From d635711daa98be86d4c7fd01499c34f566b54ccb Mon Sep 17 00:00:00 2001 From: André Fabian Silva Delgado Date: Fri, 10 Jun 2016 05:30:17 -0300 Subject: Linux-libre 4.6.2-gnu --- drivers/gpu/drm/vc4/vc4_bo.c | 5 +- drivers/gpu/drm/vc4/vc4_crtc.c | 210 ++++++++------ drivers/gpu/drm/vc4/vc4_drv.c | 10 - drivers/gpu/drm/vc4/vc4_drv.h | 50 +++- drivers/gpu/drm/vc4/vc4_gem.c | 123 +++++--- drivers/gpu/drm/vc4/vc4_hdmi.c | 42 ++- drivers/gpu/drm/vc4/vc4_hvs.c | 97 +++++++ drivers/gpu/drm/vc4/vc4_irq.c | 58 +++- drivers/gpu/drm/vc4/vc4_kms.c | 9 + drivers/gpu/drm/vc4/vc4_plane.c | 603 ++++++++++++++++++++++++++++++++++++---- drivers/gpu/drm/vc4/vc4_regs.h | 120 +++++++- drivers/gpu/drm/vc4/vc4_v3d.c | 1 + 12 files changed, 1115 insertions(+), 213 deletions(-) (limited to 'drivers/gpu/drm/vc4') diff --git a/drivers/gpu/drm/vc4/vc4_bo.c b/drivers/gpu/drm/vc4/vc4_bo.c index ac8eafea6..9807bc9d2 100644 --- a/drivers/gpu/drm/vc4/vc4_bo.c +++ b/drivers/gpu/drm/vc4/vc4_bo.c @@ -398,9 +398,8 @@ int vc4_mmap(struct file *filp, struct vm_area_struct *vma) vma->vm_flags &= ~VM_PFNMAP; vma->vm_pgoff = 0; - ret = dma_mmap_writecombine(bo->base.base.dev->dev, vma, - bo->base.vaddr, bo->base.paddr, - vma->vm_end - vma->vm_start); + ret = dma_mmap_wc(bo->base.base.dev->dev, vma, bo->base.vaddr, + bo->base.paddr, vma->vm_end - vma->vm_start); if (ret) drm_gem_vm_close(vma); diff --git a/drivers/gpu/drm/vc4/vc4_crtc.c b/drivers/gpu/drm/vc4/vc4_crtc.c index 018145e0b..355ee4b09 100644 --- a/drivers/gpu/drm/vc4/vc4_crtc.c +++ b/drivers/gpu/drm/vc4/vc4_crtc.c @@ -49,22 +49,27 @@ struct vc4_crtc { /* Which HVS channel we're using for our CRTC. */ int channel; - /* Pointer to the actual hardware display list memory for the - * crtc. - */ - u32 __iomem *dlist; - - u32 dlist_size; /* in dwords */ - struct drm_pending_vblank_event *event; }; +struct vc4_crtc_state { + struct drm_crtc_state base; + /* Dlist area for this CRTC configuration. */ + struct drm_mm_node mm; +}; + static inline struct vc4_crtc * to_vc4_crtc(struct drm_crtc *crtc) { return (struct vc4_crtc *)crtc; } +static inline struct vc4_crtc_state * +to_vc4_crtc_state(struct drm_crtc_state *crtc_state) +{ + return (struct vc4_crtc_state *)crtc_state; +} + struct vc4_crtc_data { /* Which channel of the HVS this pixelvalve sources from. */ int hvs_channel; @@ -83,7 +88,7 @@ static const struct { } crtc_regs[] = { CRTC_REG(PV_CONTROL), CRTC_REG(PV_V_CONTROL), - CRTC_REG(PV_VSYNCD), + CRTC_REG(PV_VSYNCD_EVEN), CRTC_REG(PV_HORZA), CRTC_REG(PV_HORZB), CRTC_REG(PV_VERTA), @@ -183,6 +188,8 @@ static int vc4_get_clock_select(struct drm_crtc *crtc) static void vc4_crtc_mode_set_nofb(struct drm_crtc *crtc) { + struct drm_device *dev = crtc->dev; + struct vc4_dev *vc4 = to_vc4_dev(dev); struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc); struct drm_crtc_state *state = crtc->state; struct drm_display_mode *mode = &state->adjusted_mode; @@ -212,6 +219,16 @@ static void vc4_crtc_mode_set_nofb(struct drm_crtc *crtc) PV_HORZB_HFP) | VC4_SET_FIELD(mode->hdisplay, PV_HORZB_HACTIVE)); + CRTC_WRITE(PV_VERTA, + VC4_SET_FIELD(mode->vtotal - mode->vsync_end, + PV_VERTA_VBP) | + VC4_SET_FIELD(mode->vsync_end - mode->vsync_start, + PV_VERTA_VSYNC)); + CRTC_WRITE(PV_VERTB, + VC4_SET_FIELD(mode->vsync_start - mode->vdisplay, + PV_VERTB_VFP) | + VC4_SET_FIELD(vactive, PV_VERTB_VACTIVE)); + if (interlace) { CRTC_WRITE(PV_VERTA_EVEN, VC4_SET_FIELD(mode->vtotal - mode->vsync_end - 1, @@ -241,6 +258,10 @@ static void vc4_crtc_mode_set_nofb(struct drm_crtc *crtc) PV_CONTROL_FIFO_CLR | PV_CONTROL_EN); + HVS_WRITE(SCALER_DISPBKGNDX(vc4_crtc->channel), + SCALER_DISPBKGND_AUTOHS | + (interlace ? SCALER_DISPBKGND_INTERLACE : 0)); + if (debug_dump_regs) { DRM_INFO("CRTC %d regs after:\n", drm_crtc_index(crtc)); vc4_crtc_dump_regs(vc4_crtc); @@ -319,11 +340,13 @@ static void vc4_crtc_enable(struct drm_crtc *crtc) static int vc4_crtc_atomic_check(struct drm_crtc *crtc, struct drm_crtc_state *state) { + struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(state); struct drm_device *dev = crtc->dev; struct vc4_dev *vc4 = to_vc4_dev(dev); struct drm_plane *plane; - struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc); + unsigned long flags; u32 dlist_count = 0; + int ret; /* The pixelvalve can only feed one encoder (and encoders are * 1:1 with connectors.) @@ -346,18 +369,12 @@ static int vc4_crtc_atomic_check(struct drm_crtc *crtc, dlist_count++; /* Account for SCALER_CTL0_END. */ - if (!vc4_crtc->dlist || dlist_count > vc4_crtc->dlist_size) { - vc4_crtc->dlist = ((u32 __iomem *)vc4->hvs->dlist + - HVS_BOOTLOADER_DLIST_END); - vc4_crtc->dlist_size = ((SCALER_DLIST_SIZE >> 2) - - HVS_BOOTLOADER_DLIST_END); - - if (dlist_count > vc4_crtc->dlist_size) { - DRM_DEBUG_KMS("dlist too large for CRTC (%d > %d).\n", - dlist_count, vc4_crtc->dlist_size); - return -EINVAL; - } - } + spin_lock_irqsave(&vc4->hvs->mm_lock, flags); + ret = drm_mm_insert_node(&vc4->hvs->dlist_mm, &vc4_state->mm, + dlist_count, 1, 0); + spin_unlock_irqrestore(&vc4->hvs->mm_lock, flags); + if (ret) + return ret; return 0; } @@ -368,47 +385,29 @@ static void vc4_crtc_atomic_flush(struct drm_crtc *crtc, struct drm_device *dev = crtc->dev; struct vc4_dev *vc4 = to_vc4_dev(dev); struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc); + struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(crtc->state); struct drm_plane *plane; bool debug_dump_regs = false; - u32 __iomem *dlist_next = vc4_crtc->dlist; + u32 __iomem *dlist_start = vc4->hvs->dlist + vc4_state->mm.start; + u32 __iomem *dlist_next = dlist_start; if (debug_dump_regs) { DRM_INFO("CRTC %d HVS before:\n", drm_crtc_index(crtc)); vc4_hvs_dump_state(dev); } - /* Copy all the active planes' dlist contents to the hardware dlist. - * - * XXX: If the new display list was large enough that it - * overlapped a currently-read display list, we need to do - * something like disable scanout before putting in the new - * list. For now, we're safe because we only have the two - * planes. - */ + /* Copy all the active planes' dlist contents to the hardware dlist. */ drm_atomic_crtc_for_each_plane(plane, crtc) { dlist_next += vc4_plane_write_dlist(plane, dlist_next); } - if (dlist_next == vc4_crtc->dlist) { - /* If no planes were enabled, use the SCALER_CTL0_END - * at the start of the display list memory (in the - * bootloader section). We'll rewrite that - * SCALER_CTL0_END, just in case, though. - */ - writel(SCALER_CTL0_END, vc4->hvs->dlist); - HVS_WRITE(SCALER_DISPLISTX(vc4_crtc->channel), 0); - } else { - writel(SCALER_CTL0_END, dlist_next); - dlist_next++; - - HVS_WRITE(SCALER_DISPLISTX(vc4_crtc->channel), - (u32 __iomem *)vc4_crtc->dlist - - (u32 __iomem *)vc4->hvs->dlist); - - /* Make the next display list start after ours. */ - vc4_crtc->dlist_size -= (dlist_next - vc4_crtc->dlist); - vc4_crtc->dlist = dlist_next; - } + writel(SCALER_CTL0_END, dlist_next); + dlist_next++; + + WARN_ON_ONCE(dlist_next - dlist_start != vc4_state->mm.size); + + HVS_WRITE(SCALER_DISPLISTX(vc4_crtc->channel), + vc4_state->mm.start); if (debug_dump_regs) { DRM_INFO("CRTC %d HVS after:\n", drm_crtc_index(crtc)); @@ -544,6 +543,7 @@ static int vc4_async_page_flip(struct drm_crtc *crtc, /* Make sure all other async modesetes have landed. */ ret = down_interruptible(&vc4->async_modeset); if (ret) { + drm_framebuffer_unreference(fb); kfree(flip_state); return ret; } @@ -573,6 +573,36 @@ static int vc4_page_flip(struct drm_crtc *crtc, return drm_atomic_helper_page_flip(crtc, fb, event, flags); } +static struct drm_crtc_state *vc4_crtc_duplicate_state(struct drm_crtc *crtc) +{ + struct vc4_crtc_state *vc4_state; + + vc4_state = kzalloc(sizeof(*vc4_state), GFP_KERNEL); + if (!vc4_state) + return NULL; + + __drm_atomic_helper_crtc_duplicate_state(crtc, &vc4_state->base); + return &vc4_state->base; +} + +static void vc4_crtc_destroy_state(struct drm_crtc *crtc, + struct