From c182be37ed7cb04c344501b88b8fdb747016e6cf Mon Sep 17 00:00:00 2001 From: Kristian Høgsberg Date: Sat, 12 Sep 2009 04:33:34 +1000 Subject: drm: Add async event synchronization for drmWaitVblank MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This patch adds a new flag to the drmWaitVblank ioctl, which asks the drm to return immediately and notify userspace when the specified vblank sequence happens by sending an event back on the drm fd. The event mechanism works with the other flags supported by the ioctls, specifically, the vblank sequence can be specified relatively or absolutely, and works for primary and seconday crtc. The signal field of the vblank request is used to provide user data, which will be sent back to user space in the vblank event. Signed-off-by: Kristian Høgsberg Reviewed-by: Jesse Barnes Signed-off-by: Dave Airlie --- include/drm/drm.h | 33 +++++++++++++++++++++++++++++++-- include/drm/drmP.h | 26 ++++++++++++++++++++++++++ 2 files changed, 57 insertions(+), 2 deletions(-) (limited to 'include/drm') diff --git a/include/drm/drm.h b/include/drm/drm.h index 7cb50bdde46d..fa6d9155873d 100644 --- a/include/drm/drm.h +++ b/include/drm/drm.h @@ -454,6 +454,7 @@ struct drm_irq_busid { enum drm_vblank_seq_type { _DRM_VBLANK_ABSOLUTE = 0x0, /**< Wait for specific vblank sequence number */ _DRM_VBLANK_RELATIVE = 0x1, /**< Wait for given number of vblanks */ + _DRM_VBLANK_EVENT = 0x4000000, /**< Send event instead of blocking */ _DRM_VBLANK_FLIP = 0x8000000, /**< Scheduled buffer swap should flip */ _DRM_VBLANK_NEXTONMISS = 0x10000000, /**< If missed, wait for next vblank */ _DRM_VBLANK_SECONDARY = 0x20000000, /**< Secondary display controller */ @@ -461,8 +462,8 @@ enum drm_vblank_seq_type { }; #define _DRM_VBLANK_TYPES_MASK (_DRM_VBLANK_ABSOLUTE | _DRM_VBLANK_RELATIVE) -#define _DRM_VBLANK_FLAGS_MASK (_DRM_VBLANK_SIGNAL | _DRM_VBLANK_SECONDARY | \ - _DRM_VBLANK_NEXTONMISS) +#define _DRM_VBLANK_FLAGS_MASK (_DRM_VBLANK_EVENT | _DRM_VBLANK_SIGNAL | \ + _DRM_VBLANK_SECONDARY | _DRM_VBLANK_NEXTONMISS) struct drm_wait_vblank_request { enum drm_vblank_seq_type type; @@ -698,6 +699,34 @@ struct drm_gem_open { #define DRM_COMMAND_BASE 0x40 #define DRM_COMMAND_END 0xA0 +/** + * Header for events written back to userspace on the drm fd. The + * type defines the type of event, the length specifies the total + * length of the event (including the header), and user_data is + * typically a 64 bit value passed with the ioctl that triggered the + * event. A read on the drm fd will always only return complete + * events, that is, if for example the read buffer is 100 bytes, and + * there are two 64 byte events pending, only one will be returned. + * + * Event types 0 - 0x7fffffff are generic drm events, 0x80000000 and + * up are chipset specific. + */ +struct drm_event { + __u32 type; + __u32 length; +}; + +#define DRM_EVENT_VBLANK 0x01 + +struct drm_event_vblank { + struct drm_event base; + __u64 user_data; + __u32 tv_sec; + __u32 tv_usec; + __u32 sequence; + __u32 reserved; +}; + /* typedef area */ #ifndef __KERNEL__ typedef struct drm_clip_rect drm_clip_rect_t; diff --git a/include/drm/drmP.h b/include/drm/drmP.h index eeefb6369e19..fe52254df60c 100644 --- a/include/drm/drmP.h +++ b/include/drm/drmP.h @@ -426,6 +426,14 @@ struct drm_buf_entry { struct drm_freelist freelist; }; +/* Event queued up for userspace to read */ +struct drm_pending_event { + struct drm_event *event; + struct list_head link; + struct drm_file *file_priv; + void (*destroy)(struct drm_pending_event *event); +}; + /** File private data */ struct drm_file { int authenticated; @@ -449,6 +457,10 @@ struct drm_file { struct drm_master *master; /* master this node is currently associated with N.B. not always minor->master */ struct list_head fbs; + + wait_queue_head_t event_wait; + struct list_head event_list; + int event_space; }; /** Wait queue */ @@ -897,6 +909,12 @@ struct drm_minor { struct drm_mode_group mode_group; }; +struct drm_pending_vblank_event { + struct drm_pending_event base; + int pipe; + struct drm_event_vblank event; +}; + /** * DRM device structure. This structure represent a complete card that * may contain multiple heads. @@ -996,6 +1014,12 @@ struct drm_device { u32 max_vblank_count; /**< size of vblank counter register */ + /** + * List of events + */ + struct list_head vblank_event_list; + spinlock_t event_lock; + /*@} */ cycles_t ctx_start; cycles_t lck_start; @@ -1132,6 +1156,8 @@ extern int drm_lastclose(struct drm_device *dev); extern int drm_open(struct inode *inode, struct file *filp); extern int drm_stub_open(struct inode *inode, struct file *filp); extern int drm_fasync(int fd, struct file *filp, int on); +extern ssize_t drm_read(struct file *filp, char __user *buffer, + size_t count, loff_t *offset); extern int drm_release(struct inode *inode, struct file *filp); /* Mapping support (drm_vm.h) */ -- cgit v1.2.3 From 7a9c906094de8b3dc227de448019dbc386cd25d4 Mon Sep 17 00:00:00 2001 From: Daniel Vetter Date: Tue, 15 Sep 2009 22:57:31 +0200 Subject: drm: make drm_mode_object_find typesafe I've wasted half a day hunting a bug that could easily be spotted by gcc. Prevent this from reoccurring. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt --- drivers/gpu/drm/drm_crtc.c | 3 ++- include/drm/drm_crtc.h | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-) (limited to 'include/drm') diff --git a/drivers/gpu/drm/drm_crtc.c b/drivers/gpu/drm/drm_crtc.c index 5cae0b3eee9b..ee0cde1ecca5 100644 --- a/drivers/gpu/drm/drm_crtc.c +++ b/drivers/gpu/drm/drm_crtc.c @@ -247,7 +247,8 @@ static void drm_mode_object_put(struct drm_device *dev, mutex_unlock(&dev->mode_config.idr_mutex); } -void *drm_mode_object_find(struct drm_device *dev, uint32_t id, uint32_t type) +struct drm_mode_object *drm_mode_object_find(struct drm_device *dev, + uint32_t id, uint32_t type) { struct drm_mode_object *obj = NULL; diff --git a/include/drm/drm_crtc.h b/include/drm/drm_crtc.h index b69347b8904f..bfcc60d101db 100644 --- a/include/drm/drm_crtc.h +++ b/include/drm/drm_crtc.h @@ -711,7 +711,8 @@ extern void drm_mode_connector_detach_encoder(struct drm_connector *connector, struct drm_encoder *encoder); extern bool drm_mode_crtc_set_gamma_size(struct drm_crtc *crtc, int gamma_size); -extern void *drm_mode_object_find(struct drm_device *dev, uint32_t id, uint32_t type); +extern struct drm_mode_object *drm_mode_object_find(struct drm_device *dev, + uint32_t id, uint32_t type); /* IOCTLs */ extern int drm_mode_getresources(struct drm_device *dev, void *data, struct drm_file *file_priv); -- cgit v1.2.3 From 48764bf43f746113fc77877d7e80f2df23ca4cbb Mon Sep 17 00:00:00 2001 From: Daniel Vetter Date: Tue, 15 Sep 2009 22:57:32 +0200 Subject: drm/i915: add i915_lp_ring_sync helper This just waits until the hw passed the current ring position with cmd execution. This slightly changes the existing i915_wait_request function to make uninterruptible waiting possible - no point in returning to userspace while mucking around with the overlay, that piece of hw is just too fragile. Also replace a magic 0 with the symbolic constant (and kill the then superflous comment) while I was looking at the code. Signed-off-by: Daniel Vetter Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 49 ++++++++++++++++++++++++++++++++--------- include/drm/drm_os_linux.h | 2 +- 3 files changed, 41 insertions(+), 11 deletions(-) (limited to 'include/drm') diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index 95391191316a..e440f70e477a 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -803,6 +803,7 @@ void i915_gem_cleanup_ringbuffer(struct drm_device *dev); int i915_gem_do_init(struct drm_device *dev, unsigned long start, unsigned long end); int i915_gem_idle(struct drm_device *dev); +int i915_lp_ring_sync(struct drm_device *dev); int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf); int i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj, int write); diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index abfc27b0c2ea..7d1e9adf0f4c 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -1820,12 +1820,8 @@ i915_gem_retire_work_handler(struct work_struct *work) mutex_unlock(&dev->struct_mutex); } -/** - * Waits for a sequence number to be signaled, and cleans up the - * request and object lists appropriately for that event. - */ static int -i915_wait_request(struct drm_device *dev, uint32_t seqno) +i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible) { drm_i915_private_t *dev_priv = dev->dev_private; u32 ier; @@ -1852,10 +1848,15 @@ i915_wait_request(struct drm_device *dev, uint32_t seqno) dev_priv->mm.waiting_gem_seqno = seqno; i915_user_irq_get(dev); - ret = wait_event_interruptible(dev_priv->irq_queue, - i915_seqno_passed(i915_get_gem_seqno(dev), - seqno) || - atomic_read(&dev_priv->mm.wedged)); + if (interruptible) + ret = wait_event_interruptible(dev_priv->irq_queue, + i915_seqno_passed(i915_get_gem_seqno(dev), seqno) || + atomic_read(&dev_priv->mm.wedged)); + else + wait_event(dev_priv->irq_queue, + i915_seqno_passed(i915_get_gem_seqno(dev), seqno) || + atomic_read(&dev_priv->mm.wedged)); + i915_user_irq_put(dev); dev_priv->mm.waiting_gem_seqno = 0; @@ -1879,6 +1880,34 @@ i915_wait_request(struct drm_device *dev, uint32_t seqno) return ret; } +/** + * Waits for a sequence number to be signaled, and cleans up the + * request and object lists appropriately for that event. + */ +static int +i915_wait_request(struct drm_device *dev, uint32_t seqno) +{ + return i915_do_wait_request(dev, seqno, 1); +} + +/** + * Waits for the ring to finish up to the latest request. Usefull for waiting + * for flip events, e.g for the overlay support. */ +int i915_lp_ring_sync(struct drm_device *dev) +{ + uint32_t seqno; + int ret; + + seqno = i915_add_request(dev, NULL, 0); + + if (seqno == 0) + return -ENOMEM; + + ret = i915_do_wait_request(dev, seqno, 0); + BUG_ON(ret == -ERESTARTSYS); + return ret; +} + static void i915_gem_flush(struct drm_device *dev, uint32_t invalidate_domains, @@ -1947,7 +1976,7 @@ i915_gem_flush(struct drm_device *dev, #endif BEGIN_LP_RING(2); OUT_RING(cmd); - OUT_RING(0); /* noop */ + OUT_RING(MI_NOOP); ADVANCE_LP_RING(); } } diff --git a/include/drm/drm_os_linux.h b/include/drm/drm_os_linux.h index 26641e95e0a4..393369147a2d 100644 --- a/include/drm/drm_os_linux.h +++ b/include/drm/drm_os_linux.h @@ -123,5 +123,5 @@ do { \ remove_wait_queue(&(queue), &entry); \ } while (0) -#define DRM_WAKEUP( queue ) wake_up_interruptible( queue ) +#define DRM_WAKEUP( queue ) wake_up( queue ) #define DRM_INIT_WAITQUEUE( queue ) init_waitqueue_head( queue ) -- cgit v1.2.3 From 02e792fbaadb75dec8e476a05b610e49908fc6a4 Mon Sep 17 00:00:00 2001 From: Daniel Vetter Date: Tue, 15 Sep 2009 22:57:34 +0200 Subject: drm/i915: implement drmmode overlay support v4 This implements intel overlay support for kms via a device-specific ioctl. Thomas Hellstrom brought up the idea of a general ioctl (on dri-devel). We've reached the conclusion that such an infrastructure only makes sense when multiple kms overlay implementations exists, which atm don't (and it doesn't look like this is gonna change). Open issues: - Runs in sync with the gpu, i.e. unnecessary waiting. I've decided to wait on this because the hw tends to hang when changing something in this area. I left some dummy functions as infrastructure. - polyphase filtering uses a static table. - uses uninterruptible sleeps. Unfortunately the alternatives may unnecessarily wedged the hw if/when we timeout too early (and userspace only overloaded the batch buffers with stuff worth a few secs of gpu time). Changes since v1: - fix off-by-one misconception on my side. This fixes fullscreen playback. Changes since v2: - add underrun detection as spec'ed for i965. - flush caches properly, fixing visual corruptions. Changes since v4: - fix up cache flushing of overlay memory regs. - killed require_pipe_a logic - it hangs the chip. Tested-By: diego.abelenda@gmail.com (on a 865G) Signed-off-by: Daniel Vetter [anholt: Resolved against the MADVISE ioctl going in before this one] Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_dma.c | 7 + drivers/gpu/drm/i915/i915_drv.h | 5 + drivers/gpu/drm/i915/i915_reg.h | 5 + drivers/gpu/drm/i915/intel_display.c | 25 +- drivers/gpu/drm/i915/intel_drv.h | 28 + drivers/gpu/drm/i915/intel_overlay.c | 1293 ++++++++++++++++++++++++++++++++++ include/drm/i915_drm.h | 71 ++ 8 files changed, 1432 insertions(+), 3 deletions(-) create mode 100644 drivers/gpu/drm/i915/intel_overlay.c (limited to 'include/drm') diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile index fa7b9be096bc..87b21996cd6a 100644 --- a/drivers/gpu/drm/i915/Makefile +++ b/drivers/gpu/drm/i915/Makefile @@ -23,6 +23,7 @@ i915-y := i915_drv.o i915_dma.o i915_irq.o i915_mem.o \ intel_fb.o \ intel_tv.o \ intel_dvo.o \ + intel_overlay.o \ dvo_ch7xxx.o \ dvo_ch7017.o \ dvo_ivch.o \ diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c index e5b138be45fa..138be49259c3 100644 --- a/drivers/gpu/drm/i915/i915_dma.c +++ b/drivers/gpu/drm/i915/i915_dma.c @@ -807,6 +807,9 @@ static int i915_getparam(struct drm_device *dev, void *data, case I915_PARAM_NUM_FENCES_AVAIL: value = dev_priv->num_fence_regs - dev_priv->fence_reg_start; break; + case I915_PARAM_HAS_OVERLAY: + value = dev_priv->overlay ? 1 : 0; + break; default: DRM_DEBUG_DRIVER("Unknown parameter %d\n", param->param); @@ -1548,6 +1551,8 @@ int i915_driver_unload(struct drm_device *dev) mutex_unlock(&dev->struct_mutex); drm_mm_takedown(&dev_priv->vram); i915_gem_lastclose(dev); + + intel_cleanup_overlay(dev); } pci_dev_put(dev_priv->bridge_dev); @@ -1656,6 +1661,8 @@ struct drm_ioctl_desc i915_ioctls[] = { DRM_IOCTL_DEF(DRM_I915_GEM_GET_APERTURE, i915_gem_get_aperture_ioctl, 0), DRM_IOCTL_DEF(DRM_I915_GET_PIPE_FROM_CRTC_ID, intel_get_pipe_from_crtc_id, 0), DRM_IOCTL_DEF(DRM_I915_GEM_MADVISE, i915_gem_madvise_ioctl, 0), + DRM_IOCTL_DEF(DRM_I915_OVERLAY_PUT_IMAGE, intel_overlay_put_image, DRM_MASTER|DRM_CONTROL_ALLOW), + DRM_IOCTL_DEF(DRM_I915_OVERLAY_ATTRS, intel_overlay_attrs, DRM_MASTER|DRM_CONTROL_ALLOW), }; int i915_max_ioctl = DRM_ARRAY_SIZE(i915_ioctls); diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index 4e8b26161a74..ce03fd5b3f5b 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -170,6 +170,8 @@ struct drm_i915_display_funcs { /* clock gating init */ }; +struct intel_overlay; + typedef struct drm_i915_private { struct drm_device *dev; @@ -241,6 +243,9 @@ typedef struct drm_i915_private { struct intel_opregion opregion; + /* overlay */ + struct intel_overlay *overlay; + /* LVDS info */ int backlight_duty_cycle; /* restore backlight to this value */ bool panel_wants_dither; diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h index 51fd152f47f3..d1be1849580d 100644 --- a/drivers/gpu/drm/i915/i915_reg.h +++ b/drivers/gpu/drm/i915/i915_reg.h @@ -140,6 +140,7 @@ #define MI_NOOP MI_INSTR(0, 0) #define MI_USER_INTERRUPT MI_INSTR(0x02, 0) #define MI_WAIT_FOR_EVENT MI_INSTR(0x03, 0) +#define MI_WAIT_FOR_OVERLAY_FLIP (1<<16) #define MI_WAIT_FOR_PLANE_B_FLIP (1<<6) #define MI_WAIT_FOR_PLANE_A_FLIP (1<<2) #define MI_WAIT_FOR_PLANE_A_SCANLINES (1<<1) @@ -151,6 +152,10 @@ #define MI_END_SCENE (1 << 4) /* flush binner and incr scene count */ #define MI_BATCH_BUFFER_END MI_INSTR(0x0a, 0) #define MI_REPORT_HEAD MI_INSTR(0x07, 0) +#define MI_OVERLAY_FLIP MI_INSTR(0x11,0) +#define MI_OVERLAY_CONTINUE (0x0<<21) +#define MI_OVERLAY_ON (0x1<<21) +#define MI_OVERLAY_OFF (0x2<<21) #define MI_LOAD_SCAN_LINES_INCL MI_INSTR(0x12, 0) #define MI_STORE_DWORD_IMM MI_INSTR(0x20, 1) #define MI_MEM_VIRTUAL (1 << 22) /* 965+ only */ diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index 0be624a52e50..6f818fadcbe3 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -1781,6 +1781,22 @@ static void igdng_crtc_dpms(struct drm_crtc *crtc, int mode) } } +static void intel_crtc_dpms_overlay(struct intel_crtc *intel_crtc, bool enable) +{ + struct intel_overlay *overlay; + + if (!enable && intel_crtc->overlay) { + overlay = intel_crtc->overlay; + mutex_lock(&overlay->dev->struct_mutex); + intel_overlay_switch_off(overlay); + mutex_unlock(&overlay->dev->struct_mutex); + } + /* Let userspace switch the overlay on again. In most cases userspace + * has to recompute where to put it anyway. */ + + return; +} + static void i9xx_crtc_dpms(struct drm_crtc *crtc, int mode) { struct drm_device *dev = crtc->dev; @@ -1839,12 +1855,13 @@ static void i9xx_crtc_dpms(struct drm_crtc *crtc, int mode) intel_update_fbc(crtc, &crtc->mode); /* Give the overlay scaler a chance to enable if it's on this pipe */ - //intel_crtc_dpms_video(crtc, true); TODO + intel_crtc_dpms_overlay(intel_crtc, true); break; case DRM_MODE_DPMS_OFF: intel_update_watermarks(dev); + /* Give the overlay scaler a chance to disable if it's on this pipe */ - //intel_crtc_dpms_video(crtc, FALSE); TODO + intel_crtc_dpms_overlay(intel_crtc, false); if (dev_priv->cfb_plane == plane && dev_priv->display.disable_fbc) @@ -2039,7 +2056,7 @@ static int i830_get_display_clock_speed(struct drm_device *dev) * Return the pipe currently connected to the panel fitter, * or -1 if the panel fitter is not present or not in use */ -static int intel_panel_fitter_pipe (struct drm_device *dev) +int intel_panel_fitter_pipe (struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; u32 pfit_control; @@ -4458,6 +4475,8 @@ void intel_modeset_init(struct drm_device *dev) INIT_WORK(&dev_priv->idle_work, intel_idle_update); setup_timer(&dev_priv->idle_timer, intel_gpu_idle_timer, (unsigned long)dev); + + intel_setup_overlay(dev); } void intel_modeset_cleanup(struct drm_device *dev) diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h index ef61fe9507e2..c9b1b97ab792 100644 --- a/drivers/gpu/drm/i915/intel_drv.h +++ b/drivers/gpu/drm/i915/intel_drv.h @@ -110,6 +110,25 @@ struct intel_output { int clone_mask; }; +struct intel_crtc; +struct intel_overlay { + struct drm_device *dev; + struct intel_crtc *crtc; + struct drm_i915_gem_object *vid_bo; + struct drm_i915_gem_object *old_vid_bo; + int active; + int pfit_active; + u32 pfit_vscale_ratio; /* shifted-point number, (1<<12) == 1.0 */ + u32 color_key; + u32 brightness, contrast, saturation; + u32 old_xscale, old_yscale; + /* register access */ + u32 flip_addr; + struct drm_i915_gem_object *reg_bo; + void *virt_addr; + int hw_wedged; +}; + struct intel_crtc { struct drm_crtc base; enum pipe pipe; @@ -121,6 +140,7 @@ struct intel_crtc { bool busy; /* is scanout buffer being updated frequently? */ struct timer_list idle_timer; bool lowfreq_avail; + struct intel_overlay *overlay; }; #define to_intel_crtc(x) container_of(x, struct intel_crtc, base) @@ -148,6 +168,7 @@ intel_dp_set_m_n(struct drm_crtc *crtc, struct drm_display_mode *mode, extern void intel_edp_link_config (struct intel_output *, int *, int *); +extern int intel_panel_fitter_pipe (struct drm_device *dev); extern void intel_crtc_load_lut(struct drm_crtc *crtc); extern void intel_encoder_prepare (struct drm_encoder *encoder); extern void intel_encoder_commit (struct drm_encoder *encoder); @@ -183,4 +204,11 @@ extern int intel_framebuffer_create(struct drm_device *dev, struct drm_framebuffer **fb, struct drm_gem_object *obj); +extern void intel_setup_overlay(struct drm_device *dev); +extern void intel_cleanup_overlay(struct drm_device *dev); +extern int intel_overlay_switch_off(struct intel_overlay *overlay); +extern int intel_overlay_put_image(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int intel_overlay_attrs(struct drm_device *dev, void *data, + struct drm_file *file_priv); #endif /* __INTEL_DRV_H__ */ diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c new file mode 100644 index 000000000000..3f6f3a36929f --- /dev/null +++ b/drivers/gpu/drm/i915/intel_overlay.c @@ -0,0 +1,1293 @@ +/* + * Copyright © 2009 + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + * + * Authors: + * Daniel Vetter + * + * Derived from Xorg ddx, xf86-video-intel, src/i830_video.c + */ +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "i915_reg.h" +#include "intel_drv.h" + +/* Limits for overlay size. According to intel doc, the real limits are: + * Y width: 4095, UV width (planar): 2047, Y height: 2047, + * UV width (planar): * 1023. But the xorg thinks 2048 for height and width. Use + * the mininum of both. */ +#define IMAGE_MAX_WIDTH 2048 +#define IMAGE_MAX_HEIGHT 2046 /* 2 * 1023 */ +/* on 830 and 845 these large limits result in the card hanging */ +#define IMAGE_MAX_WIDTH_LEGACY 1024 +#define IMAGE_MAX_HEIGHT_LEGACY 1088 + +/* overlay register definitions */ +/* OCMD register */ +#define OCMD_TILED_SURFACE (0x1<<19) +#define OCMD_MIRROR_MASK (0x3<<17) +#define OCMD_MIRROR_MODE (0x3<<17) +#define OCMD_MIRROR_HORIZONTAL (0x1<<17) +#define OCMD_MIRROR_VERTICAL (0x2<<17) +#define OCMD_MIRROR_BOTH (0x3<<17) +#define OCMD_BYTEORDER_MASK (0x3<<14) /* zero for YUYV or FOURCC YUY2 */ +#define OCMD_UV_SWAP (0x1<<14) /* YVYU */ +#define OCMD_Y_SWAP (0x2<<14) /* UYVY or FOURCC UYVY */ +#define OCMD_Y_AND_UV_SWAP (0x3<<14) /* VYUY */ +#define OCMD_SOURCE_FORMAT_MASK (0xf<<10) +#define OCMD_RGB_888 (0x1<<10) /* not in i965 Intel docs */ +#define OCMD_RGB_555 (0x2<<10) /* not in i965 Intel docs */ +#define OCMD_RGB_565 (0x3<<10) /* not in i965 Intel docs */ +#define OCMD_YUV_422_PACKED (0x8<<10) +#define OCMD_YUV_411_PACKED (0x9<<10) /* not in i965 Intel docs */ +#define OCMD_YUV_420_PLANAR (0xc<<10) +#define OCMD_YUV_422_PLANAR (0xd<<10) +#define OCMD_YUV_410_PLANAR (0xe<<10) /* also 411 */ +#define OCMD_TVSYNCFLIP_PARITY (0x1<<9) +#define OCMD_TVSYNCFLIP_ENABLE (0x1<<7) +#define OCMD_BUF_TYPE_MASK (Ox1<<5) +#define OCMD_BUF_TYPE_FRAME (0x0<<5) +#define OCMD_BUF_TYPE_FIELD (0x1<<5) +#define OCMD_TEST_MODE (0x1<<4) +#define OCMD_BUFFER_SELECT (0x3<<2) +#define OCMD_BUFFER0 (0x0<<2) +#define OCMD_BUFFER1 (0x1<<2) +#define OCMD_FIELD_SELECT (0x1<<2) +#define OCMD_FIELD0 (0x0<<1) +#define OCMD_FIELD1 (0x1<<1) +#define OCMD_ENABLE (0x1<<0) + +/* OCONFIG register */ +#define OCONF_PIPE_MASK (0x1<<18) +#define OCONF_PIPE_A (0x0<<18) +#define OCONF_PIPE_B (0x1<<18) +#define OCONF_GAMMA2_ENABLE (0x1<<16) +#define OCONF_CSC_MODE_BT601 (0x0<<5) +#define OCONF_CSC_MODE_BT709 (0x1<<5) +#define OCONF_CSC_BYPASS (0x1<<4) +#define OCONF_CC_OUT_8BIT (0x1<<3) +#define OCONF_TEST_MODE (0x1<<2) +#define OCONF_THREE_LINE_BUFFER (0x1<<0) +#define OCONF_TWO_LINE_BUFFER (0x0<<0) + +/* DCLRKM (dst-key) register */ +#define DST_KEY_ENABLE (0x1<<31) +#define CLK_RGB24_MASK 0x0 +#define CLK_RGB16_MASK 0x070307 +#define CLK_RGB15_MASK 0x070707 +#define CLK_RGB8I_MASK 0xffffff + +#define RGB16_TO_COLORKEY(c) \ + (((c & 0xF800) << 8) | ((c & 0x07E0) << 5) | ((c & 0x001F) << 3)) +#define RGB15_TO_COLORKEY(c) \ + (((c & 0x7c00) << 9) | ((c & 0x03E0) << 6) | ((c & 0x001F) << 3)) + +/* overlay flip addr flag */ +#define OFC_UPDATE 0x1 + +/* polyphase filter coefficients */ +#define N_HORIZ_Y_TAPS 5 +#define N_VERT_Y_TAPS 3 +#define N_HORIZ_UV_TAPS 3 +#define N_VERT_UV_TAPS 3 +#define N_PHASES 17 +#define MAX_TAPS 5 + +/* memory bufferd overlay registers */ +struct overlay_registers { + u32 OBUF_0Y; + u32 OBUF_1Y; + u32 OBUF_0U; + u32 OBUF_0V; + u32 OBUF_1U; + u32 OBUF_1V; + u32 OSTRIDE; + u32 YRGB_VPH; + u32 UV_VPH; + u32 HORZ_PH; + u32 INIT_PHS; + u32 DWINPOS; + u32 DWINSZ; + u32 SWIDTH; + u32 SWIDTHSW; + u32 SHEIGHT; + u32 YRGBSCALE; + u32 UVSCALE; + u32 OCLRC0; + u32 OCLRC1; + u32 DCLRKV; + u32 DCLRKM; + u32 SCLRKVH; + u32 SCLRKVL; + u32 SCLRKEN; + u32 OCONFIG; + u32 OCMD; + u32 RESERVED1; /* 0x6C */ + u32 OSTART_0Y; + u32 OSTART_1Y; + u32 OSTART_0U; + u32 OSTART_0V; + u32 OSTART_1U; + u32 OSTART_1V; + u32 OTILEOFF_0Y; + u32 OTILEOFF_1Y; + u32 OTILEOFF_0U; + u32 OTILEOFF_0V; + u32 OTILEOFF_1U; + u32 OTILEOFF_1V; + u32 FASTHSCALE; /* 0xA0 */ + u32 UVSCALEV; /* 0xA4 */ + u32 RESERVEDC[(0x200 - 0xA8) / 4]; /* 0xA8 - 0x1FC */ + u16 Y_VCOEFS[N_VERT_Y_TAPS * N_PHASES]; /* 0x200 */ + u16 RESERVEDD[0x100 / 2 - N_VERT_Y_TAPS * N_PHASES]; + u16 Y_HCOEFS[N_HORIZ_Y_TAPS * N_PHASES]; /* 0x300 */ + u16 RESERVEDE[0x200 / 2 - N_HORIZ_Y_TAPS * N_PHASES]; + u16 UV_VCOEFS[N_VERT_UV_TAPS * N_PHASES]; /* 0x500 */ + u16 RESERVEDF[0x100 / 2 - N_VERT_UV_TAPS * N_PHASES]; + u16 UV_HCOEFS[N_HORIZ_UV_TAPS * N_PHASES]; /* 0x600 */ + u16 RESERVEDG[0x100 / 2 - N_HORIZ_UV_TAPS * N_PHASES]; +}; + +/* overlay flip addr flag */ +#define OFC_UPDATE 0x1 + +#define OVERLAY_NONPHYSICAL(dev) (IS_G33(dev) || IS_I965G(dev)) +#define OVERLAY_EXISTS(dev) (!IS_G4X(dev) && !IS_IGDNG(dev)) + + +static struct overlay_registers *intel_overlay_map_regs_atomic(struct intel_overlay *overlay) +{ + drm_i915_private_t *dev_priv = overlay->dev->dev_private; + struct overlay_registers *regs; + + /* no recursive mappings */ + BUG_ON(overlay->virt_addr); + + if (OVERLAY_NONPHYSICAL(overlay->dev)) { + regs = io_mapping_map_atomic_wc(dev_priv->mm.gtt_mapping, + overlay->reg_bo->gtt_offset); + + if (!regs) { + DRM_ERROR("failed to map overlay regs in GTT\n"); + return NULL; + } + } else + regs = overlay->reg_bo->phys_obj->handle->vaddr; + + return overlay->virt_addr = regs; +} + +static void intel_overlay_unmap_regs_atomic(struct intel_overlay *overlay) +{ + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + if (OVERLAY_NONPHYSICAL(overlay->dev)) + io_mapping_unmap_atomic(overlay->virt_addr); + + overlay->virt_addr = NULL; + + I915_READ(OVADD); /* flush wc cashes */ + + return; +} + +/* overlay needs to be disable in OCMD reg */ +static int intel_overlay_on(struct intel_overlay *overlay) +{ + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + RING_LOCALS; + + BUG_ON(overlay->active); + + BEGIN_LP_RING(6); + OUT_RING(MI_FLUSH); + OUT_RING(MI_NOOP); + OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_ON); + OUT_RING(overlay->flip_addr | OFC_UPDATE); + OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); + + ret = i915_lp_ring_sync(dev); + if (ret != 0) { + DRM_ERROR("intel overlay: ring sync failed, hw likely wedged\n"); + overlay->hw_wedged = 1; + return 0; + } + + overlay->active = 1; + + return 0; +} + +/* overlay needs to be enabled in OCMD reg */ +static void intel_overlay_continue(struct intel_overlay *overlay, + bool load_polyphase_filter) +{ + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 flip_addr = overlay->flip_addr; + u32 tmp; + int ret; + RING_LOCALS; + + BUG_ON(!overlay->active); + + if (load_polyphase_filter) + flip_addr |= OFC_UPDATE; + + /* check for underruns */ + tmp = I915_READ(DOVSTA); + if (tmp & (1 << 17)) + DRM_DEBUG("overlay underrun, DOVSTA: %x\n", tmp); + + BEGIN_LP_RING(6); + OUT_RING(MI_FLUSH); + OUT_RING(MI_NOOP); + OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE); + OUT_RING(flip_addr); + OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); + + /* run in lockstep with the hw for easier testing */ + ret = i915_lp_ring_sync(dev); + if (ret != 0) { + DRM_ERROR("intel overlay: ring sync failed, hw likely wedged\n"); + overlay->hw_wedged = 1; + } +} + +static int intel_overlay_wait_flip(struct intel_overlay *overlay) +{ + /* don't overcomplicate things for now with asynchronous operations + * see comment above */ + return 0; +} + +/* overlay needs to be disabled in OCMD reg */ +static int intel_overlay_off(struct intel_overlay *overlay) +{ + u32 flip_addr = overlay->flip_addr; + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + RING_LOCALS; + + BUG_ON(!overlay->active); + + /* According to intel docs the overlay hw may hang (when switching + * off) without loading the filter coeffs. It is however unclear whether + * this applies to the disabling of the overlay or to the switching off + * of the hw. Do it in both cases */ + flip_addr |= OFC_UPDATE; + + /* wait for overlay to go idle */ + BEGIN_LP_RING(6); + OUT_RING(MI_FLUSH); + OUT_RING(MI_NOOP); + OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE); + OUT_RING(flip_addr); + OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); + + ret = i915_lp_ring_sync(dev); + if (ret != 0) { + DRM_ERROR("intel overlay: ring sync failed, hw likely wedged\n"); + overlay->hw_wedged = 1; + return ret; + } + + /* turn overlay off */ + /* this is not done in userspace! + BEGIN_LP_RING(6); + OUT_RING(MI_FLUSH); + OUT_RING(MI_NOOP); + OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_OFF); + OUT_RING(flip_addr); + OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); + + ret = i915_lp_ring_sync(dev); + if (ret != 0) { + DRM_ERROR("intel overlay: ring sync failed, hw likely wedged\n"); + overlay->hw_wedged = 1; + return ret; + }*/ + + overlay->active = 0; + + return ret; +} + +/* wait for pending overlay flip and release old frame */ +static int intel_overlay_release_old_vid(struct intel_overlay *overlay) +{ + int ret; + struct drm_gem_object *obj; + + ret = intel_overlay_wait_flip(overlay); + if (ret != 0) + return ret; + + if (!overlay->old_vid_bo) + return 0; + + obj = overlay->old_vid_bo->obj; + i915_gem_object_unpin(obj); + drm_gem_object_unreference(obj); + overlay->old_vid_bo = NULL; + + return 0; +} + +struct put_image_params { + int format; + short dst_x; + short dst_y; + short dst_w; + short dst_h; + short src_w; + short src_scan_h; + short src_scan_w; + short src_h; + short stride_Y; + short stride_UV; + int offset_Y; + int offset_U; + int offset_V; +}; + +static int packed_depth_bytes(u32 format) +{ + switch (format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + return 4; + case I915_OVERLAY_YUV411: + /* return 6; not implemented */ + default: + return -EINVAL; + } +} + +static int packed_width_bytes(u32 format, short width) +{ + switch (format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + return width << 1; + default: + return -EINVAL; + } +} + +static int uv_hsubsampling(u32 format) +{ + switch (format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + case I915_OVERLAY_YUV420: + return 2; + case I915_OVERLAY_YUV411: + case I915_OVERLAY_YUV410: + return 4; + default: + return -EINVAL; + } +} + +static int uv_vsubsampling(u32 format) +{ + switch (format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV420: + case I915_OVERLAY_YUV410: + return 2; + case I915_OVERLAY_YUV422: + case I915_OVERLAY_YUV411: + return 1; + default: + return -EINVAL; + } +} + +static u32 calc_swidthsw(struct drm_device *dev, u32 offset, u32 width) +{ + u32 mask, shift, ret; + if (IS_I9XX(dev)) { + mask = 0x3f; + shift = 6; + } else { + mask = 0x1f; + shift = 5; + } + ret = ((offset + width + mask) >> shift) - (offset >> shift); + if (IS_I9XX(dev)) + ret <<= 1; + ret -=1; + return ret << 2; +} + +static const u16 y_static_hcoeffs[N_HORIZ_Y_TAPS * N_PHASES] = { + 0x3000, 0xb4a0, 0x1930, 0x1920, 0xb4a0, + 0x3000, 0xb500, 0x19d0, 0x1880, 0xb440, + 0x3000, 0xb540, 0x1a88, 0x2f80, 0xb3e0, + 0x3000, 0xb580, 0x1b30, 0x2e20, 0xb380, + 0x3000, 0xb5c0, 0x1bd8, 0x2cc0, 0xb320, + 0x3020, 0xb5e0, 0x1c60, 0x2b80, 0xb2c0, + 0x3020, 0xb5e0, 0x1cf8, 0x2a20, 0xb260, + 0x3020, 0xb5e0, 0x1d80, 0x28e0, 0xb200, + 0x3020, 0xb5c0, 0x1e08, 0x3f40, 0xb1c0, + 0x3020, 0xb580, 0x1e78, 0x3ce0, 0xb160, + 0x3040, 0xb520, 0x1ed8, 0x3aa0, 0xb120, + 0x3040, 0xb4a0, 0x1f30, 0x3880, 0xb0e0, + 0x3040, 0xb400, 0x1f78, 0x3680, 0xb0a0, + 0x3020, 0xb340, 0x1fb8, 0x34a0, 0xb060, + 0x3020, 0xb240, 0x1fe0, 0x32e0, 0xb040, + 0x3020, 0xb140, 0x1ff8, 0x3160, 0xb020, + 0xb000, 0x3000, 0x0800, 0x3000, 0xb000}; +static const u16 uv_static_hcoeffs[N_HORIZ_UV_TAPS * N_PHASES] = { + 0x3000, 0x1800, 0x1800, 0xb000, 0x18d0, 0x2e60, + 0xb000, 0x1990, 0x2ce0, 0xb020, 0x1a68, 0x2b40, + 0xb040, 0x1b20, 0x29e0, 0xb060, 0x1bd8, 0x2880, + 0xb080, 0x1c88, 0x3e60, 0xb0a0, 0x1d28, 0x3c00, + 0xb0c0, 0x1db8, 0x39e0, 0xb0e0, 0x1e40, 0x37e0, + 0xb100, 0x1eb8, 0x3620, 0xb100, 0x1f18, 0x34a0, + 0xb100, 0x1f68, 0x3360, 0xb0e0, 0x1fa8, 0x3240, + 0xb0c0, 0x1fe0, 0x3140, 0xb060, 0x1ff0, 0x30a0, + 0x3000, 0x0800, 0x3000}; + +static void update_polyphase_filter(struct overlay_registers *regs) +{ + memcpy(regs->Y_HCOEFS, y_static_hcoeffs, sizeof(y_static_hcoeffs)); + memcpy(regs->UV_HCOEFS, uv_static_hcoeffs, sizeof(uv_static_hcoeffs)); +} + +static bool update_scaling_factors(struct intel_overlay *overlay, + struct overlay_registers *regs, + struct put_image_params *params) +{ + /* fixed point with a 12 bit shift */ + u32 xscale, yscale, xscale_UV, yscale_UV; +#define FP_SHIFT 12 +#define FRACT_MASK 0xfff + bool scale_changed = false; + int uv_hscale = uv_hsubsampling(params->format); + int uv_vscale = uv_vsubsampling(params->format); + + if (params->dst_w > 1) + xscale = ((params->src_scan_w - 1) << FP_SHIFT) + /(params->dst_w); + else + xscale = 1 << FP_SHIFT; + + if (params->dst_h > 1) + yscale = ((params->src_scan_h - 1) << FP_SHIFT) + /(params->dst_h); + else + yscale = 1 << FP_SHIFT; + + /*if (params->format & I915_OVERLAY_YUV_PLANAR) {*/ + xscale_UV = xscale/uv_hscale; + yscale_UV = yscale/uv_vscale; + /* make the Y scale to UV scale ratio an exact multiply */ + xscale = xscale_UV * uv_hscale; + yscale = yscale_UV * uv_vscale; + /*} else { + xscale_UV = 0; + yscale_UV = 0; + }*/ + + if (xscale != overlay->old_xscale || yscale != overlay->old_yscale) + scale_changed = true; + overlay->old_xscale = xscale; + overlay->old_yscale = yscale; + + regs->YRGBSCALE = ((yscale & FRACT_MASK) << 20) + | ((xscale >> FP_SHIFT) << 16) + | ((xscale & FRACT_MASK) << 3); + regs->UVSCALE = ((yscale_UV & FRACT_MASK) << 20) + | ((xscale_UV >> FP_SHIFT) << 