OpenCloudOS-Kernel/include/drm/drmP.h

1529 lines
50 KiB
C
Raw Normal View History

/**
* \file drmP.h
* Private header for Direct Rendering Manager
*
* \author Rickard E. (Rik) Faith <faith@valinux.com>
* \author Gareth Hughes <gareth@valinux.com>
*/
/*
* Copyright 1999 Precision Insight, Inc., Cedar Park, Texas.
* Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California.
* Copyright (c) 2009-2010, Code Aurora Forum.
* All rights reserved.
*
* 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
* VA LINUX SYSTEMS AND/OR ITS SUPPLIERS 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.
*/
#ifndef _DRM_P_H_
#define _DRM_P_H_
#include <linux/agp_backend.h>
#include <linux/cdev.h>
#include <linux/dma-mapping.h>
#include <linux/file.h>
#include <linux/fs.h>
#include <linux/highmem.h>
#include <linux/idr.h>
#include <linux/init.h>
#include <linux/io.h>
#include <linux/jiffies.h>
#include <linux/kernel.h>
#include <linux/kref.h>
#include <linux/miscdevice.h>
#include <linux/mm.h>
#include <linux/mutex.h>
#include <linux/pci.h>
#include <linux/platform_device.h>
#include <linux/poll.h>
#include <linux/ratelimit.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/types.h>
#include <linux/vmalloc.h>
#include <linux/workqueue.h>
#include <asm/mman.h>
#include <asm/pgalloc.h>
#include <asm/uaccess.h>
#include <drm/drm.h>
#include <drm/drm_hashtab.h>
#include <drm/drm_mm.h>
#include <drm/drm_os_linux.h>
#include <drm/drm_sarea.h>
#include <drm/drm_vma_manager.h>
struct module;
struct drm_file;
struct drm_device;
struct drm_agp_head;
struct device_node;
struct videomode;
struct reservation_object;
/*
* 4 debug categories are defined:
*
* CORE: Used in the generic drm code: drm_ioctl.c, drm_mm.c, drm_memory.c, ...
* This is the category used by the DRM_DEBUG() macro.
*
* DRIVER: Used in the vendor specific part of the driver: i915, radeon, ...
* This is the category used by the DRM_DEBUG_DRIVER() macro.
*
* KMS: used in the modesetting code.
* This is the category used by the DRM_DEBUG_KMS() macro.
*
* PRIME: used in the prime code.
* This is the category used by the DRM_DEBUG_PRIME() macro.
*
* Enabling verbose debug messages is done through the drm.debug parameter,
* each category being enabled by a bit.
*
* drm.debug=0x1 will enable CORE messages
* drm.debug=0x2 will enable DRIVER messages
* drm.debug=0x3 will enable CORE and DRIVER messages
* ...
* drm.debug=0xf will enable all messages
*
* An interesting feature is that it's possible to enable verbose logging at
* run-time by echoing the debug value in its sysfs node:
* # echo 0xf > /sys/module/drm/parameters/debug
*/
#define DRM_UT_CORE 0x01
#define DRM_UT_DRIVER 0x02
#define DRM_UT_KMS 0x04
#define DRM_UT_PRIME 0x08
2009-06-02 14:09:47 +08:00
extern __printf(2, 3)
void drm_ut_debug_printk(const char *function_name,
const char *format, ...);
extern __printf(2, 3)
int drm_err(const char *func, const char *format, ...);
/***********************************************************************/
/** \name DRM template customization defaults */
/*@{*/
/* driver capabilities and requirements mask */
#define DRIVER_USE_AGP 0x1
#define DRIVER_PCI_DMA 0x8
#define DRIVER_SG 0x10
#define DRIVER_HAVE_DMA 0x20
#define DRIVER_HAVE_IRQ 0x40
#define DRIVER_IRQ_SHARED 0x80
#define DRIVER_GEM 0x1000
#define DRIVER_MODESET 0x2000
#define DRIVER_PRIME 0x4000
drm: implement experimental render nodes Render nodes provide an API for userspace to use non-privileged GPU commands without any running DRM-Master. It is useful for offscreen rendering, GPGPU clients, and normal render clients which do not perform modesetting. Compared to legacy clients, render clients no longer need any authentication to perform client ioctls. Instead, user-space controls render/client access to GPUs via filesystem access-modes on the render-node. Once a render-node was opened, a client has full access to the client/render operations on the GPU. However, no modesetting or ioctls that affect global state are allowed on render nodes. To prevent privilege-escalation, drivers must explicitly state that they support render nodes. They must mark their render-only ioctls as DRM_RENDER_ALLOW so render clients can use them. Furthermore, they must support clients without any attached master. If filesystem access-modes are not enough for fine-grained access control to render nodes (very unlikely, considering the versaitlity of FS-ACLs), you may still fall-back to fd-passing from server to client (which allows arbitrary access-control). However, note that revoking access is currently impossible and unlikely to get implemented. Note: Render clients no longer have any associated DRM-Master as they are supposed to be independent of any server state. DRM core highly depends on file_priv->master to be non-NULL for modesetting/ctx/etc. commands. Therefore, drivers must be very careful to not require DRM-Master if they support DRIVER_RENDER. So far render-nodes are protected by "drm_rnodes". As long as this module-parameter is not set to 1, a driver will not create render nodes. This allows us to experiment with the API a bit before we stabilize it. v2: drop insecure GEM_FLINK to force use of dmabuf Signed-off-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-08-26 00:29:00 +08:00
#define DRIVER_RENDER 0x8000
/***********************************************************************/
/** \name Begin the DRM... */
/*@{*/
#define DRM_MAGIC_HASH_ORDER 4 /**< Size of key hash table. Must be power of 2. */
/*@}*/
/***********************************************************************/
/** \name Macros to make printk easier */
/*@{*/
/**
* Error output.