drm_crtc_state *state) +{ + struct vc4_dev *vc4 = to_vc4_dev(crtc->dev); + struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(state); + + if (vc4_state->mm.allocated) { + unsigned long flags; + + spin_lock_irqsave(&vc4->hvs->mm_lock, flags); + drm_mm_remove_node(&vc4_state->mm); + spin_unlock_irqrestore(&vc4->hvs->mm_lock, flags); + + } + + __drm_atomic_helper_crtc_destroy_state(crtc, state); +} + static const struct drm_crtc_funcs vc4_crtc_funcs = { .set_config = drm_atomic_helper_set_config, .destroy = vc4_crtc_destroy, @@ -581,8 +611,8 @@ static const struct drm_crtc_funcs vc4_crtc_funcs = { .cursor_set = NULL, /* handled by drm_mode_cursor_universal */ .cursor_move = NULL, /* handled by drm_mode_cursor_universal */ .reset = drm_atomic_helper_crtc_reset, - .atomic_duplicate_state = drm_atomic_helper_crtc_duplicate_state, - .atomic_destroy_state = drm_atomic_helper_crtc_destroy_state, + .atomic_duplicate_state = vc4_crtc_duplicate_state, + .atomic_destroy_state = vc4_crtc_destroy_state, }; static const struct drm_crtc_helper_funcs vc4_crtc_helper_funcs = { @@ -593,26 +623,6 @@ static const struct drm_crtc_helper_funcs vc4_crtc_helper_funcs = { .atomic_flush = vc4_crtc_atomic_flush, }; -/* Frees the page flip event when the DRM device is closed with the - * event still outstanding. - */ -void vc4_cancel_page_flip(struct drm_crtc *crtc, struct drm_file *file) -{ - struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc); - struct drm_device *dev = crtc->dev; - unsigned long flags; - - spin_lock_irqsave(&dev->event_lock, flags); - - if (vc4_crtc->event && vc4_crtc->event->base.file_priv == file) { - vc4_crtc->event->base.destroy(&vc4_crtc->event->base); - drm_crtc_vblank_put(crtc); - vc4_crtc->event = NULL; - } - - spin_unlock_irqrestore(&dev->event_lock, flags); -} - static const struct vc4_crtc_data pv0_data = { .hvs_channel = 0, .encoder0_type = VC4_ENCODER_TYPE_DSI0, @@ -664,9 +674,9 @@ static int vc4_crtc_bind(struct device *dev, struct device *master, void *data) struct vc4_dev *vc4 = to_vc4_dev(drm); struct vc4_crtc *vc4_crtc; struct drm_crtc *crtc; - struct drm_plane *primary_plane, *cursor_plane; + struct drm_plane *primary_plane, *cursor_plane, *destroy_plane, *temp; const struct of_device_id *match; - int ret; + int ret, i; vc4_crtc = devm_kzalloc(dev, sizeof(*vc4_crtc), GFP_KERNEL); if (!vc4_crtc) @@ -695,27 +705,49 @@ static int vc4_crtc_bind(struct device *dev, struct device *master, void *data) goto err; } - cursor_plane = vc4_plane_init(drm, DRM_PLANE_TYPE_CURSOR); - if (IS_ERR(cursor_plane)) { - dev_err(dev, "failed to construct cursor plane\n"); - ret = PTR_ERR(cursor_plane); - goto err_primary; - } - - drm_crtc_init_with_planes(drm, crtc, primary_plane, cursor_plane, + drm_crtc_init_with_planes(drm, crtc, primary_plane, NULL, &vc4_crtc_funcs, NULL); drm_crtc_helper_add(crtc, &vc4_crtc_helper_funcs); primary_plane->crtc = crtc; - cursor_plane->crtc = crtc; vc4->crtc[drm_crtc_index(crtc)] = vc4_crtc; vc4_crtc->channel = vc4_crtc->data->hvs_channel; + /* Set up some arbitrary number of planes. We're not limited + * by a set number of physical registers, just the space in + * the HVS (16k) and how small an plane can be (28 bytes). + * However, each plane we set up takes up some memory, and + * increases the cost of looping over planes, which atomic + * modesetting does quite a bit. As a result, we pick a + * modest number of planes to expose, that should hopefully + * still cover any sane usecase. + */ + for (i = 0; i < 8; i++) { + struct drm_plane *plane = + vc4_plane_init(drm, DRM_PLANE_TYPE_OVERLAY); + + if (IS_ERR(plane)) + continue; + + plane->possible_crtcs = 1 << drm_crtc_index(crtc); + } + + /* Set up the legacy cursor after overlay initialization, + * since we overlay planes on the CRTC in the order they were + * initialized. + */ + cursor_plane = vc4_plane_init(drm, DRM_PLANE_TYPE_CURSOR); + if (!IS_ERR(cursor_plane)) { + cursor_plane->possible_crtcs = 1 << drm_crtc_index(crtc); + cursor_plane->crtc = crtc; + crtc->cursor = cursor_plane; + } + CRTC_WRITE(PV_INTEN, 0); CRTC_WRITE(PV_INTSTAT, PV_INT_VFP_START); ret = devm_request_irq(dev, platform_get_irq(pdev, 0), vc4_crtc_irq_handler, 0, "vc4 crtc", vc4_crtc); if (ret) - goto err_cursor; + goto err_destroy_planes; vc4_set_crtc_possible_masks(drm, crtc); @@ -723,10 +755,12 @@ static int vc4_crtc_bind(struct device *dev, struct device *master, void *data) return 0; -err_cursor: - cursor_plane->funcs->destroy(cursor_plane); -err_primary: - primary_plane->funcs->destroy(primary_plane); +err_destroy_planes: + list_for_each_entry_safe(destroy_plane, temp, + &drm->mode_config.plane_list, head) { + if (destroy_plane->possible_crtcs == 1 << drm_crtc_index(crtc)) + destroy_plane->funcs->destroy(destroy_plane); + } err: return ret; } diff --git a/drivers/gpu/drm/vc4/vc4_drv.c b/drivers/gpu/drm/vc4/vc4_drv.c index f1655fff8..b7d2ff0e6 100644 --- a/drivers/gpu/drm/vc4/vc4_drv.c +++ b/drivers/gpu/drm/vc4/vc4_drv.c @@ -43,14 +43,6 @@ void __iomem *vc4_ioremap_regs(struct platform_device *dev, int index) return map; } -static void vc4_drm_preclose(struct drm_device *dev, struct drm_file *file) -{ - struct drm_crtc *crtc; - - list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) - vc4_cancel_page_flip(crtc, file); -} - static void vc4_lastclose(struct drm_device *dev) { struct vc4_dev *vc4 = to_vc4_dev(dev); @@ -91,8 +83,6 @@ static struct drm_driver vc4_drm_driver = { DRIVER_HAVE_IRQ | DRIVER_PRIME), .lastclose = vc4_lastclose, - .preclose = vc4_drm_preclose, - .irq_handler = vc4_irq, .irq_preinstall = vc4_irq_preinstall, .irq_postinstall = vc4_irq_postinstall, diff --git a/drivers/gpu/drm/vc4/vc4_drv.h b/drivers/gpu/drm/vc4/vc4_drv.h index 51a63330d..fa2ad15d4 100644 --- a/drivers/gpu/drm/vc4/vc4_drv.h +++ b/drivers/gpu/drm/vc4/vc4_drv.h @@ -52,7 +52,7 @@ struct vc4_dev { /* Protects bo_cache and the BO stats. */ struct mutex bo_lock; - /* Sequence number for the last job queued in job_list. + /* Sequence number for the last job queued in bin_job_list. * Starts at 0 (no jobs emitted). */ uint64_t emit_seqno; @@ -62,11 +62,19 @@ struct vc4_dev { */ uint64_t finished_seqno; - /* List of all struct vc4_exec_info for jobs to be executed. - * The first job in the list is the one currently programmed - * into ct0ca/ct1ca for execution. + /* List of all struct vc4_exec_info for jobs to be executed in + * the binner. The first job in the list is the one currently + * programmed into ct0ca for execution. */ - struct list_head job_list; + struct list_head bin_job_list; + + /* List of all struct vc4_exec_info for jobs that have + * completed binning and are ready for rendering. The first + * job in the list is the one currently programmed into ct1ca + * for execution. + */ + struct list_head render_job_list; + /* List of the finished vc4_exec_infos waiting to be freed by * job_done_work. */ @@ -154,7 +162,17 @@ struct vc4_v3d { struct vc4_hvs { struct platform_device *pdev; void __iomem *regs; - void __iomem *dlist; + u32 __iomem *dlist; + + /* Memory manager for CRTCs to allocate space in the display + * list. Units are dwords. + */ + struct drm_mm dlist_mm; + /* Memory manager for the LBM memory used by HVS scaling. */ + struct drm_mm lbm_mm; + spinlock_t mm_lock; + + struct drm_mm_node mitchell_netravali_filter; }; struct vc4_plane { @@ -286,11 +304,20 @@ struct vc4_exec_info { }; static inline struct vc4_exec_info * -vc4_first_job(struct vc4_dev *vc4) +vc4_first_bin_job(struct vc4_dev *vc4) +{ + if (list_empty(&vc4->bin_job_list)) + return NULL; + return list_first_entry(&vc4->bin_job_list, struct vc4_exec_info, head); +} + +static inline struct vc4_exec_info * +vc4_first_render_job(struct vc4_dev *vc4) { - if (list_empty(&vc4->job_list)) + if (list_empty(&vc4->render_job_list)) return NULL; - return list_first_entry(&vc4->job_list, struct vc4_exec_info, head); + return