16) + | ((xscale_UV & FRACT_MASK) << 3); + regs->UVSCALEV = ((yscale >> FP_SHIFT) << 16) + | ((yscale_UV >> FP_SHIFT) << 0); + + if (scale_changed) + update_polyphase_filter(regs); + + return scale_changed; +} + +static void update_colorkey(struct intel_overlay *overlay, + struct overlay_registers *regs) +{ + u32 key = overlay->color_key; + switch (overlay->crtc->base.fb->bits_per_pixel) { + case 8: + regs->DCLRKV = 0; + regs->DCLRKM = CLK_RGB8I_MASK | DST_KEY_ENABLE; + case 16: + if (overlay->crtc->base.fb->depth == 15) { + regs->DCLRKV = RGB15_TO_COLORKEY(key); + regs->DCLRKM = CLK_RGB15_MASK | DST_KEY_ENABLE; + } else { + regs->DCLRKV = RGB16_TO_COLORKEY(key); + regs->DCLRKM = CLK_RGB16_MASK | DST_KEY_ENABLE; + } + case 24: + case 32: + regs->DCLRKV = key; + regs->DCLRKM = CLK_RGB24_MASK | DST_KEY_ENABLE; + } +} + +static u32 overlay_cmd_reg(struct put_image_params *params) +{ + u32 cmd = OCMD_ENABLE | OCMD_BUF_TYPE_FRAME | OCMD_BUFFER0; + + if (params->format & I915_OVERLAY_YUV_PLANAR) { + switch (params->format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + cmd |= OCMD_YUV_422_PLANAR; + break; + case I915_OVERLAY_YUV420: + cmd |= OCMD_YUV_420_PLANAR; + break; + case I915_OVERLAY_YUV411: + case I915_OVERLAY_YUV410: + cmd |= OCMD_YUV_410_PLANAR; + break; + } + } else { /* YUV packed */ + switch (params->format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + cmd |= OCMD_YUV_422_PACKED; + break; + case I915_OVERLAY_YUV411: + cmd |= OCMD_YUV_411_PACKED; + break; + } + + switch (params->format & I915_OVERLAY_SWAP_MASK) { + case I915_OVERLAY_NO_SWAP: + break; + case I915_OVERLAY_UV_SWAP: + cmd |= OCMD_UV_SWAP; + break; + case I915_OVERLAY_Y_SWAP: + cmd |= OCMD_Y_SWAP; + break; + case I915_OVERLAY_Y_AND_UV_SWAP: + cmd |= OCMD_Y_AND_UV_SWAP; + break; + } + } + + return cmd; +} + +int intel_overlay_do_put_image(struct intel_overlay *overlay, + struct drm_gem_object *new_bo, + struct put_image_params *params) +{ + int ret, tmp_width; + struct overlay_registers *regs; + bool scale_changed = false; + struct drm_i915_gem_object *bo_priv = new_bo->driver_private; + struct drm_device *dev = overlay->dev; + + BUG_ON(!mutex_is_locked(&dev->struct_mutex)); + BUG_ON(!mutex_is_locked(&dev->mode_config.mutex)); + BUG_ON(!overlay); + + if (overlay->hw_wedged) + return -EBUSY; + + ret = intel_overlay_release_old_vid(overlay); + if (ret != 0) + return ret; + + ret = i915_gem_object_pin(new_bo, PAGE_SIZE); + if (ret != 0) + return ret; + + ret = i915_gem_object_set_to_gtt_domain(new_bo, 0); + if (ret != 0) + goto out_unpin; + + if (!overlay->active) { + regs = intel_overlay_map_regs_atomic(overlay); + if (!regs) { + ret = -ENOMEM; + goto out_unpin; + } + regs->OCONFIG = OCONF_CC_OUT_8BIT; + if (IS_I965GM(overlay->dev)) + regs->OCONFIG |= OCONF_CSC_MODE_BT709; + regs->OCONFIG |= overlay->crtc->pipe == 0 ? + OCONF_PIPE_A : OCONF_PIPE_B; + intel_overlay_unmap_regs_atomic(overlay); + + ret = intel_overlay_on(overlay); + if (ret != 0) + goto out_unpin; + } + + regs = intel_overlay_map_regs_atomic(overlay); + if (!regs) { + ret = -ENOMEM; + goto out_unpin; + } + + regs->DWINPOS = (params->dst_y << 16) | params->dst_x; + regs->DWINSZ = (params->dst_h << 16) | params->dst_w; + + if (params->format & I915_OVERLAY_YUV_PACKED) + tmp_width = packed_width_bytes(params->format, params->src_w); + else + tmp_width = params->src_w; + + regs->SWIDTH = params->src_w; + regs->SWIDTHSW = calc_swidthsw(overlay->dev, + params->offset_Y, tmp_width); + regs->SHEIGHT = params->src_h; + regs->OBUF_0Y = bo_priv->gtt_offset + params-> offset_Y; + regs->OSTRIDE = params->stride_Y; + + if (params->format & I915_OVERLAY_YUV_PLANAR) { + int uv_hscale = uv_hsubsampling(params->format); + int uv_vscale = uv_vsubsampling(params->format); + u32 tmp_U, tmp_V; + regs->SWIDTH |= (params->src_w/uv_hscale) << 16; + tmp_U = calc_swidthsw(overlay->dev, params->offset_U, + params->src_w/uv_hscale); + tmp_V = calc_swidthsw(overlay->dev, params->offset_V, + params->src_w/uv_hscale); + regs->SWIDTHSW |= max_t(u32, tmp_U, tmp_V) << 16; + regs->SHEIGHT |= (params->src_h/uv_vscale) << 16; + regs->OBUF_0U = bo_priv->gtt_offset + params->offset_U; + regs->OBUF_0V = bo_priv->gtt_offset + params->offset_V; + regs->OSTRIDE |= params->stride_UV << 16; + } + + scale_changed = update_scaling_factors(overlay, regs, params); + + update_colorkey(overlay, regs); + + regs->OCMD = overlay_cmd_reg(params); + + intel_overlay_unmap_regs_atomic(overlay); + + intel_overlay_continue(overlay, scale_changed); + + overlay->old_vid_bo = overlay->vid_bo; + overlay->vid_bo = new_bo->driver_private; + + return 0; + +out_unpin: + i915_gem_object_unpin(new_bo); + return ret; +} + +int intel_overlay_switch_off(struct intel_overlay *overlay) +{ + int ret; + struct overlay_registers *regs; + struct drm_gem_object *obj; + struct drm_device *dev = overlay->dev; + + BUG_ON(!mutex_is_locked(&dev->struct_mutex)); + BUG_ON(!mutex_is_locked(&dev->mode_config.mutex)); + + if (!overlay->active) + return 0; + + if (overlay->hw_wedged) + return -EBUSY; + + ret = intel_overlay_release_old_vid(overlay); + if (ret != 0) + return ret; + + regs = intel_overlay_map_regs_atomic(overlay); + regs->OCMD = 0; + intel_overlay_unmap_regs_atomic(overlay); + + ret = intel_overlay_off(overlay); + /* never have the overlay hw on without showing a frame */ + BUG_ON(!overlay->vid_bo); + obj = overlay->vid_bo->obj; + + i915_gem_object_unpin(obj); + drm_gem_object_unreference(obj); + overlay->vid_bo = NULL; + + overlay->crtc->overlay = NULL; + overlay->crtc = NULL; + + return 0; +} + +static int check_overlay_possible_on_crtc(struct intel_overlay *overlay, + struct intel_crtc *crtc) +{ + drm_i915_private_t *dev_priv = overlay->dev->dev_private; + u32 pipeconf; + int pipeconf_reg = (crtc->pipe == 0) ? PIPEACONF : PIPEBCONF; + + if (!crtc->base.enabled || crtc->dpms_mode != DRM_MODE_DPMS_ON) + return -EINVAL; + + pipeconf = I915_READ(pipeconf_reg); + + /* can't use the overlay with double wide pipe */ + if (!IS_I965G(overlay->dev) && pipeconf & PIPEACONF_DOUBLE_WIDE) + return -EINVAL; + + return 0; +} + +static void update_pfit_vscale_ratio(struct intel_overlay *overlay) +{ + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 ratio; + u32 pfit_control = I915_READ(PFIT_CONTROL); + + /* XXX: This is not the same logic as in the xorg driver, but more in + * line with the intel documentation for the i965 */ + if (!IS_I965G(dev) && (pfit_control & VERT_AUTO_SCALE)) { + ratio = I915_READ(PFIT_AUTO_RATIOS) >> PFIT_VERT_SCALE_SHIFT; + } else { /* on i965 use the PGM reg to read out the autoscaler values */ + ratio = I915_READ(PFIT_PGM_RATIOS); + if (IS_I965G(dev)) + ratio >>= PFIT_VERT_SCALE_SHIFT_965; + else + ratio >>= PFIT_VERT_SCALE_SHIFT; + } + + overlay->pfit_vscale_ratio = ratio; +} + +static int check_overlay_dst(struct intel_overlay *overlay, + struct drm_intel_overlay_put_image *rec) +{ + struct drm_display_mode *mode = &overlay->crtc->base.mode; + + if ((rec->dst_x < mode->crtc_hdisplay) + && (rec->dst_x + rec->dst_width + <= mode->crtc_hdisplay) + && (rec->dst_y < mode->crtc_vdisplay) + && (rec->dst_y + rec->dst_height + <= mode->crtc_vdisplay)) + return 0; + else + return -EINVAL; +} + +static int check_overlay_scaling(struct put_image_params *rec) +{ + u32 tmp; + + /* downscaling limit is 8.0 */ + tmp = ((rec->src_scan_h << 16) / rec->dst_h) >> 16; + if (tmp > 7) + return -EINVAL; + tmp = ((rec->src_scan_w << 16) / rec->dst_w) >> 16; + if (tmp > 7) + return -EINVAL; + + return 0; +} + +static int check_overlay_src(struct drm_device *dev, + struct drm_intel_overlay_put_image *rec, + struct drm_gem_object *new_bo) +{ + u32 stride_mask; + int depth; + int uv_hscale = uv_hsubsampling(rec->flags); + int uv_vscale = uv_vsubsampling(rec->flags); + size_t tmp; + + /* check src dimensions */ + if (IS_845G(dev) || IS_I830(dev)) { + if (rec->src_height > IMAGE_MAX_HEIGHT_LEGACY + || rec->src_width > IMAGE_MAX_WIDTH_LEGACY) + return -EINVAL; + } else { + if (rec->src_height > IMAGE_MAX_HEIGHT + || rec->src_width > IMAGE_MAX_WIDTH) + return -EINVAL; + } + /* better safe than sorry, use 4 as the maximal subsampling ratio */ + if (rec->src_height < N_VERT_Y_TAPS*4 + || rec->src_width < N_HORIZ_Y_TAPS*4) + return -EINVAL; + + /* check alingment constrains */ + switch (rec->flags & I915_OVERLAY_TYPE_MASK) { + case