*
* \param fmt printf() like format string.
* \param arg arguments
*/
#define DRM_ERROR(fmt, ...) \
drm_err(__func__, fmt, ##__VA_ARGS__)
/**
* Rate limited error output. Like DRM_ERROR() but won't flood the log.
*
* \param fmt printf() like format string.
* \param arg arguments
*/
#define DRM_ERROR_RATELIMITED(fmt, ...) \
({ \
static DEFINE_RATELIMIT_STATE(_rs, \
DEFAULT_RATELIMIT_INTERVAL, \
DEFAULT_RATELIMIT_BURST); \
\
if (__ratelimit(&_rs)) \
drm_err(__func__, fmt, ##__VA_ARGS__); \
})
#define DRM_INFO(fmt, ...) \
printk(KERN_INFO "[" DRM_NAME "] " fmt, ##__VA_ARGS__)
#define DRM_INFO_ONCE(fmt, ...) \
printk_once(KERN_INFO "[" DRM_NAME "] " fmt, ##__VA_ARGS__)
/**
* Debug output.
*
* \param fmt printf() like format string.
* \param arg arguments
*/
2009-06-02 14:09:47 +08:00
#define DRM_DEBUG(fmt, args...) \
do { \
if (unlikely(drm_debug & DRM_UT_CORE)) \
drm_ut_debug_printk(__func__, fmt, ##args); \
2009-06-02 14:09:47 +08:00
} while (0)
#define DRM_DEBUG_DRIVER(fmt, args...) \
2009-06-02 14:09:47 +08:00
do { \
if (unlikely(drm_debug & DRM_UT_DRIVER)) \
drm_ut_debug_printk(__func__, fmt, ##args); \
2009-06-02 14:09:47 +08:00
} while (0)
#define DRM_DEBUG_KMS(fmt, args...) \
2009-06-02 14:09:47 +08:00
do { \
if (unlikely(drm_debug & DRM_UT_KMS)) \
drm_ut_debug_printk(__func__, fmt, ##args); \
2009-06-02 14:09:47 +08:00
} while (0)
#define DRM_DEBUG_PRIME(fmt, args...) \
do { \
if (unlikely(drm_debug & DRM_UT_PRIME)) \
drm_ut_debug_printk(__func__, fmt, ##args); \
} while (0)
/*@}*/
/***********************************************************************/
/** \name Internal types and structures */
/*@{*/
#define DRM_IF_VERSION(maj, min) (maj << 16 | min)
/**
* Test that the hardware lock is held by the caller, returning otherwise.
*
* \param dev DRM device.
* \param filp file pointer of the caller.
*/
#define LOCK_TEST_WITH_RETURN( dev, _file_priv ) \
do { \
if (!_DRM_LOCK_IS_HELD(_file_priv->master->lock.hw_lock->lock) || \
_file_priv->master->lock.file_priv != _file_priv) { \
DRM_ERROR( "%s called without lock held, held %d owner %p %p\n",\
__func__, _DRM_LOCK_IS_HELD(_file_priv->master->lock.hw_lock->lock),\
_file_priv->master->lock.file_priv, _file_priv); \
return -EINVAL; \
} \
} while (0)
/**
* Ioctl function type.
*
* \param inode device inode.
* \param file_priv DRM file private pointer.
* \param cmd command.
* \param arg argument.