list_first_entry(&vc4->render_job_list, + struct vc4_exec_info, head); } /** @@ -386,7 +413,6 @@ int vc4_bo_stats_debugfs(struct seq_file *m, void *arg); extern struct platform_driver vc4_crtc_driver; int vc4_enable_vblank(struct drm_device *dev, unsigned int crtc_id); void vc4_disable_vblank(struct drm_device *dev, unsigned int crtc_id); -void vc4_cancel_page_flip(struct drm_crtc *crtc, struct drm_file *file); int vc4_crtc_debugfs_regs(struct seq_file *m, void *arg); /* vc4_debugfs.c */ @@ -405,7 +431,9 @@ int vc4_wait_seqno_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv); int vc4_wait_bo_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv); -void vc4_submit_next_job(struct drm_device *dev); +void vc4_submit_next_bin_job(struct drm_device *dev); +void vc4_submit_next_render_job(struct drm_device *dev); +void vc4_move_job_to_render(struct drm_device *dev, struct vc4_exec_info *exec); int vc4_wait_for_seqno(struct drm_device *dev, uint64_t seqno, uint64_t timeout_ns, bool interruptible); void vc4_job_handle_completed(struct vc4_dev *vc4); diff --git a/drivers/gpu/drm/vc4/vc4_gem.c b/drivers/gpu/drm/vc4/vc4_gem.c index 202aa1544..8d4384f8b 100644 --- a/drivers/gpu/drm/vc4/vc4_gem.c +++ b/drivers/gpu/drm/vc4/vc4_gem.c @@ -141,10 +141,10 @@ vc4_save_hang_state(struct drm_device *dev) struct vc4_dev *vc4 = to_vc4_dev(dev); struct drm_vc4_get_hang_state *state; struct vc4_hang_state *kernel_state; - struct vc4_exec_info *exec; + struct vc4_exec_info *exec[2]; struct vc4_bo *bo; unsigned long irqflags; - unsigned int i, unref_list_count; + unsigned int i, j, unref_list_count, prev_idx; kernel_state = kcalloc(1, sizeof(*kernel_state), GFP_KERNEL); if (!kernel_state) @@ -153,37 +153,55 @@ vc4_save_hang_state(struct drm_device *dev) state = &kernel_state->user_state; spin_lock_irqsave(&vc4->job_lock, irqflags); - exec = vc4_first_job(vc4); - if (!exec) { + exec[0] = vc4_first_bin_job(vc4); + exec[1] = vc4_first_render_job(vc4); + if (!exec[0] && !exec[1]) { spin_unlock_irqrestore(&vc4->job_lock, irqflags); return; } - unref_list_count = 0; - list_for_each_entry(bo, &exec->unref_list, unref_head) - unref_list_count++; + /* Get the bos from both binner and renderer into hang state. */ + state->bo_count = 0; + for (i = 0; i < 2; i++) { + if (!exec[i]) + continue; + + unref_list_count = 0; + list_for_each_entry(bo, &exec[i]->unref_list, unref_head) + unref_list_count++; + state->bo_count += exec[i]->bo_count + unref_list_count; + } + + kernel_state->bo = kcalloc(state->bo_count, + sizeof(*kernel_state->bo), GFP_ATOMIC); - state->bo_count = exec->bo_count + unref_list_count; - kernel_state->bo = kcalloc(state->bo_count, sizeof(*kernel_state->bo), - GFP_ATOMIC); if (!kernel_state->bo) { spin_unlock_irqrestore(&vc4->job_lock, irqflags); return; } - for (i = 0; i < exec->bo_count; i++) { - drm_gem_object_reference(&exec->bo[i]->base); - kernel_state->bo[i] = &exec->bo[i]->base; - } + prev_idx = 0; + for (i = 0; i < 2; i++) { + if (!exec[i]) + continue; - list_for_each_entry(bo, &exec->unref_list, unref_head) { - drm_gem_object_reference(&bo->base.base); - kernel_state->bo[i] = &bo->base.base; - i++; + for (j = 0; j < exec[i]->bo_count; j++) { + drm_gem_object_reference(&exec[i]->bo[j]->base); + kernel_state->bo[j + prev_idx] = &exec[i]->bo[j]->base; + } + + list_for_each_entry(bo, &exec[i]->unref_list, unref_head) { + drm_gem_object_reference(&bo->base.base); + kernel_state->bo[j + prev_idx] = &bo->base.base; + j++; + } + prev_idx = j + 1; } - state->start_bin = exec->ct0ca; - state->start_render = exec->ct1ca; + if (exec[0]) + state->start_bin = exec[0]->ct0ca; + if (exec[1]) + state->start_render = exec[1]->ct1ca; spin_unlock_irqrestore(&vc4->job_lock, irqflags); @@ -267,13 +285,15 @@ vc4_hangcheck_elapsed(unsigned long data) struct vc4_dev *vc4 = to_vc4_dev(dev); uint32_t ct0ca, ct1ca; unsigned long irqflags; - struct vc4_exec_info *exec; + struct vc4_exec_info *bin_exec, *render_exec; spin_lock_irqsave(&vc4->job_lock, irqflags); - exec = vc4_first_job(vc4); + + bin_exec = vc4_first_bin_job(vc4); + render_exec = vc4_first_render_job(vc4); /* If idle, we can stop watching for hangs. */ - if (!exec) { + if (!bin_exec && !render_exec) { spin_unlock_irqrestore(&vc4->job_lock, irqflags); return; } @@ -284,9 +304,12 @@ vc4_hangcheck_elapsed(unsigned long data) /* If we've made any progress in execution, rearm the timer * and wait. */ - if (ct0ca != exec->last_ct0ca || ct1ca != exec->last_ct1ca) { - exec->last_ct0ca = ct0ca; - exec->last_ct1ca = ct1ca; + if ((bin_exec && ct0ca != bin_exec->last_ct0ca) || + (render_exec && ct1ca != render_exec->last_ct1ca)) { + if (bin_exec) + bin_exec->last_ct0ca = ct0ca; + if (render_exec) + render_exec->last_ct1ca = ct1ca; spin_unlock_irqrestore(&vc4->job_lock, irqflags); vc4_queue_hangcheck(dev); return; @@ -386,11 +409,13 @@ vc4_flush_caches(struct drm_device *dev) * The job_lock should be held during this. */ void -vc4_submit_next_job(struct drm_device *dev) +vc4_submit_next_bin_job(struct drm_device *dev) { struct vc4_dev *vc4 = to_vc4_dev(dev); - struct vc4_exec_info *exec = vc4_first_job(vc4); + struct vc4_exec_info *exec; +again: + exec = vc4_first_bin_job(vc4); if (!exec) return; @@ -400,11 +425,40 @@ vc4_submit_next_job(struct drm_device *dev) V3D_WRITE(V3D_BPOA, 0); V3D_WRITE(V3D_BPOS, 0); - if (exec->ct0ca != exec->ct0ea) + /* Either put the job in the binner if it uses the binner, or + * immediately move it to the to-be-rendered queue. + */ + if (exec->ct0ca != exec->ct0ea) { submit_cl(dev, 0, exec->ct0ca, exec->ct0ea); + } else { + vc4_move_job_to_render(dev, exec); + goto again; + } +} + +void +vc4_submit_next_render_job(struct drm_device *dev) +{ + struct vc4_dev *vc4 = to_vc4_dev(dev); + struct vc4_exec_info *exec = vc4_first_render_job(vc4); + + if (!exec) + return; + submit_cl(dev, 1, exec->ct1ca, exec->ct1ea); } +void +vc4_move_job_to_render(struct drm_device *dev, struct vc4_exec_info *exec) +{ + struct vc4_dev *vc4 = to_vc4_dev(dev); + bool was_empty = list_empty(&vc4->render_job_list); + + list_move_tail(&exec->head, &vc4->render_job_list); + if (was_empty) + vc4_submit_next_render_job(dev); +} + static void vc4_update_bo_seqnos(struct vc4_exec_info *exec, uint64_t seqno) { @@ -443,14 +497,14 @@ vc4_queue_submit(struct drm_device *dev, struct vc4_exec_info *exec) exec->seqno = seqno; vc4_update_bo_seqnos(exec, seqno); - list_add_tail(&exec->head, &vc4->job_list); + list_add_tail(&exec->head, &vc4->bin_job_list); /* If no job was executing, kick ours off. Otherwise, it'll - * get started when the previous job's frame done interrupt + * get started when the previous job's flush done interrupt * occurs. */ - if (vc4_first_job(vc4) == exec) { - vc4_submit_next_job(dev); + if (vc4_first_bin_job(vc4) == exec) { + vc4_submit_next_bin_job(dev); vc4_queue_hangcheck(dev); } @@ -859,7 +913,8 @@ vc4_gem_init(struct drm_device *dev) { struct vc4_dev *vc4 = to_vc4_dev(dev); - INIT_LIST_HEAD(&vc4->job_list); + INIT_LIST_HEAD(&vc4->bin_job_list); + INIT_LIST_HEAD(&vc4->render_job_list); INIT_LIST_HEAD(&vc4->job_done_list); INIT_LIST_HEAD(&vc4->seqno_cb_list); spin_lock_init(&vc4->job_lock); diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.c b/drivers/gpu/drm/vc4/vc4_hdmi.c index c69c04601..d8b864925 100644 --- a/drivers/gpu/drm/vc4/vc4_hdmi.c +++ b/drivers/gpu/drm/vc4/vc4_hdmi.c @@ -47,6 +47,7 @@ struct vc4_hdmi { void __iomem *hdmicore_regs; void __iomem *hd_regs; int hpd_gpio; + bool hpd_active_low; struct clk *pixel_clock; struct clk *hsm_clock; @@ -95,6 +96,7 @@ static const struct { HDMI_REG(VC4_HDMI_SW_RESET_CONTROL), HDMI_REG(VC4_HDMI_HOTPLUG_INT), HDMI_REG(VC4_HDMI_HOTPLUG), + HDMI_REG(VC4_HDMI_RAM_PACKET_CONFIG), HDMI_REG(VC4_HDMI_HORZA), HDMI_REG(VC4_HDMI_HORZB), HDMI_REG(VC4_HDMI_FIFO_CTL), @@ -165,7 +167,8 @@ vc4_hdmi_connector_detect(struct drm_connector *connector, bool force) struct vc4_dev *vc4 = to_vc4_dev(dev); if (vc4->hdmi->hpd_gpio) { - if (gpio_get_value(vc4->hdmi->hpd_gpio)) + if (gpio_get_value_cansleep(vc4->hdmi->hpd_gpio) ^ + vc4->hdmi->hpd_active_low) return connector_status_connected; else return connector_status_disconnected; @@ -495,6 +498,16 @@ static int vc4_hdmi_bind(struct device *dev, struct device *master, void *data) goto err_put_i2c; } + /* This is the rate that is set by the firmware. The number + * needs to be a bit higher than the pixel clock rate + * (generally 148.5Mhz). + */ + ret = clk_set_rate(hdmi->hsm_clock, 163682864); + if (ret) { + DRM_ERROR("Failed to set HSM clock rate: %d\n", ret); + goto err_unprepare_pix; + } + ret = clk_prepare_enable(hdmi->hsm_clock); if (ret) { DRM_ERROR("Failed to turn on HDMI state machine clock: %d\n", @@ -506,17 +519,40 @@ static int vc4_hdmi_bind(struct device *dev, struct device *master, void *data) * we'll use the HDMI core's register. */ if (of_find_property(dev->of_node, "hpd-gpios", &value)) { - hdmi->hpd_gpio = of_get_named_gpio(dev->of_node, "hpd-gpios", 0); + enum of_gpio_flags hpd_gpio_flags; + + hdmi->hpd_gpio = of_get_named_gpio_flags(dev->of_node, + "hpd-gpios", 0, + &hpd_gpio_flags); if (hdmi->hpd_gpio < 0) { ret = hdmi->hpd_gpio; goto err_unprepare_hsm; } + + hdmi->hpd_active_low = hpd_gpio_flags & OF_GPIO_ACTIVE_LOW; } vc4->hdmi = hdmi; /* HDMI core must be enabled. */ - WARN_ON_ONCE((HD_READ(VC4_HD_M_CTL) & VC4_HD_M_ENABLE) == 0); + if (!(HD_READ(VC4_HD_M_CTL) & VC4_HD_M_ENABLE)) { + HD_WRITE(VC4_HD_M_CTL, VC4_HD_M_SW_RST); + udelay(1); + HD_WRITE(VC4_HD_M_CTL, 0); + + HD_WRITE(VC4_HD_M_CTL, VC4_HD_M_ENABLE); + + HDMI_WRITE(VC4_HDMI_SW_RESET_CONTROL, + VC4_HDMI_SW_RESET_HDMI | + VC4_HDMI_SW_RESET_FORMAT_DETECT); + + HDMI_WRITE(VC4_HDMI_SW_RESET_CONTROL, 0); + + /* PHY should be in reset, like + * vc4_hdmi_encoder_disable() does. + */ + HDMI_WRITE(VC4_HDMI_TX_PHY_RESET_CTL, 0xf << 16); + } drm_encoder_init(drm, hdmi->encoder, &vc4_hdmi_encoder_funcs, DRM_MODE_ENCODER_TMDS, NULL); diff --git a/drivers/gpu/drm/vc4/vc4_hvs.c b/drivers/gpu/drm/vc4/vc4_hvs.c index 8098c5b21..6fbab1c82 100644 --- a/drivers/gpu/drm/vc4/vc4_hvs.c +++ b/drivers/gpu/drm/vc4/vc4_hvs.c @@ -100,12 +100,76 @@ int vc4_hvs_debugfs_regs(struct seq_file *m, void *unused) } #endif +/* The filter kernel is composed of dwords each containing 3 9-bit + * signed integers packed next to each other. + */ +#define VC4_INT_TO_COEFF(coeff) (coeff & 0x1ff) +#define VC4_PPF_FILTER_WORD(c0, c1, c2) \ + ((((c0) & 0x1ff) << 0) | \ + (((c1) & 0x1ff) << 9) | \ + (((c2) & 0x1ff) << 18)) + +/* The whole filter kernel is arranged as the coefficients 0-16 going + * up, then a pad, then 17-31 going down and reversed within the + * dwords. This means that a linear phase kernel (where it's + * symmetrical at the boundary between 15 and 16) has the last 5 + * dwords matching the first 5, but reversed. + */ +#define VC4_LINEAR_PHASE_KERNEL(c0, c1, c2, c3, c4, c5, c6, c7, c8, \ + c9, c10, c11, c12, c13, c14, c15) \ + {VC4_PPF_FILTER_WORD(c0, c1, c2), \ + VC4_PPF_FILTER_WORD(c3, c4, c5), \ + VC4_PPF_FILTER_WORD(c6, c7, c8), \ + VC4_PPF_FILTER_WORD(c9, c10, c11), \ + VC4_PPF_FILTER_WORD(c12, c13, c14), \ + VC4_PPF_FILTER_WORD(c15, c15, 0)} + +#define VC4_LINEAR_PHASE_KERNEL_DWORDS 6 +#define VC4_KERNEL_DWORDS (VC4_LINEAR_PHASE_KERNEL_DWORDS * 2 - 1) + +/* Recommended B=1/3, C=1/3 filter choice from Mitchell/Netravali. + * http://www.cs.utexas.edu/~fussell/courses/cs384g/lectures/mitchell/Mitchell.pdf + */ +static const u32 mitchell_netravali_1_3_1_3_kernel[] = + VC4_LINEAR_PHASE_KERNEL(0, -2, -6, -8, -10, -8, -3, 2, 18, + 50, 82, 119, 155, 187, 213, 227); + +static int vc4_hvs_upload_linear_kernel(struct vc4_hvs *hvs, + struct drm_mm_node *space, + const u32 *kernel) +{ + int ret, i; + u32 __iomem *dst_kernel; + + ret = drm_mm_insert_node(&hvs->dlist_mm, space, VC4_KERNEL_DWORDS, 1, + 0); + if (ret) { + DRM_ERROR("Failed to allocate space for filter kernel: %d\n", + ret); + return ret; + } + + dst_kernel = hvs->dlist + space->start; + + for (i = 0; i < VC4_KERNEL_DWORDS; i++) { + if (i < VC4_LINEAR_PHASE_KERNEL_DWORDS) + writel(kernel[i], &dst_kernel[i]); + else { + writel(kernel[VC4_KERNEL_DWORDS - i - 1], + &dst_kernel[i]); + } + } + + return 0; +} + static int vc4_hvs_bind(struct device *dev, struct device *master, void *data) { struct platform_device *pdev = to_platform_device(dev); struct drm_device *drm = dev_get_drvdata(master); struct vc4_dev *vc4 = drm->dev_private; struct vc4_hvs *hvs = NULL; + int ret; hvs = devm_kzalloc(&pdev->dev, sizeof(*hvs), GFP_KERNEL); if (!hvs) @@ -119,6 +183,33 @@ static int vc4_hvs_bind(struct device *dev, struct device *master, void *data) hvs->dlist = hvs->regs + SCALER_DLIST_START; + spin_lock_init(&hvs->mm_lock); + + /* Set up the HVS display list memory manager. We never + * overwrite the setup from the bootloader (just 128b out of + * our 16K), since we don't want to scramble the screen when + * transitioning from the firmware's boot setup to runtime. + */ + drm_mm_init(&hvs->dlist_mm, + HVS_BOOTLOADER_DLIST_END, + (SCALER_DLIST_SIZE >> 2) - HVS_BOOTLOADER_DLIST_END); + + /* Set up the HVS LBM memory manager. We could have some more + * complicated data structure that allowed reuse of LBM areas + * between planes when they don't overlap on the screen, but + * for now we just allocate globally. + */ + drm_mm_init(&hvs->lbm_mm, 0, 96 * 1024); + + /* Upload filter kernels. We only have the one for now, so we + * keep it around for the lifetime of the driver. + */ + ret = vc4_hvs_upload_linear_kernel(hvs, + &hvs->mitchell_netravali_filter, + mitchell_netravali_1_3_1_3_kernel); + if (ret) + return ret; + vc4->hvs = hvs; return 0; } @@ -129,6 +220,12 @@ static void vc4_hvs_unbind(struct device *dev, struct device *master, struct drm_device *drm = dev_get_drvdata(master); struct vc4_dev *vc4 = drm->dev_private; + if (vc4->hvs->mitchell_netravali_filter.allocated) + drm_mm_remove_node(&vc4->hvs->mitchell_netravali_filter); + + drm_mm_takedown(&vc4->hvs->dlist_mm); + drm_mm_takedown(&vc4->hvs->lbm_mm); + vc4->hvs = NULL; } diff --git a/drivers/gpu/drm/vc4/vc4_irq.c b/drivers/gpu/drm/vc4/vc4_irq.c index 78a21357f..b0104a346 100644 --- a/drivers/gpu/drm/vc4/vc4_irq.c +++ b/drivers/gpu/drm/vc4/vc4_irq.c @@ -30,6 +30,10 @@ * disables that specific interrupt, and 0s written are ignored * (reading either one returns the set of enabled interrupts). * + * When we take a binning flush done interrupt, we need to submit the + * next frame for binning and move the finished frame to the render + * thread. + * * When we take a render frame interrupt, we need to wake the * processes waiting for some frame to be done, and get the next frame * submitted ASAP (so the hardware doesn't sit idle when there's work @@ -44,6 +48,7 @@ #include "vc4_regs.h" #define V3D_DRIVER_IRQS (V3D_INT_OUTOMEM | \ + V3D_INT_FLDONE | \ V3D_INT_FRDONE) DECLARE_WAIT_QUEUE_HEAD(render_wait); @@ -77,7 +82,7 @@ vc4_overflow_mem_work(struct work_struct *work) unsigned long irqflags; spin_lock_irqsave(&vc4->job_lock, irqflags); - current_exec = vc4_first_job(vc4); + current_exec = vc4_first_bin_job(vc4); if (current_exec) { vc4->overflow_mem->seqno = vc4->finished_seqno + 1; list_add_tail(&vc4->overflow_mem->unref_head, @@ -98,17 +103,43 @@ vc4_overflow_mem_work(struct work_struct *work) } static void -vc4_irq_finish_job(struct drm_device *dev) +vc4_irq_finish_bin_job(struct drm_device *dev) +{ + struct vc4_dev *vc4 = to_vc4_dev(dev); + struct vc4_exec_info *exec = vc4_first_bin_job(vc4); + + if (!exec) + return; + + vc4_move_job_to_render(dev, exec); + vc4_submit_next_bin_job(dev); +} + +static void +vc4_cancel_bin_job(struct drm_device *dev) +{ + struct vc4_dev *vc4 = to_vc4_dev(dev); + struct vc4_exec_info *exec = vc4_first_bin_job(vc4); + + if (!exec) + return; + + list_move_tail(&exec->head, &vc4->bin_job_list); + vc4_submit_next_bin_job(dev); +} + +static void +vc4_irq_finish_render_job(struct drm_device *dev) { struct vc4_dev *vc4 = to_vc4_dev(dev); - struct vc4_exec_info *exec = vc4_first_job(vc4); + struct vc4_exec_info *exec = vc4_first_render_job(vc4); if (!exec) return; vc4->finished_seqno++; list_move_tail(&exec->head, &vc4->job_done_list); - vc4_submit_next_job(dev); + vc4_submit_next_render_job(dev); wake_up_all(&vc4->job_wait_queue); schedule_work(&vc4->job_done_work); @@ -125,9 +156,10 @@ vc4_irq(int irq, void *arg) barrier(); intctl = V3D_READ(V3D_INTCTL); - /* Acknowledge the interrupts we're handling here. The render - * frame done interrupt will be cleared, while OUTOMEM will - * stay high until the underlying cause is cleared. + /* Acknowledge the interrupts we're handling here. The binner + * last flush / render frame done interrupt will be cleared, + * while OUTOMEM will stay high until the underlying cause is + * cleared. */ V3D_WRITE(V3D_INTCTL, intctl); @@ -138,9 +170,16 @@ vc4_irq(int irq, void *arg) status = IRQ_HANDLED; } + if (intctl & V3D_INT_FLDONE) { + spin_lock(&vc4->job_lock); + vc4_irq_finish_bin_job(dev); + spin_unlock(&vc4->job_lock); + status = IRQ_HANDLED; + } + if (intctl & V3D_INT_FRDONE) { spin_lock(&vc4->job_lock); - vc4_irq_finish_job(dev); + vc4_irq_finish_render_job(dev); spin_unlock(&vc4->job_lock); status = IRQ_HANDLED; } @@ -205,6 +244,7 @@ void vc4_irq_reset(struct drm_device *dev) V3D_WRITE(V3D_INTENA, V3D_DRIVER_IRQS); spin_lock_irqsave(&vc4->job_lock, irqflags); - vc4_irq_finish_job(dev); + vc4_cancel_bin_job(dev); + vc4_irq_finish_render_job(dev); spin_unlock_irqrestore(&vc4->job_lock, irqflags); } diff --git a/drivers/gpu/drm/vc4/vc4_kms.c b/drivers/gpu/drm/vc4/vc4_kms.c index f95f2df5f..4718ae517 100644 --- a/drivers/gpu/drm/vc4/vc4_kms.c +++ b/drivers/gpu/drm/vc4/vc4_kms.c @@ -49,6 +49,15 @@ vc4_atomic_complete_commit(struct vc4_commit *c) drm_atomic_helper_commit_modeset_enables(dev, state); + /* Make sure that drm_atomic_helper_wait_for_vblanks() + * actually waits for vblank. If we're doing a full atomic + * modeset (as opposed to a vc4_update_plane() short circuit), + * then we need to wait for scanout to be done with our + * display lists before we free it and potentially reallocate + * and overwrite the dlist memory with a new modeset. + */ + state->legacy_cursor_update = false; + drm_atomic_helper_wait_for_vblanks(dev, state); drm_atomic_helper_cleanup_planes(dev, state); diff --git a/drivers/gpu/drm/vc4/vc4_plane.c b/drivers/gpu/drm/vc4/vc4_plane.c index 0addbad15..7b0c72ae0 100644 --- a/drivers/gpu/drm/vc4/vc4_plane.c +++ b/drivers/gpu/drm/vc4/vc4_plane.c @@ -24,19 +24,52 @@ #include "drm_fb_cma_helper.h" #include "drm_plane_helper.h" +enum vc4_scaling_mode { + VC4_SCALING_NONE, + VC4_SCALING_TPZ, + VC4_SCALING_PPF, +}; + struct vc4_plane_state { struct drm_plane_state base; + /* System memory copy of the display list for this element, computed + * at atomic_check time. + */ u32 *dlist; - u32 dlist_size; /* Number of dwords in allocated for the display list */ + u32 dlist_size; /* Number of dwords allocated for the display list */ u32 dlist_count; /* Number of used dwords in the display list. */ - /* Offset in the dlist to pointer word 0. */ - u32 pw0_offset; + /* Offset in the dlist to various words, for pageflip or + * cursor updates. + */ + u32 pos0_offset; + u32 pos2_offset; + u32 ptr0_offset; /* Offset where the plane's dlist was last stored in the - hardware at vc4_crtc_atomic_flush() time. - */ - u32 *hw_dlist; + * hardware at vc4_crtc_atomic_flush() time. + */ + u32 __iomem *hw_dlist; + + /* Clipped coordinates of the plane on the display. */ + int crtc_x, crtc_y, crtc_w, crtc_h; + /* Clipped area being scanned from in the FB. */ + u32 src_x, src_y; + + u32 src_w[2], src_h[2]; + + /* Scaling selection for the RGB/Y plane and the Cb/Cr planes. */ + enum vc4_scaling_mode x_scaling[2], y_scaling[2]; + bool is_unity; + bool is_yuv; + + /* Offset to start scanning out from the start of the plane's + * BO. + */ + u32 offsets[3]; + + /* Our allocation in LBM for temporary storage during scaling. */ + struct drm_mm_node lbm; }; static inline struct vc4_plane_state * @@ -50,6 +83,7 @@ static const struct hvs_format { u32 hvs; /* HVS_FORMAT_* */ u32 pixel_order; bool has_alpha; + bool flip_cbcr; } hvs_formats[] = { { .drm = DRM_FORMAT_XRGB8888, .hvs = HVS_PIXEL_FORMAT_RGBA8888, @@ -59,6 +93,48 @@ static const struct hvs_format { .drm = DRM_FORMAT_ARGB8888, .hvs = HVS_PIXEL_FORMAT_RGBA8888, .pixel_order = HVS_PIXEL_ORDER_ABGR, .has_alpha = true, }, + { + .drm = DRM_FORMAT_RGB565, .hvs = HVS_PIXEL_FORMAT_RGB565, + .pixel_order = HVS_PIXEL_ORDER_XRGB, .has_alpha = false, + }, + { + .drm = DRM_FORMAT_BGR565, .hvs = HVS_PIXEL_FORMAT_RGB565, + .pixel_order = HVS_PIXEL_ORDER_XBGR, .has_alpha = false, + }, + { + .drm = DRM_FORMAT_ARGB1555, .hvs = HVS_PIXEL_FORMAT_RGBA5551, + .pixel_order = HVS_PIXEL_ORDER_ABGR, .has_alpha = true, + }, + { + .drm = DRM_FORMAT_XRGB1555, .hvs = HVS_PIXEL_FORMAT_RGBA5551, + .pixel_order = HVS_PIXEL_ORDER_ABGR, .has_alpha = false, + }, + { + .drm = DRM_FORMAT_YUV422, + .hvs = HVS_PIXEL_FORMAT_YCBCR_YUV422_3PLANE, + }, + { + .drm = DRM_FORMAT_YVU422, + .hvs = HVS_PIXEL_FORMAT_YCBCR_YUV422_3PLANE, + .flip_cbcr = true, + }, + { + .drm = DRM_FORMAT_YUV420, + .hvs = HVS_PIXEL_FORMAT_YCBCR_YUV420_3PLANE, + }, + { + .drm = DRM_FORMAT_YVU420, + .hvs = HVS_PIXEL_FORMAT_YCBCR_YUV420_3PLANE, + .flip_cbcr = true, + }, + { + .drm = DRM_FORMAT_NV12, + .hvs = HVS_PIXEL_FORMAT_YCBCR_YUV420_2PLANE, + }, + { + .drm = DRM_FORMAT_NV16, + .hvs = HVS_PIXEL_FORMAT_YCBCR_YUV422_2PLANE, + }, }; static const struct hvs_format *vc4_get_hvs_format(u32 drm_format) @@ -73,6 +149,16 @@ static const struct hvs_format *vc4_get_hvs_format(u32 drm_format) return NULL; } +static enum vc4_scaling_mode vc4_get_scaling_mode(u32 src, u32 dst) +{ + if (dst > src) + return VC4_SCALING_PPF; + else if (dst < src) + return VC4_SCALING_TPZ; + else + return VC4_SCALING_NONE; +} + static bool plane_enabled(struct drm_plane_state *state) { return state->fb && state->crtc; @@ -89,6 +175,8 @@ static struct drm_plane_state *vc4_plane_duplicate_state(struct drm_plane *plane if (!vc4_state) return NULL; + memset(&vc4_state->lbm, 0, sizeof(vc4_state->lbm)); + __drm_atomic_helper_plane_duplicate_state(plane, &vc4_state->base); if (vc4_state->dlist) { @@ -108,8 +196,17 @@ static struct drm_plane_state *vc4_plane_duplicate_state(struct drm_plane *plane static void vc4_plane_destroy_state(struct drm_plane *plane, struct drm_plane_state *state) { + struct vc4_dev *vc4 = to_vc4_dev(plane->dev); struct vc4_plane_state *vc4_state = to_vc4_plane_state(state); + if (vc4_state->lbm.allocated) { + unsigned long irqflags; + + spin_lock_irqsave(&vc4->hvs->mm_lock, irqflags); + drm_mm_remove_node(&vc4_state->lbm); + spin_unlock_irqrestore(&vc4->hvs->mm_lock, irqflags); + } + kfree(vc4_state->dlist); __drm_atomic_helper_plane_destroy_state(plane, &vc4_state->base); kfree(state); @@ -148,84 +245,400 @@ static void vc4_dlist_write(struct vc4_plane_state *vc4_state, u32 val) vc4_state->dlist[vc4_state->dlist_count++] = val; } +/* Returns the scl0/scl1 field based on whether the dimensions need to + * be up/down/non-scaled. + * + * This is a replication of a table from the spec. + */ +static u32 vc4_get_scl_field(struct drm_plane_state *state, int plane) +{ + struct vc4_plane_state *vc4_state = to_vc4_plane_state(state); + + switch (vc4_state->x_scaling[plane] << 2 | vc4_state->y_scaling[plane]) { + case VC4_SCALING_PPF << 2 | VC4_SCALING_PPF: + return SCALER_CTL0_SCL_H_PPF_V_PPF; + case VC4_SCALING_TPZ << 2 | VC4_SCALING_PPF: + return SCALER_CTL0_SCL_H_TPZ_V_PPF; + case VC4_SCALING_PPF << 2 | VC4_SCALING_TPZ: + return SCALER_CTL0_SCL_H_PPF_V_TPZ; + case VC4_SCALING_TPZ << 2 | VC4_SCALING_TPZ: + return SCALER_CTL0_SCL_H_TPZ_V_TPZ; + case VC4_SCALING_PPF << 2 | VC4_SCALING_NONE: + return SCALER_CTL0_SCL_H_PPF_V_NONE; + case