I915_OVERLAY_RGB: + /* not implemented */ + return -EINVAL; + case I915_OVERLAY_YUV_PACKED: + depth = packed_depth_bytes(rec->flags); + if (uv_vscale != 1) + return -EINVAL; + if (depth < 0) + return depth; + /* ignore UV planes */ + rec->stride_UV = 0; + rec->offset_U = 0; + rec->offset_V = 0; + /* check pixel alignment */ + if (rec->offset_Y % depth) + return -EINVAL; + break; + case I915_OVERLAY_YUV_PLANAR: + if (uv_vscale < 0 || uv_hscale < 0) + return -EINVAL; + /* no offset restrictions for planar formats */ + break; + default: + return -EINVAL; + } + + if (rec->src_width % uv_hscale) + return -EINVAL; + + /* stride checking */ + stride_mask = 63; + + if (rec->stride_Y & stride_mask || rec->stride_UV & stride_mask) + return -EINVAL; + if (IS_I965G(dev) && rec->stride_Y < 512) + return -EINVAL; + + tmp = (rec->flags & I915_OVERLAY_TYPE_MASK) == I915_OVERLAY_YUV_PLANAR ? + 4 : 8; + if (rec->stride_Y > tmp*1024 || rec->stride_UV > 2*1024) + return -EINVAL; + + /* check buffer dimensions */ + switch (rec->flags & I915_OVERLAY_TYPE_MASK) { + case I915_OVERLAY_RGB: + case I915_OVERLAY_YUV_PACKED: + /* always 4 Y values per depth pixels */ + if (packed_width_bytes(rec->flags, rec->src_width) + > rec->stride_Y) + return -EINVAL; + + tmp = rec->stride_Y*rec->src_height; + if (rec->offset_Y + tmp > new_bo->size) + return -EINVAL; + break; + case I915_OVERLAY_YUV_PLANAR: + if (rec->src_width > rec->stride_Y) + return -EINVAL; + if (rec->src_width/uv_hscale > rec->stride_UV) + return -EINVAL; + + tmp = rec->stride_Y*rec->src_height; + if (rec->offset_Y + tmp > new_bo->size) + return -EINVAL; + tmp = rec->stride_UV*rec->src_height; + tmp /= uv_vscale; + if (rec->offset_U + tmp > new_bo->size + || rec->offset_V + tmp > new_bo->size) + return -EINVAL; + break; + } + + return 0; +} + +int intel_overlay_put_image(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_intel_overlay_put_image *put_image_rec = data; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_overlay *overlay; + struct drm_mode_object *drmmode_obj; + struct intel_crtc *crtc; + struct drm_gem_object *new_bo; + struct put_image_params *params; + int ret; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + overlay = dev_priv->overlay; + if (!overlay) { + DRM_DEBUG("userspace bug: no overlay\n"); + return -ENODEV; + } + + if (!(put_image_rec->flags & I915_OVERLAY_ENABLE)) { + mutex_lock(&dev->mode_config.mutex); + mutex_lock(&dev->struct_mutex); + + ret = intel_overlay_switch_off(overlay); + + mutex_unlock(&dev->struct_mutex); + mutex_unlock(&dev->mode_config.mutex); + + return ret; + } + + params = kmalloc(sizeof(struct put_image_params), GFP_KERNEL); + if (!params) + return -ENOMEM; + + drmmode_obj = drm_mode_object_find(dev, put_image_rec->crtc_id, + DRM_MODE_OBJECT_CRTC); + if (!drmmode_obj) + return -ENOENT; + crtc = to_intel_crtc(obj_to_crtc(drmmode_obj)); + + new_bo = drm_gem_object_lookup(dev, file_priv, + put_image_rec->bo_handle); + if (!new_bo) + return -ENOENT; + + mutex_lock(&dev->mode_config.mutex); + mutex_lock(&dev->struct_mutex); + + if (overlay->crtc != crtc) { + struct drm_display_mode *mode = &crtc->base.mode; + ret = intel_overlay_switch_off(overlay); + if (ret != 0) + goto out_unlock; + + ret = check_overlay_possible_on_crtc(overlay, crtc); + if (ret != 0) + goto out_unlock; + + overlay->crtc = crtc; + crtc->overlay = overlay; + + if (intel_panel_fitter_pipe(dev) == crtc->pipe + /* and line to wide, i.e. one-line-mode */ + && mode->hdisplay > 1024) { + overlay->pfit_active = 1; + update_pfit_vscale_ratio(overlay); + } else + overlay->pfit_active = 0; + } + + ret = check_overlay_dst(overlay, put_image_rec); + if (ret != 0) + goto out_unlock; + + if (overlay->pfit_active) { + params->dst_y = ((((u32)put_image_rec->dst_y) << 12) / + overlay->pfit_vscale_ratio); + /* shifting right rounds downwards, so add 1 */ + params->dst_h = ((((u32)put_image_rec->dst_height) << 12) / + overlay->pfit_vscale_ratio) + 1; + } else { + params->dst_y = put_image_rec->dst_y; + params->dst_h = put_image_rec->dst_height; + } + params->dst_x = put_image_rec->dst_x; + params->dst_w = put_image_rec->dst_width; + + params->src_w = put_image_rec->src_width; + params->src_h = put_image_rec->src_height; + params->src_scan_w = put_image_rec->src_scan_width; + params->src_scan_h = put_image_rec->src_scan_height; + if (params->src_scan_h > params->src_h + || params->src_scan_w > params->src_w) { + ret = -EINVAL; + goto out_unlock; + } + + ret = check_overlay_src(dev, put_image_rec, new_bo); + if (ret != 0) + goto out_unlock; + params->format = put_image_rec->flags & ~I915_OVERLAY_FLAGS_MASK; + params->stride_Y = put_image_rec->stride_Y; + params->stride_UV = put_image_rec->stride_UV; + params->offset_Y = put_image_rec->offset_Y; + params->offset_U = put_image_rec->offset_U; + params->offset_V = put_image_rec->offset_V; + + /* Check scaling after src size to prevent a divide-by-zero. */ + ret = check_overlay_scaling(params); + if (ret != 0) + goto out_unlock; + + ret = intel_overlay_do_put_image(overlay, new_bo, params); + if (ret != 0) + goto out_unlock; + + mutex_unlock(&dev->struct_mutex); + mutex_unlock(&dev->mode_config.mutex); + + kfree(params); + + return 0; + +out_unlock: + mutex_unlock(&dev->struct_mutex); + mutex_unlock(&dev->mode_config.mutex); + drm_gem_object_unreference(new_bo); + kfree(params); + + return ret; +} + +static void update_reg_attrs(struct intel_overlay *overlay, + struct overlay_registers *regs) +{ + regs->OCLRC0 = (overlay->contrast << 18) | (overlay->brightness & 0xff); + regs->OCLRC1 = overlay->saturation; +} + +static bool check_gamma_bounds(u32 gamma1, u32 gamma2) +{ + int i; + + if (gamma1 & 0xff000000 || gamma2 & 0xff000000) + return false; + + for (i = 0; i < 3; i++) { + if (((gamma1 >> i * 8) & 0xff) >= ((gamma2 >> i*8) & 0xff)) + return false; + } + + return true; +} + +static bool check_gamma5_errata(u32 gamma5) +{ + int i; + + for (i = 0; i < 3; i++) { + if (((gamma5 >> i*8) & 0xff) == 0x80) + return false; + } + + return true; +} + +static int check_gamma(struct drm_intel_overlay_attrs *attrs) +{ + if (!check_gamma_bounds(0, attrs->gamma0) + || !check_gamma_bounds(attrs->gamma0, attrs->gamma1) + || !check_gamma_bounds(attrs->gamma1, attrs->gamma2) + || !check_gamma_bounds(attrs->gamma2, attrs->gamma3) + || !check_gamma_bounds(attrs->gamma3, attrs->gamma4) + || !check_gamma_bounds(attrs->gamma4, attrs->gamma5) + || !check_gamma_bounds(attrs->gamma5, 0x00ffffff)) + return -EINVAL; + if (!check_gamma5_errata(attrs->gamma5)) + return -EINVAL; + return 0; +} + +int intel_overlay_attrs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_intel_overlay_attrs *attrs = data; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_overlay *overlay; + struct overlay_registers *regs; + int ret; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + overlay = dev_priv->overlay; + if (!overlay) { + DRM_DEBUG("userspace bug: no overlay\n"); + return -ENODEV; + } + + mutex_lock(&dev->mode_config.mutex); + mutex_lock(&dev->struct_mutex); + + if (!(attrs->flags & I915_OVERLAY_UPDATE_ATTRS)) { + attrs->color_key = overlay->color_key; + attrs->brightness = overlay->brightness; + attrs->contrast = overlay->contrast; + attrs->saturation = overlay->saturation; + + if (IS_I9XX(dev)) { + attrs->gamma0 = I915_READ(OGAMC0); + attrs->gamma1 = I915_READ(OGAMC1); + attrs->gamma2 = I915_READ(OGAMC2); + attrs->gamma3 = I915_READ(OGAMC3); + attrs->gamma4 = I915_READ(OGAMC4); + attrs->gamma5 = I915_READ(OGAMC5); + } + ret = 0; + } else { + overlay->color_key = attrs->color_key; + if (attrs->brightness >= -128 && attrs->brightness <= 127) { + overlay->brightness = attrs->brightness; + } else { + ret = -EINVAL; + goto out_unlock; + } + if (attrs->contrast <= 255) { + overlay->contrast = attrs->contrast; + } else { + ret = -EINVAL; + goto out_unlock; + } + if (attrs->saturation <= 1023) { + overlay->saturation = attrs->saturation; + } else { + ret = -EINVAL; + goto out_unlock; + } + + regs = intel_overlay_map_regs_atomic(overlay); + if (!regs) { + ret = -ENOMEM; + goto out_unlock; + } + + update_reg_attrs(overlay, regs); + + intel_overlay_unmap_regs_atomic(overlay); + + if (attrs->flags & I915_OVERLAY_UPDATE_GAMMA) { + if (!IS_I9XX(dev)) { + ret = -EINVAL; + goto out_unlock; + } + + if (overlay->active) { + ret = -EBUSY; + goto out_unlock; + } + + ret = check_gamma(attrs); + if (ret != 0) + goto out_unlock; + + I915_WRITE(OGAMC0, attrs->gamma0); + I915_WRITE(OGAMC1, attrs->gamma1); + I915_WRITE(OGAMC2, attrs->gamma2); + I915_WRITE(OGAMC3, attrs->gamma3); + I915_WRITE(OGAMC4, attrs->gamma4); + I915_WRITE(OGAMC5, attrs->gamma5); + } + ret = 0; + } + +out_unlock: + mutex_unlock(&dev->struct_mutex); + mutex_unlock(&dev->mode_config.mutex); + + return ret; +} + +void intel_setup_overlay(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_overlay *overlay; + struct drm_gem_object *reg_bo; + struct overlay_registers *regs; + int ret; + + if (!OVERLAY_EXISTS(dev)) + return; + + overlay = kzalloc(sizeof(struct intel_overlay), GFP_KERNEL); + if (!overlay) + return; + overlay->dev = dev; + + reg_bo = drm_gem_object_alloc(dev, PAGE_SIZE); + if (!reg_bo) + goto out_free; + overlay->reg_bo = reg_bo->driver_private; + + if (OVERLAY_NONPHYSICAL(dev)) { + ret = i915_gem_object_pin(reg_bo, PAGE_SIZE); + if (ret) { + DRM_ERROR("failed to pin overlay register bo\n"); + goto out_free_bo; + } + overlay->flip_addr = overlay->reg_bo->gtt_offset; + } else { + ret = i915_gem_attach_phys_object(dev, reg_bo, + I915_GEM_PHYS_OVERLAY_REGS); + if (ret) { + DRM_ERROR("failed to attach phys overlay regs\n"); + goto out_free_bo; + } + overlay->flip_addr = overlay->reg_bo->phys_obj->handle->busaddr; + } + + /* init all values */ + overlay->color_key = 0x0101fe; + overlay->brightness = -19; + overlay->contrast = 75; + overlay->saturation = 146; + + regs = intel_overlay_map_regs_atomic(overlay); + if (!regs) + goto out_free_bo; + + memset(regs, 0, sizeof(struct overlay_registers)); + update_polyphase_filter(regs); + + update_reg_attrs(overlay, regs); + + intel_overlay_unmap_regs_atomic(overlay); + + dev_priv->overlay = overlay; + DRM_INFO("initialized overlay support\n"); + return; + +out_free_bo: + drm_gem_object_unreference(reg_bo); +out_free: + kfree(overlay); + return; +} + +void intel_cleanup_overlay(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + + if (dev_priv->overlay) { + /* The bo's should be free'd by the generic code already. + * Furthermore modesetting teardown happens beforehand so the + * hardware should be off already */ + BUG_ON(dev_priv->overlay->active); + + kfree(dev_priv->overlay); + } +} diff --git a/include/drm/i915_drm.h b/include/drm/i915_drm.h index 7e0cb1da92e6..c900499f2f63 100644 --- a/include/drm/i915_drm.h +++ b/include/drm/i915_drm.h @@ -186,6 +186,8 @@ typedef struct _drm_i915_sarea { #define DRM_I915_GEM_MMAP_GTT 0x24 #define DRM_I915_GET_PIPE_FROM_CRTC_ID 0x25 #define DRM_I915_GEM_MADVISE 0x26 +#define DRM_I915_OVERLAY_PUT_IMAGE 0x27 +#define DRM_I915_OVERLAY_ATTRS 0x28 #define DRM_IOCTL_I915_INIT DRM_IOW( DRM_COMMAND_BASE + DRM_I915_INIT, drm_i915_init_t) #define DRM_IOCTL_I915_FLUSH DRM_IO ( DRM_COMMAND_BASE + DRM_I915_FLUSH) @@ -223,6 +225,8 @@ typedef struct _drm_i915_sarea { #define DRM_IOCTL_I915_GEM_GET_APERTURE DRM_IOR (DRM_COMMAND_BASE + DRM_I915_GEM_GET_APERTURE, struct drm_i915_gem_get_aperture) #define DRM_IOCTL_I915_GET_PIPE_FROM_CRTC_ID DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_GET_PIPE_FROM_CRTC_ID, struct drm_intel_get_pipe_from_crtc_id) #define DRM_IOCTL_I915_GEM_MADVISE DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_GEM_MADVISE, struct drm_i915_gem_madvise) +#define DRM_IOCTL_I915_OVERLAY_PUT_IMAGE DRM_IOW(DRM_COMMAND_BASE + DRM_IOCTL_I915_OVERLAY_ATTRS, struct drm_intel_overlay_put_image) +#define DRM_IOCTL_I915_OVERLAY_ATTRS DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_OVERLAY_ATTRS, struct drm_intel_overlay_attrs) /* Allow drivers to submit batchbuffers directly to hardware, relying * on the security mechanisms provided by hardware. @@ -266,6 +270,7 @@ typedef struct drm_i915_irq_wait { #define I915_PARAM_CHIPSET_ID 4 #define I915_PARAM_HAS_GEM 5 #define I915_PARAM_NUM_FENCES_AVAIL 6 +#define I915_PARAM_HAS_OVERLAY 7 typedef struct drm_i915_getparam { int param; @@ -686,4 +691,70 @@ struct drm_i915_gem_madvise { __u32 retained; }; +/* flags */ +#define I915_OVERLAY_TYPE_MASK 0xff +#define I915_OVERLAY_YUV_PLANAR 0x01 +#define I915_OVERLAY_YUV_PACKED 0x02 +#define I915_OVERLAY_RGB 0x03 + +#define I915_OVERLAY_DEPTH_MASK 0xff00 +#define I915_OVERLAY_RGB24 0x1000 +#define I915_OVERLAY_RGB16 0x2000 +#define I915_OVERLAY_RGB15 0x3000 +#define I915_OVERLAY_YUV422 0x0100 +#define I915_OVERLAY_YUV411 0x0200 +#define I915_OVERLAY_YUV420 0x0300 +#define I915_OVERLAY_YUV410 0x0400 + +#define I915_OVERLAY_SWAP_MASK 0xff0000 +#define I915_OVERLAY_NO_SWAP 0x000000 +#define I915_OVERLAY_UV_SWAP 0x010000 +#define I915_OVERLAY_Y_SWAP 0x020000 +#define I915_OVERLAY_Y_AND_UV_SWAP 0x030000 + +#define I915_OVERLAY_FLAGS_MASK 0xff000000 +#define I915_OVERLAY_ENABLE 0x01000000 + +struct drm_intel_overlay_put_image { + /* various flags and src format description */ + __u32 flags; + /* source picture description */ + __u32 bo_handle; + /* stride values and offsets are in bytes, buffer relative */ + __u16 stride_Y; /* stride for packed formats */ + __u16 stride_UV; + __u32 offset_Y; /* offset for packet formats */ + __u32 offset_U; + __u32 offset_V; + /* in pixels */ + __u16 src_width; + __u16 src_height; + /* to compensate the scaling factors for partially covered surfaces */ + __u16 src_scan_width; + __u16 src_scan_height; + /* output crtc description */ + __u32 crtc_id; + __u16 dst_x; + __u16 dst_y; + __u16 dst_width; + __u16 dst_height; +}; + +/* flags */ +#define I915_OVERLAY_UPDATE_ATTRS (1<<0) +#define I915_OVERLAY_UPDATE_GAMMA (1<<1) +struct drm_intel_overlay_attrs { + __u32 flags; + __u32 color_key; + __s32 brightness; + __u32 contrast; + __u32 saturation; + __u32 gamma0; + __u32 gamma1; + __u32 gamma2; + __u32 gamma3; + __u32 gamma4; + __u32 gamma5; +}; + #endif /* _I915_DRM_H_ */ -- cgit v1.2.3 From c8e0f93a381d1d76135e567f13a4418fce66fd95 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Sun, 22 Nov 2009 03:49:37 +0100 Subject: drm/i915: Replace a calloc followed by copying data over it with malloc. Execbufs involve quite a bit of payload, to the extent that cache misses show up in the profiles here, and a suspicion that some of those cachelines may get evicted and then reloaded in the subsequent copy. This is still abstracted like drm_calloc_large since we want to check for size overflow, and because we want to choose between kmalloc and vmalloc on the fly. cairo's interface for malloc-with-calloc's-args was used as the model. Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/i915_gem.c | 4 ++-- include/drm/drmP.h | 15 ++++++++++++++- 2 files changed, 16 insertions(+), 3 deletions(-) (limited to 'include/drm') diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 2065b8f7e875..481c0ab888c8 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -3563,8 +3563,8 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, return -EINVAL; } /* Copy in the exec list from userland */ - exec_list = drm_calloc_large(sizeof(*exec_list), args->buffer_count); - object_list = drm_calloc_large(sizeof(*object_list), args->buffer_count); + exec_list = drm_malloc_ab(sizeof(*exec_list), args->buffer_count); + object_list = drm_malloc_ab(sizeof(*object_list), args->buffer_count); if (exec_list == NULL || object_list == NULL) { DRM_ERROR("Failed to allocate exec or object list " "for %d buffers\n", diff --git a/include/drm/drmP.h b/include/drm/drmP.h index b0b36838ab11..1b807d0f6cdb 100644 --- a/include/drm/drmP.h +++ b/include/drm/drmP.h @@ -1545,14 +1545,27 @@ static __inline__ void drm_core_dropmap(struct drm_local_map *map) static __inline__ void *drm_calloc_large(size_t nmemb, size_t size) { + if (size != 0 && nmemb > ULONG_MAX / size) + return NULL; + if (size * nmemb <= PAGE_SIZE) return kcalloc(nmemb, size, GFP_KERNEL); + return __vmalloc(size * nmemb, + GFP_KERNEL | __GFP_HIGHMEM | __GFP_ZERO, PAGE_KERNEL); +} + +/* Modeled after cairo's malloc_ab, it's like calloc but without the zeroing. */ +static __inline__ void *drm_malloc_ab(size_t nmemb, size_t size) +{ if (size != 0 && nmemb > ULONG_MAX / size) return NULL; + if (size * nmemb <= PAGE_SIZE) + return kmalloc(nmemb * size, GFP_KERNEL); + return __vmalloc(size * nmemb, - GFP_KERNEL | __GFP_HIGHMEM | __GFP_ZERO, PAGE_KERNEL); + GFP_KERNEL | __GFP_HIGHMEM, PAGE_KERNEL); } static __inline void drm_free_large(void *ptr) -- cgit v1.2.3 From 7bd4d7be5c8c35a401b1589201e5d43a64d3f05b Mon Sep 17 00:00:00 2001 From: Jesse Barnes Date: Thu, 19 Nov 2009 10:50:22 -0800 Subject: drm: use page flip event to signal flip completion MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We don't actually know which frame number the flip will complete on, so userspace needs a specific flip notification to tell it when the last flip completed. Signed-off-by: Jesse Barnes Signed-off-by: Eric Anholt Acked-by: Kristian Høgsberg --- drivers/gpu/drm/drm_crtc.c | 2 +- include/drm/drm.