes.
In reply to Maarten Lankhorst== Series Details ==
== Series Details ==
Series: Cleaned up revert of remaining changes.
URL : https://patchwork.freedesktop.org/series/125685/
State : failure
== Summary ==
=== Applying kernel patches on branch 'drm-xe-next' with base: ===
Base commit: 1832821c7 drm/xe: Print virtualization mode during probe
=== git am output follows ===
error: patch failed: drivers/gpu/drm/i915/display/intel_fbdev.c:48
error: drivers/gpu/drm/i915/display/intel_fbdev.c: patch does not apply
hint: Use 'git am --show-current-patch' to see the failed patch
Applying: drm/i915/display: Revert all before remaining changes to make xe compile Patch failed at 0001 drm/i915/display: Revert all before remaining changes to make xe compile When you have resolved this problem, run "git am --continue".
If you prefer to skip this patch, run "git am --skip" instead.
To restore the original branch and stop patching, run "git am --abort".