VC4_SCALING_NONE << 2 | VC4_SCALING_PPF: + return SCALER_CTL0_SCL_H_NONE_V_PPF; + case VC4_SCALING_NONE << 2 | VC4_SCALING_TPZ: + return SCALER_CTL0_SCL_H_NONE_V_TPZ; + case VC4_SCALING_TPZ << 2 | VC4_SCALING_NONE: + return SCALER_CTL0_SCL_H_TPZ_V_NONE; + default: + case VC4_SCALING_NONE << 2 | VC4_SCALING_NONE: + /* The unity case is independently handled by + * SCALER_CTL0_UNITY. + */ + return 0; + } +} + +static int vc4_plane_setup_clipping_and_scaling(struct drm_plane_state *state) +{ + struct drm_plane *plane = state->plane; + struct vc4_plane_state *vc4_state = to_vc4_plane_state(state); + struct drm_framebuffer *fb = state->fb; + struct drm_gem_cma_object *bo = drm_fb_cma_get_gem_obj(fb, 0); + u32 subpixel_src_mask = (1 << 16) - 1; + u32 format = fb->pixel_format; + int num_planes = drm_format_num_planes(format); + u32 h_subsample = 1; + u32 v_subsample = 1; + int i; + + for (i = 0; i < num_planes; i++) + vc4_state->offsets[i] = bo->paddr + fb->offsets[i]; + + /* We don't support subpixel source positioning for scaling. */ + if ((state->src_x & subpixel_src_mask) || + (state->src_y & subpixel_src_mask) || + (state->src_w & subpixel_src_mask) || + (state->src_h & subpixel_src_mask)) { + return -EINVAL; + } + + vc4_state->src_x = state->src_x >> 16; + vc4_state->src_y = state->src_y >> 16; + vc4_state->src_w[0] = state->src_w >> 16; + vc4_state->src_h[0] = state->src_h >> 16; + + vc4_state->crtc_x = state->crtc_x; + vc4_state->crtc_y = state->crtc_y; + vc4_state->crtc_w = state->crtc_w; + vc4_state->crtc_h = state->crtc_h; + + vc4_state->x_scaling[0] = vc4_get_scaling_mode(vc4_state->src_w[0], + vc4_state->crtc_w); + vc4_state->y_scaling[0] = vc4_get_scaling_mode(vc4_state->src_h[0], + vc4_state->crtc_h); + + if (num_planes > 1) { + vc4_state->is_yuv = true; + + h_subsample = drm_format_horz_chroma_subsampling(format); + v_subsample = drm_format_vert_chroma_subsampling(format); + vc4_state->src_w[1] = vc4_state->src_w[0] / h_subsample; + vc4_state->src_h[1] = vc4_state->src_h[0] / v_subsample; + + vc4_state->x_scaling[1] = + vc4_get_scaling_mode(vc4_state->src_w[1], + vc4_state->crtc_w); + vc4_state->y_scaling[1] = + vc4_get_scaling_mode(vc4_state->src_h[1], + vc4_state->crtc_h); + + /* YUV conversion requires that scaling be enabled, + * even on a plane that's otherwise 1:1. Choose TPZ + * for simplicity. + */ + if (vc4_state->x_scaling[0] == VC4_SCALING_NONE) + vc4_state->x_scaling[0] = VC4_SCALING_TPZ; + if (vc4_state->y_scaling[0] == VC4_SCALING_NONE) + vc4_state->y_scaling[0] = VC4_SCALING_TPZ; + } + + vc4_state->is_unity = (vc4_state->x_scaling[0] == VC4_SCALING_NONE && + vc4_state->y_scaling[0] == VC4_SCALING_NONE && + vc4_state->x_scaling[1] == VC4_SCALING_NONE && + vc4_state->y_scaling[1] == VC4_SCALING_NONE); + + /* No configuring scaling on the cursor plane, since it gets + non-vblank-synced updates, and scaling requires requires + LBM changes which have to be vblank-synced. + */ + if (plane->type == DRM_PLANE_TYPE_CURSOR && !vc4_state->is_unity) + return -EINVAL; + + /* Clamp the on-screen start x/y to 0. The hardware doesn't + * support negative y, and negative x wastes bandwidth. + */ + if (vc4_state->crtc_x < 0) { + for (i = 0; i < num_planes; i++) { + u32 cpp = drm_format_plane_cpp(fb->pixel_format, i); + u32 subs = ((i == 0) ? 1 : h_subsample); + + vc4_state->offsets[i] += (cpp * + (-vc4_state->crtc_x) / subs); + } + vc4_state->src_w[0] += vc4_state->crtc_x; + vc4_state->src_w[1] += vc4_state->crtc_x / h_subsample; + vc4_state->crtc_x = 0; + } + + if (vc4_state->crtc_y < 0) { + for (i = 0; i < num_planes; i++) { + u32 subs = ((i == 0) ? 1 : v_subsample); + + vc4_state->offsets[i] += (fb->pitches[i] * + (-vc4_state->crtc_y) / subs); + } + vc4_state->src_h[0] += vc4_state->crtc_y; + vc4_state->src_h[1] += vc4_state->crtc_y / v_subsample; + vc4_state->crtc_y = 0; + } + + return 0; +} + +static void vc4_write_tpz(struct vc4_plane_state *vc4_state, u32 src, u32 dst) +{ + u32 scale, recip; + + scale = (1 << 16) * src / dst; + + /* The specs note that while the reciprocal would be defined + * as (1<<32)/scale, ~0 is close enough. + */ + recip = ~0 / scale; + + vc4_dlist_write(vc4_state, + VC4_SET_FIELD(scale, SCALER_TPZ0_SCALE) | + VC4_SET_FIELD(0, SCALER_TPZ0_IPHASE)); + vc4_dlist_write(vc4_state, + VC4_SET_FIELD(recip, SCALER_TPZ1_RECIP)); +} + +static void vc4_write_ppf(struct vc4_plane_state *vc4_state, u32 src, u32 dst) +{ + u32 scale = (1 << 16) * src / dst; + + vc4_dlist_write(vc4_state, + SCALER_PPF_AGC | + VC4_SET_FIELD(scale, SCALER_PPF_SCALE) | + VC4_SET_FIELD(0, SCALER_PPF_IPHASE)); +} + +static u32 vc4_lbm_size(struct drm_plane_state *state) +{ + struct vc4_plane_state *vc4_state = to_vc4_plane_state(state); + /* This is the worst case number. One of the two sizes will + * be used depending on the scaling configuration. + */ + u32 pix_per_line = max(vc4_state->src_w[0], (u32)vc4_state->crtc_w); + u32 lbm; + + if (!vc4_state->is_yuv) { + if (vc4_state->is_unity) + return 0; + else if (vc4_state->y_scaling[0] == VC4_SCALING_TPZ) + lbm = pix_per_line * 8; + else { + /* In special cases, this multiplier might be 12. */ + lbm = pix_per_line * 16; + } + } else { + /* There are cases for this going down to a multiplier + * of 2, but according to the firmware source, the + * table in the docs is somewhat wrong. + */ + lbm = pix_per_line * 16; + } + + lbm = roundup(lbm, 32); + + return lbm; +} + +static void vc4_write_scaling_parameters(struct drm_plane_state *state, + int channel) +{ + struct vc4_plane_state *vc4_state = to_vc4_plane_state(state); + + /* Ch0 H-PPF Word 0: Scaling Parameters */ + if (vc4_state->x_scaling[channel] == VC4_SCALING_PPF) { + vc4_write_ppf(vc4_state, + vc4_state->src_w[channel], vc4_state->crtc_w); + } + + /* Ch0 V-PPF Words 0-1: Scaling Parameters, Context */ + if (vc4_state->y_scaling[channel] == VC4_SCALING_PPF) { + vc4_write_ppf(vc4_state, + vc4_state->src_h[channel], vc4_state->crtc_h); + vc4_dlist_write(vc4_state, 0xc0c0c0c0); + } + + /* Ch0 H-TPZ Words 0-1: Scaling Parameters, Recip */ + if (vc4_state->x_scaling[channel] == VC4_SCALING_TPZ) { + vc4_write_tpz(vc4_state, + vc4_state->src_w[channel], vc4_state->crtc_w); + } + + /* Ch0 V-TPZ Words 0-2: Scaling Parameters, Recip, Context */ + if (vc4_state->y_scaling[channel] == VC4_SCALING_TPZ) { + vc4_write_tpz(vc4_state, + vc4_state->src_h[channel], vc4_state->crtc_h); + vc4_dlist_write(vc4_state, 0xc0c0c0c0); + } +} + /* Writes out a full display list for an active plane to the plane's * private dlist state. */ static int vc4_plane_mode_set(struct drm_plane *plane, struct drm_plane_state *state) { + struct vc4_dev *vc4 = to_vc4_dev(plane->dev); struct vc4_plane_state *vc4_state = to_vc4_plane_state(state); struct drm_framebuffer *fb = state->fb; - struct drm_gem_cma_object *bo = drm_fb_cma_get_gem_obj(fb, 0); u32 ctl0_offset = vc4_state->dlist_count; const struct hvs_format *format = vc4_get_hvs_format(fb->pixel_format); - uint32_t offset = fb->offsets[0]; - int crtc_x = state->crtc_x; - int crtc_y = state->crtc_y; - int crtc_w = state->crtc_w; - int crtc_h = state->crtc_h; - - if (state->crtc_w << 16 != state->src_w || - state->crtc_h << 16 != state->src_h) { - /* We don't support scaling yet, which involves - * allocating the LBM memory for scaling temporary - * storage, and putting filter kernels in the HVS - * context. - */ - return -EINVAL; + int num_planes = drm_format_num_planes(format->drm); + u32 scl0, scl1; + u32 lbm_size; + unsigned long irqflags; + int ret, i; + + ret = vc4_plane_setup_clipping_and_scaling(state); + if (ret) + return ret; + + /* Allocate the LBM memory that the HVS will use for temporary + * storage due to our scaling/format conversion. + */ + lbm_size = vc4_lbm_size(state); + if (lbm_size) { + if (!vc4_state->lbm.allocated) { + spin_lock_irqsave(&vc4->hvs->mm_lock, irqflags); + ret = drm_mm_insert_node(&vc4->hvs->lbm_mm, + &vc4_state->lbm, + lbm_size, 