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) (limited to 'include/drm') diff --git a/drivers/gpu/drm/drm_crtc.c b/drivers/gpu/drm/drm_crtc.c index ac2fa193072b..3bc870d38a97 100644 --- a/drivers/gpu/drm/drm_crtc.c +++ b/drivers/gpu/drm/drm_crtc.c @@ -2527,7 +2527,7 @@ int drm_mode_page_flip_ioctl(struct drm_device *dev, goto out; } - e->event.base.type = DRM_EVENT_VBLANK; + e->event.base.type = DRM_EVENT_FLIP_COMPLETE; e->event.base.length = sizeof e->event; e->event.user_data = page_flip->user_data; e->base.event = &e->event.base; diff --git a/include/drm/drm.h b/include/drm/drm.h index 3919a4f792ae..309d0a5ed68d 100644 --- a/include/drm/drm.h +++ b/include/drm/drm.h @@ -718,6 +718,7 @@ struct drm_event { }; #define DRM_EVENT_VBLANK 0x01 +#define DRM_EVENT_FLIP_COMPLETE 0x02 struct drm_event_vblank { struct drm_event base; -- cgit v1.2.3 From e9560f7cb20722e0e7db46bbb6f43c2194a238d5 Mon Sep 17 00:00:00 2001 From: Jesse Barnes Date: Thu, 19 Nov 2009 10:49:07 -0800 Subject: drm/i915: add GETPARAM request for page flipping Add a GETPARAM request for checking if page flipping is supported. Useful for the 2D driver to enable the flipping path. Signed-off-by: Jesse Barnes Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/i915_dma.c | 3 +++ include/drm/i915_drm.h | 1 + 2 files changed, 4 insertions(+) (limited to 'include/drm') diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c index a8efe78e5c57..fe89d0c723e6 100644 --- a/drivers/gpu/drm/i915/i915_dma.c +++ b/drivers/gpu/drm/i915/i915_dma.c @@ -810,6 +810,9 @@ static int i915_getparam(struct drm_device *dev, void *data, case I915_PARAM_HAS_OVERLAY: value = dev_priv->overlay ? 1 : 0; break; + case I915_PARAM_HAS_PAGEFLIPPING: + value = 1; + break; default: DRM_DEBUG_DRIVER("Unknown parameter %d\n", param->param); diff --git a/include/drm/i915_drm.h b/include/drm/i915_drm.h index c900499f2f63..1b4f3a565d78 100644 --- a/include/drm/i915_drm.h +++ b/include/drm/i915_drm.h @@ -271,6 +271,7 @@ typedef struct drm_i915_irq_wait { #define I915_PARAM_HAS_GEM 5 #define I915_PARAM_NUM_FENCES_AVAIL 6 #define I915_PARAM_HAS_OVERLAY 7 +#define I915_PARAM_HAS_PAGEFLIPPING 8 typedef struct drm_i915_getparam { int param; -- cgit v1.2.3 From 778c902640530371a169ad1c03566e7c51b09874 Mon Sep 17 00:00:00 2001 From: Li Peng Date: Mon, 9 Nov 2009 12:51:22 +0800 Subject: drm/i915: Fix sync to vblank when VGA output is turned off In current vblank-wait implementation, if we turn off VGA output, drm_wait_vblank will still wait on the disabled pipe until timeout, because vblank on the pipe is assumed be enabled. This would cause slow system response on some system such as moblin. This patch resolve the issue by adding a drm helper function drm_vblank_off which explicitly clear vblank_enabled[crtc], wake up any waiting queue and save last vblank counter before turning off crtc. It also slightly change drm_vblank_get to ensure that we will will return immediately if trying to wait on a disabled pipe. Signed-off-by: Li Peng Reviewed-by: Jesse Barnes [anholt: hand-applied for conflicts with overlay changes] Signed-off-by: Eric Anholt --- drivers/gpu/drm/drm_irq.c | 34 ++++++++++++++++++++++++++-------- drivers/gpu/drm/i915/intel_display.c | 1 + include/drm/drmP.h | 1 + 3 files changed, 28 insertions(+), 8 deletions(-) (limited to 'include/drm') diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c index 6b3ce6d38848..7998ee66b317 100644 --- a/drivers/gpu/drm/drm_irq.c +++ b/drivers/gpu/drm/drm_irq.c @@ -429,15 +429,21 @@ int drm_vblank_get(struct drm_device *dev, int crtc) spin_lock_irqsave(&dev->vbl_lock, irqflags); /* Going from 0->1 means we have to enable interrupts again */ - if (atomic_add_return(1, &dev->vblank_refcount[crtc]) == 1 && - !dev->vblank_enabled[crtc]) { - ret = dev->driver->enable_vblank(dev, crtc); - DRM_DEBUG("enabling vblank on crtc %d, ret: %d\n", crtc, ret); - if (ret) + if (atomic_add_return(1, &dev->vblank_refcount[crtc]) == 1) { + if (!dev->vblank_enabled[crtc]) { + ret = dev->driver->enable_vblank(dev, crtc); + DRM_DEBUG("enabling vblank on crtc %d, ret: %d\n", crtc, ret); + if (ret) + atomic_dec(&dev->vblank_refcount[crtc]); + else { + dev->vblank_enabled[crtc] = 1; + drm_update_vblank_count(dev, crtc); + } + } + } else { + if (!dev->vblank_enabled[crtc]) { atomic_dec(&dev->vblank_refcount[crtc]); - else { - dev->vblank_enabled[crtc] = 1; - drm_update_vblank_count(dev, crtc); + ret = -EINVAL; } } spin_unlock_irqrestore(&dev->vbl_lock, irqflags); @@ -464,6 +470,18 @@ void drm_vblank_put(struct drm_device *dev, int crtc) } EXPORT_SYMBOL(drm_vblank_put); +void drm_vblank_off(struct drm_device *dev, int crtc) +{ + unsigned long irqflags; + + spin_lock_irqsave(&dev->vbl_lock, irqflags); + DRM_WAKEUP(&dev->vbl_queue[crtc]); + dev->vblank_enabled[crtc] = 0; + dev->last_vblank[crtc] = dev->driver->get_vblank_counter(dev, crtc); + spin_unlock_irqrestore(&dev->vbl_lock, irqflags); +} +EXPORT_SYMBOL(drm_vblank_off); + /** * drm_vblank_pre_modeset - account for vblanks across mode sets * @dev: DRM device diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index 267adc6fbfc1..65b76fffd9e3 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -1924,6 +1924,7 @@ static void i9xx_crtc_dpms(struct drm_crtc *crtc, int mode) /* Give the overlay scaler a chance to disable if it's on this pipe */ intel_crtc_dpms_overlay(intel_crtc, false); + drm_vblank_off(dev, pipe); if (dev_priv->cfb_plane == plane && dev_priv->display.disable_fbc) diff --git a/include/drm/drmP.h b/include/drm/drmP.h index febf6c530c66..fd919959d146 100644 --- a/include/drm/drmP.h +++ b/include/drm/drmP.h @@ -1288,6 +1288,7 @@ extern u32 drm_vblank_count(struct drm_device *dev, int crtc); extern void drm_handle_vblank(struct drm_device *dev, int crtc); extern int drm_vblank_get(struct drm_device *dev, int crtc); extern void drm_vblank_put(struct drm_device *dev, int crtc); +extern void drm_vblank_off(struct drm_device *dev, int crtc); extern void drm_vblank_cleanup(struct drm_device *dev); /* Modesetting support */ extern void drm_vblank_pre_modeset(struct drm_device *dev, int crtc); -- cgit v1.2.3 From 04b2d218001bdddb41b84c9f78d6bb3cd3b5b31c Mon Sep 17 00:00:00 2001 From: Kristian Høgsberg Date: Fri, 6 Nov 2009 08:39:18 -0500 Subject: drm/i915: Fix typo in ioctl struct name. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Kristian Høgsberg Signed-off-by: Eric Anholt --- include/drm/i915_drm.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/drm') diff --git a/include/drm/i915_drm.h b/include/drm/i915_drm.h index 1b4f3a565d78..02045ed08eda 100644 --- a/include/drm/i915_drm.h +++ b/include/drm/i915_drm.h @@ -223,7 +223,7 @@ typedef struct _drm_i915_sarea { #define DRM_IOCTL_I915_GEM_SET_TILING DRM_IOWR (DRM_COMMAND_BASE + DRM_I915_GEM_SET_TILING, struct drm_i915_gem_set_tiling) #define DRM_IOCTL_I915_GEM_GET_TILING DRM_IOWR (DRM_COMMAND_BASE + DRM_I915_GEM_GET_TILING, struct drm_i915_gem_get_tiling) #define DRM_IOCTL_I915_GEM_GET_APERTURE DRM_IOR (DRM_COMMAND_BASE + DRM_I915_GEM_GET_APERTURE, struct drm_i915_gem_get_aperture) -#define DRM_IOCTL_I915_GET_PIPE_FROM_CRTC_ID DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_GET_PIPE_FROM_CRTC_ID, struct drm_intel_get_pipe_from_crtc_id) +#define DRM_IOCTL_I915_GET_PIPE_FROM_CRTC_ID DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_GET_PIPE_FROM_CRTC_ID, struct drm_i915_get_pipe_from_crtc_id) #define DRM_IOCTL_I915_GEM_MADVISE DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_GEM_MADVISE, struct drm_i915_gem_madvise) #define DRM_IOCTL_I915_OVERLAY_PUT_IMAGE DRM_IOW(DRM_COMMAND_BASE + DRM_IOCTL_I915_OVERLAY_ATTRS, struct drm_intel_overlay_put_image) #define DRM_IOCTL_I915_OVERLAY_ATTRS DRM_IOWR(DRM_COMMAND_BASE + DRM_I915_OVERLAY_ATTRS, struct drm_intel_overlay_attrs) -- cgit v1.2.3