32, 0); + spin_unlock_irqrestore(&vc4->hvs->mm_lock, irqflags); + } else { + WARN_ON_ONCE(lbm_size != vc4_state->lbm.size); + } } - if (crtc_x < 0) { - offset += drm_format_plane_cpp(fb->pixel_format, 0) * -crtc_x; - crtc_w += crtc_x; - crtc_x = 0; - } + if (ret) + return ret; - if (crtc_y < 0) { - offset += fb->pitches[0] * -crtc_y; - crtc_h += crtc_y; - crtc_y = 0; + /* SCL1 is used for Cb/Cr scaling of planar formats. For RGB + * and 4:4:4, scl1 should be set to scl0 so both channels of + * the scaler do the same thing. For YUV, the Y plane needs + * to be put in channel 1 and Cb/Cr in channel 0, so we swap + * the scl fields here. + */ + if (num_planes == 1) { + scl0 = vc4_get_scl_field(state, 1); + scl1 = scl0; + } else { + scl0 = vc4_get_scl_field(state, 1); + scl1 = vc4_get_scl_field(state, 0); } + /* Control word */ vc4_dlist_write(vc4_state, SCALER_CTL0_VALID | (format->pixel_order << SCALER_CTL0_ORDER_SHIFT) | (format->hvs << SCALER_CTL0_PIXEL_FORMAT_SHIFT) | - SCALER_CTL0_UNITY); + (vc4_state->is_unity ? SCALER_CTL0_UNITY : 0) | + VC4_SET_FIELD(scl0, SCALER_CTL0_SCL0) | + VC4_SET_FIELD(scl1, SCALER_CTL0_SCL1)); /* Position Word 0: Image Positions and Alpha Value */ + vc4_state->pos0_offset = vc4_state->dlist_count; vc4_dlist_write(vc4_state, VC4_SET_FIELD(0xff, SCALER_POS0_FIXED_ALPHA) | - VC4_SET_FIELD(crtc_x, SCALER_POS0_START_X) | - VC4_SET_FIELD(crtc_y, SCALER_POS0_START_Y)); - - /* Position Word 1: Scaled Image Dimensions. - * Skipped due to SCALER_CTL0_UNITY scaling. - */ + VC4_SET_FIELD(vc4_state->crtc_x, SCALER_POS0_START_X) | + VC4_SET_FIELD(vc4_state->crtc_y, SCALER_POS0_START_Y)); + + /* Position Word 1: Scaled Image Dimensions. */ + if (!vc4_state->is_unity) { + vc4_dlist_write(vc4_state, + VC4_SET_FIELD(vc4_state->crtc_w, + SCALER_POS1_SCL_WIDTH) | + VC4_SET_FIELD(vc4_state->crtc_h, + SCALER_POS1_SCL_HEIGHT)); + } /* Position Word 2: Source Image Size, Alpha Mode */ + vc4_state->pos2_offset = vc4_state->dlist_count; vc4_dlist_write(vc4_state, VC4_SET_FIELD(format->has_alpha ? SCALER_POS2_ALPHA_MODE_PIPELINE : SCALER_POS2_ALPHA_MODE_FIXED, SCALER_POS2_ALPHA_MODE) | - VC4_SET_FIELD(crtc_w, SCALER_POS2_WIDTH) | - VC4_SET_FIELD(crtc_h, SCALER_POS2_HEIGHT)); + VC4_SET_FIELD(vc4_state->src_w[0], SCALER_POS2_WIDTH) | + VC4_SET_FIELD(vc4_state->src_h[0], SCALER_POS2_HEIGHT)); /* Position Word 3: Context. Written by the HVS. */ vc4_dlist_write(vc4_state, 0xc0c0c0c0); - vc4_state->pw0_offset = vc4_state->dlist_count; - /* Pointer Word 0: RGB / Y Pointer */ - vc4_dlist_write(vc4_state, bo->paddr + offset); + /* Pointer Word 0/1/2: RGB / Y / Cb / Cr Pointers + * + * The pointers may be any byte address. + */ + vc4_state->ptr0_offset = vc4_state->dlist_count; + if (!format->flip_cbcr) { + for (i = 0; i < num_planes; i++) + vc4_dlist_write(vc4_state, vc4_state->offsets[i]); + } else { + WARN_ON_ONCE(num_planes != 3); + vc4_dlist_write(vc4_state, vc4_state->offsets[0]); + vc4_dlist_write(vc4_state, vc4_state->offsets[2]); + vc4_dlist_write(vc4_state, vc4_state->offsets[1]); + } - /* Pointer Context Word 0: Written by the HVS */ - vc4_dlist_write(vc4_state, 0xc0c0c0c0); + /* Pointer Context Word 0/1/2: Written by the HVS */ + for (i = 0; i < num_planes; i++) + vc4_dlist_write(vc4_state, 0xc0c0c0c0); - /* Pitch word 0: Pointer 0 Pitch */ - vc4_dlist_write(vc4_state, - VC4_SET_FIELD(fb->pitches[0], SCALER_SRC_PITCH)); + /* Pitch word 0/1/2 */ + for (i = 0; i < num_planes; i++) { + vc4_dlist_write(vc4_state, + VC4_SET_FIELD(fb->pitches[i], SCALER_SRC_PITCH)); + } + + /* Colorspace conversion words */ + if (vc4_state->is_yuv) { + vc4_dlist_write(vc4_state, SCALER_CSC0_ITR_R_601_5); + vc4_dlist_write(vc4_state, SCALER_CSC1_ITR_R_601_5); + vc4_dlist_write(vc4_state, SCALER_CSC2_ITR_R_601_5); + } + + if (!vc4_state->is_unity) { + /* LBM Base Address. */ + if (vc4_state->y_scaling[0] != VC4_SCALING_NONE || + vc4_state->y_scaling[1] != VC4_SCALING_NONE) { + vc4_dlist_write(vc4_state, vc4_state->lbm.start); + } + + if (num_planes > 1) { + /* Emit Cb/Cr as channel 0 and Y as channel + * 1. This matches how we set up scl0/scl1 + * above. + */ + vc4_write_scaling_parameters(state, 1); + } + vc4_write_scaling_parameters(state, 0); + + /* If any PPF setup was done, then all the kernel + * pointers get uploaded. + */ + if (vc4_state->x_scaling[0] == VC4_SCALING_PPF || + vc4_state->y_scaling[0] == VC4_SCALING_PPF || + vc4_state->x_scaling[1] == VC4_SCALING_PPF || + vc4_state->y_scaling[1] == VC4_SCALING_PPF) { + u32 kernel = VC4_SET_FIELD(vc4->hvs->mitchell_netravali_filter.start, + SCALER_PPF_KERNEL_OFFSET); + + /* HPPF plane 0 */ + vc4_dlist_write(vc4_state, kernel); + /* VPPF plane 0 */ + vc4_dlist_write(vc4_state, kernel); + /* HPPF plane 1 */ + vc4_dlist_write(vc4_state, kernel); + /* VPPF plane 1 */ + vc4_dlist_write(vc4_state, kernel); + } + } vc4_state->dlist[ctl0_offset] |= VC4_SET_FIELD(vc4_state->dlist_count, SCALER_CTL0_SIZE); @@ -303,13 +716,13 @@ void vc4_plane_async_set_fb(struct drm_plane *plane, struct drm_framebuffer *fb) * scanout will start from this address as soon as the FIFO * needs to refill with pixels. */ - writel(addr, &vc4_state->hw_dlist[vc4_state->pw0_offset]); + writel(addr, &vc4_state->hw_dlist[vc4_state->ptr0_offset]); /* Also update the CPU-side dlist copy, so that any later * atomic updates that don't do a new modeset on our plane * also use our updated address. */ - vc4_state->dlist[vc4_state->pw0_offset] = addr; + vc4_state->dlist[vc4_state->ptr0_offset] = addr; } static const struct drm_plane_helper_funcs vc4_plane_helper_funcs = { @@ -325,8 +738,83 @@ static void vc4_plane_destroy(struct drm_plane *plane) drm_plane_cleanup(plane); } +/* Implements immediate (non-vblank-synced) updates of the cursor + * position, or falls back to the atomic helper otherwise. + */ +static int +vc4_update_plane(struct drm_plane *plane, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int crtc_x, int crtc_y, + unsigned int crtc_w, unsigned int crtc_h, + uint32_t src_x, uint32_t src_y, + uint32_t src_w, uint32_t src_h) +{ + struct drm_plane_state *plane_state; + struct vc4_plane_state *vc4_state; + + if (plane != crtc->cursor) + goto out; + + plane_state = plane->state; + vc4_state = to_vc4_plane_state(plane_state); + + if (!plane_state) + goto out; + + /* If we're changing the cursor contents, do that in the + * normal vblank-synced atomic path. + */ + if (fb != plane_state->fb) + goto out; + + /* No configuring new scaling in the fast path. */ + if (crtc_w != plane_state->crtc_w || + crtc_h != plane_state->crtc_h || + src_w != plane_state->src_w || + src_h != plane_state->src_h) { + goto out; + } + + /* Set the cursor's position on the screen. This is the + * expected change from the drm_mode_cursor_universal() + * helper. + */ + plane_state->crtc_x = crtc_x; + plane_state->crtc_y = crtc_y; + + /* Allow changing the start position within the cursor BO, if + * that matters. + */ + plane_state->src_x = src_x; + plane_state->src_y = src_y; + + /* Update the display list based on the new crtc_x/y. */ + vc4_plane_atomic_check(plane, plane_state); + + /* Note that we can't just call vc4_plane_write_dlist() + * because that would smash the context data that the HVS is + * currently using. + */ + writel(vc4_state->dlist[vc4_state->pos0_offset], + &vc4_state->hw_dlist[vc4_state->pos0_offset]); + writel(vc4_state->dlist[vc4_state->pos2_offset], + &vc4_state->hw_dlist[vc4_state->pos2_offset]); + writel(vc4_state->dlist[vc4_state->ptr0_offset], + &vc4_state->hw_dlist[vc4_state->ptr0_offset]); + + return 0; + +out: + return drm_atomic_helper_update_plane(plane, crtc, fb, + crtc_x, crtc_y, + crtc_w, crtc_h, + src_x, src_y, + src_w, src_h); +} + static const struct drm_plane_funcs vc4_plane_funcs = { - .update_plane = drm_atomic_helper_update_plane, + .update_plane = vc4_update_plane, .disable_plane = drm_atomic_helper_disable_plane, .destroy = vc4_plane_destroy, .set_property = NULL, @@ -341,6 +829,7 @@ struct drm_plane *vc4_plane_init(struct drm_device *dev, struct drm_plane *plane = NULL; struct vc4_plane *vc4_plane; u32 formats[ARRAY_SIZE(hvs_formats)]; + u32 num_formats = 0; int ret = 0; unsigned i; @@ -351,12 +840,20 @@ struct drm_plane *vc4_plane_init(struct drm_device *dev, goto fail; } - for (i = 0; i < ARRAY_SIZE(hvs_formats); i++) - formats[i] = hvs_formats[i].drm; + for (i = 0; i < ARRAY_SIZE(hvs_formats); i++) { + /* Don't allow YUV in cursor planes, since that means + * tuning on the scaler, which we don't allow for the + * cursor. + */ + if (type != DRM_PLANE_TYPE_CURSOR || + hvs_formats[i].hvs < HVS_PIXEL_FORMAT_YCBCR_YUV420_3PLANE) { + formats[num_formats++] = hvs_formats[i].drm; + } + } plane = &vc4_plane->base; ret = drm_universal_plane_init(dev, plane, 0xff, &vc4_plane_funcs, - formats, ARRAY_SIZE(formats), + formats, num_formats, type, NULL); drm_plane_helper_add(plane, &vc4_plane_helper_funcs); diff --git a/drivers/gpu/drm/vc4/vc4_regs.h b/drivers/gpu/drm/vc4/vc4_regs.h index 4e52a0a88..bf42a8e87 100644 --- a/drivers/gpu/drm/vc4/vc4_regs.h +++ b/drivers/gpu/drm/vc4/vc4_regs.h @@ -187,7 +187,7 @@ # define PV_VCONTROL_CONTINUOUS BIT(1) # define PV_VCONTROL_VIDEN BIT(0) -#define PV_VSYNCD 0x08 +#define PV_VSYNCD_EVEN 0x08 #define PV_HORZA 0x0c # define PV_HORZA_HBP_MASK VC4_MASK(31, 16) @@ -350,6 +350,17 @@ # define SCALER_DISPCTRLX_HEIGHT_SHIFT 0 #define SCALER_DISPBKGND0 0x00000044 +# define SCALER_DISPBKGND_AUTOHS BIT(31) +# define SCALER_DISPBKGND_INTERLACE BIT(30) +# define SCALER_DISPBKGND_GAMMA BIT(29) +# define SCALER_DISPBKGND_TESTMODE_MASK VC4_MASK(28, 25) +# define SCALER_DISPBKGND_TESTMODE_SHIFT 25 +/* Enables filling the scaler line with the RGB value in the low 24 + * bits before compositing. Costs cycles, so should be skipped if + * opaque display planes will cover everything. + */ +# define SCALER_DISPBKGND_FILL BIT(24) + #define SCALER_DISPSTAT0 0x00000048 #define SCALER_DISPBASE0 0x0000004c # define SCALER_DISPSTATX_MODE_MASK VC4_MASK(31, 30) @@ -362,6 +373,9 @@ # define SCALER_DISPSTATX_EMPTY BIT(28) #define SCALER_DISPCTRL1 0x00000050 #define SCALER_DISPBKGND1 0x00000054 +#define SCALER_DISPBKGNDX(x) (SCALER_DISPBKGND0 + \ + (x) * (SCALER_DISPBKGND1 - \ + SCALER_DISPBKGND0)) #define SCALER_DISPSTAT1 0x00000058 #define SCALER_DISPSTATX(x) (SCALER_DISPSTAT0 + \ (x) * (SCALER_DISPSTAT1 - \ @@ -456,6 +470,8 @@ #define VC4_HDMI_TX_PHY_RESET_CTL 0x2c0 #define VC4_HD_M_CTL 0x00c +# define VC4_HD_M_REGISTER_FILE_STANDBY (3 << 6) +# define VC4_HD_M_RAM_STANDBY (3 << 4) # define VC4_HD_M_SW_RST BIT(2) # define VC4_HD_M_ENABLE BIT(0) @@ -503,7 +519,12 @@ enum hvs_pixel_format { HVS_PIXEL_FORMAT_RGB888 = 5, HVS_PIXEL_FORMAT_RGBA6666 = 6, /* 32bpp */ - HVS_PIXEL_FORMAT_RGBA8888 = 7 + HVS_PIXEL_FORMAT_RGBA8888 = 7, + + HVS_PIXEL_FORMAT_YCBCR_YUV420_3PLANE = 8, + HVS_PIXEL_FORMAT_YCBCR_YUV420_2PLANE = 9, + HVS_PIXEL_FORMAT_YCBCR_YUV422_3PLANE = 10, + HVS_PIXEL_FORMAT_YCBCR_YUV422_2PLANE = 11, }; /* Note: the LSB is the rightmost character shown. Only valid for @@ -536,6 +557,21 @@ enum hvs_pixel_format { #define SCALER_CTL0_ORDER_MASK VC4_MASK(14, 13) #define SCALER_CTL0_ORDER_SHIFT 13 +#define SCALER_CTL0_SCL1_MASK VC4_MASK(10, 8) +#define SCALER_CTL0_SCL1_SHIFT 8 + +#define SCALER_CTL0_SCL0_MASK VC4_MASK(7, 5) +#define SCALER_CTL0_SCL0_SHIFT 5 + +#define SCALER_CTL0_SCL_H_PPF_V_PPF 0 +#define SCALER_CTL0_SCL_H_TPZ_V_PPF 1 +#define SCALER_CTL0_SCL_H_PPF_V_TPZ 2 +#define SCALER_CTL0_SCL_H_TPZ_V_TPZ 3 +#define SCALER_CTL0_SCL_H_PPF_V_NONE 4 +#define SCALER_CTL0_SCL_H_NONE_V_PPF 5 +#define SCALER_CTL0_SCL_H_NONE_V_TPZ 6 +#define SCALER_CTL0_SCL_H_TPZ_V_NONE 7 + /* Set to indicate no scaling. */ #define SCALER_CTL0_UNITY BIT(4) @@ -551,6 +587,12 @@ enum hvs_pixel_format { #define SCALER_POS0_START_X_MASK VC4_MASK(11, 0) #define SCALER_POS0_START_X_SHIFT 0 +#define SCALER_POS1_SCL_HEIGHT_MASK VC4_MASK(27, 16) +#define SCALER_POS1_SCL_HEIGHT_SHIFT 16 + +#define SCALER_POS1_SCL_WIDTH_MASK VC4_MASK(11, 0) +#define SCALER_POS1_SCL_WIDTH_SHIFT 0 + #define SCALER_POS2_ALPHA_MODE_MASK VC4_MASK(31, 30) #define SCALER_POS2_ALPHA_MODE_SHIFT 30 #define SCALER_POS2_ALPHA_MODE_PIPELINE 0 @@ -564,6 +606,80 @@ enum hvs_pixel_format { #define SCALER_POS2_WIDTH_MASK VC4_MASK(11, 0) #define SCALER_POS2_WIDTH_SHIFT 0 +/* Color Space Conversion words. Some values are S2.8 signed + * integers, except that the 2 integer bits map as {0x0: 0, 0x1: 1, + * 0x2: 2, 0x3: -1} + */ +/* bottom 8 bits of S2.8 contribution of Cr to Blue */ +#define SCALER_CSC0_COEF_CR_BLU_MASK VC4_MASK(31, 24) +#define SCALER_CSC0_COEF_CR_BLU_SHIFT 24 +/* Signed offset to apply to Y before CSC. (Y' = Y + YY_OFS) */ +#define SCALER_CSC0_COEF_YY_OFS_MASK VC4_MASK(23, 16) +#define SCALER_CSC0_COEF_YY_OFS_SHIFT 16 +/* Signed offset to apply to CB before CSC (Cb' = Cb - 128 + CB_OFS). */ +#define SCALER_CSC0_COEF_CB_OFS_MASK VC4_MASK(15, 8) +#define SCALER_CSC0_COEF_CB_OFS_SHIFT 8 +/* Signed offset to apply to CB before CSC (Cr' = Cr - 128 + CR_OFS). */ +#define SCALER_CSC0_COEF_CR_OFS_MASK VC4_MASK(7, 0) +#define SCALER_CSC0_COEF_CR_OFS_SHIFT 0 +#define SCALER_CSC0_ITR_R_601_5 0x00f00000 +#define SCALER_CSC0_ITR_R_709_3 0x00f00000 +#define SCALER_CSC0_JPEG_JFIF 0x00000000 + +/* S2.8 contribution of Cb to Green */ +#define SCALER_CSC1_COEF_CB_GRN_MASK VC4_MASK(31, 22) +#define SCALER_CSC1_COEF_CB_GRN_SHIFT 22 +/* S2.8 contribution of Cr to Green */ +#define SCALER_CSC1_COEF_CR_GRN_MASK VC4_MASK(21, 12) +#define SCALER_CSC1_COEF_CR_GRN_SHIFT 12 +/* S2.8 contribution of Y to all of RGB */ +#define SCALER_CSC1_COEF_YY_ALL_MASK VC4_MASK(11, 2) +#define SCALER_CSC1_COEF_YY_ALL_SHIFT 2 +/* top 2 bits of S2.8 contribution of Cr to Blue */ +#define SCALER_CSC1_COEF_CR_BLU_MASK VC4_MASK(1, 0) +#define SCALER_CSC1_COEF_CR_BLU_SHIFT 0 +#define SCALER_CSC1_ITR_R_601_5 0xe73304a8 +#define SCALER_CSC1_ITR_R_709_3 0xf2b784a8 +#define SCALER_CSC1_JPEG_JFIF 0xea34a400 + +/* S2.8 contribution of Cb to Red */ +#define SCALER_CSC2_COEF_CB_RED_MASK VC4_MASK(29, 20) +#define SCALER_CSC2_COEF_CB_RED_SHIFT 20 +/* S2.8 contribution of Cr to Red */ +#define SCALER_CSC2_COEF_CR_RED_MASK VC4_MASK(19, 10) +#define SCALER_CSC2_COEF_CR_RED_SHIFT 10 +/* S2.8 contribution of Cb to Blue */ +#define SCALER_CSC2_COEF_CB_BLU_MASK VC4_MASK(19, 10) +#define SCALER_CSC2_COEF_CB_BLU_SHIFT 10 +#define SCALER_CSC2_ITR_R_601_5 0x00066204 +#define SCALER_CSC2_ITR_R_709_3 0x00072a1c +#define SCALER_CSC2_JPEG_JFIF 0x000599c5 + +#define SCALER_TPZ0_VERT_RECALC BIT(31) +#define SCALER_TPZ0_SCALE_MASK VC4_MASK(28, 8) +#define SCALER_TPZ0_SCALE_SHIFT 8 +#define SCALER_TPZ0_IPHASE_MASK VC4_MASK(7, 0) +#define SCALER_TPZ0_IPHASE_SHIFT 0 +#define SCALER_TPZ1_RECIP_MASK VC4_MASK(15, 0) +#define SCALER_TPZ1_RECIP_SHIFT 0 + +/* Skips interpolating coefficients to 64 phases, so just 8 are used. + * Required for nearest neighbor. + */ +#define SCALER_PPF_NOINTERP BIT(31) +/* Replaes the highest valued coefficient with one that makes all 4 + * sum to unity. + */ +#define SCALER_PPF_AGC BIT(30) +#define SCALER_PPF_SCALE_MASK VC4_MASK(24, 8) +#define SCALER_PPF_SCALE_SHIFT 8 +#define SCALER_PPF_IPHASE_MASK VC4_MASK(6, 0) +#define SCALER_PPF_IPHASE_SHIFT 0 + +#define SCALER_PPF_KERNEL_OFFSET_MASK VC4_MASK(13, 0) +#define SCALER_PPF_KERNEL_OFFSET_SHIFT 0 +#define SCALER_PPF_KERNEL_UNCACHED BIT(31) + #define SCALER_SRC_PITCH_MASK VC4_MASK(15, 0) #define SCALER_SRC_PITCH_SHIFT 0 diff --git a/drivers/gpu/drm/vc4/vc4_v3d.c b/drivers/gpu/drm/vc4/vc4_v3d.c index 31de5d17b..e6d3c6028 100644 --- a/drivers/gpu/drm/vc4/vc4_v3d.c +++ b/drivers/gpu/drm/vc4/vc4_v3d.c @@ -268,6 +268,7 @@ static int vc4_v3d_dev_remove(struct platform_device *pdev) } static const struct of_device_id vc4_v3d_dt_match[] = { + { .compatible = "brcm,bcm2835-v3d" }, { .compatible = "brcm,vc4-v3d" }, {} }; -- cgit v1.2.3-54-g00ecf