*/
typedef int drm_ioctl_t(struct drm_device *dev, void *data,
struct drm_file *file_priv);
drm: 32/64-bit DRM ioctl compatibility patch The patch is against a 2.6.11 kernel tree. I am running this with a 32-bit X server (compiled up from X.org CVS as of a couple of weeks ago) and 32-bit DRI libraries and clients. All the userland stuff is identical to what I am using under a 32-bit kernel on my G4 powerbook (which is a 32-bit machine of course). I haven't tried compiling up a 64-bit X server or clients yet. In the compatibility routines I have assumed that the kernel can safely access user addresses after set_fs(KERNEL_DS). That is, where an ioctl argument structure contains pointers to other structures, and those other structures are already compatible between the 32-bit and 64-bit ABIs (i.e. they only contain things like chars, shorts or ints), I just check the address with access_ok() and then pass it through to the 64-bit ioctl code. I believe this approach may not work on sparc64, but it does work on ppc64 and x86_64 at least. One tricky area which may need to be revisited is the question of how to handle the handles which we pass back to userspace to identify mappings. These handles are generated in the ADDMAP ioctl and then passed in as the offset value to mmap. However, offset values for mmap seem to be generated in other ways as well, particularly for AGP mappings. The approach I have ended up with is to generate a fake 32-bit handle only for _DRM_SHM mappings. The handles for other mappings (AGP, REG, FB) are physical addresses which are already limited to 32 bits, and generating fake handles for them created all sorts of problems in the mmap/nopage code. This patch has been updated to use the new compatibility ioctls. From: Paul Mackerras <paulus@samba.org> Signed-off-by: Dave Airlie <airlied@linux.ie>
2005-06-23 19:29:18 +08:00
typedef int drm_ioctl_compat_t(struct file *filp, unsigned int cmd,
unsigned long arg);
#define DRM_IOCTL_NR(n) _IOC_NR(n)
#define DRM_MAJOR 226
#define DRM_AUTH 0x1
#define DRM_MASTER 0x2
#define DRM_ROOT_ONLY 0x4
#define DRM_CONTROL_ALLOW 0x8
#define DRM_UNLOCKED 0x10
drm: implement experimental render nodes Render nodes provide an API for userspace to use non-privileged GPU commands without any running DRM-Master. It is useful for offscreen rendering, GPGPU clients, and normal render clients which do not perform modesetting. Compared to legacy clients, render clients no longer need any authentication to perform client ioctls. Instead, user-space controls render/client access to GPUs via filesystem access-modes on the render-node. Once a render-node was opened, a client has full access to the client/render operations on the GPU. However, no modesetting or ioctls that affect global state are allowed on render nodes. To prevent privilege-escalation, drivers must explicitly state that they support render nodes. They must mark their render-only ioctls as DRM_RENDER_ALLOW so render clients can use them. Furthermore, they must support clients without any attached master. If filesystem access-modes are not enough for fine-grained access control to render nodes (very unlikely, considering the versaitlity of FS-ACLs), you may still fall-back to fd-passing from server to client (which allows arbitrary access-control). However, note that revoking access is currently impossible and unlikely to get implemented. Note: Render clients no longer have any associated DRM-Master as they are supposed to be independent of any server state. DRM core highly depends on file_priv->master to be non-NULL for modesetting/ctx/etc. commands. Therefore, drivers must be very careful to not require DRM-Master if they support DRIVER_RENDER. So far render-nodes are protected by "drm_rnodes". As long as this module-parameter is not set to 1, a driver will not create render nodes. This allows us to experiment with the API a bit before we stabilize it. v2: drop insecure GEM_FLINK to force use of dmabuf Signed-off-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-08-26 00:29:00 +08:00
#define DRM_RENDER_ALLOW 0x20
struct drm_ioctl_desc {
unsigned int cmd;
int flags;
drm_ioctl_t *func;
unsigned int cmd_drv;
drm: Use names of ioctls in debug traces The intention here is to make the output of dmesg with full verbosity a bit easier for a human to parse. This commit transforms: [drm:drm_ioctl], pid=699, cmd=0x6458, nr=0x58, dev 0xe200, auth=1 [drm:drm_ioctl], pid=699, cmd=0xc010645b, nr=0x5b, dev 0xe200, auth=1 [drm:drm_ioctl], pid=699, cmd=0xc0106461, nr=0x61, dev 0xe200, auth=1 [drm:drm_ioctl], pid=699, cmd=0xc01c64ae, nr=0xae, dev 0xe200, auth=1 [drm:drm_mode_addfb], [FB:32] [drm:drm_ioctl], pid=699, cmd=0xc0106464, nr=0x64, dev 0xe200, auth=1 [drm:drm_vm_open_locked], 0x7fd9302fe000,0x00a00000 [drm:drm_ioctl], pid=699, cmd=0x400c645f, nr=0x5f, dev 0xe200, auth=1 [drm:drm_ioctl], pid=699, cmd=0xc00464af, nr=0xaf, dev 0xe200, auth=1 [drm:intel_crtc_set_config], [CRTC:3] [NOFB] into: [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_THROTTLE [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_CREATE [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_SET_TILING [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, IOCTL_MODE_ADDFB [drm:drm_mode_addfb], [FB:32] [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_MMAP_GTT [drm:drm_vm_open_locked], 0x7fd9302fe000,0x00a00000 [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_SET_DOMAIN [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, DRM_IOCTL_MODE_RMFB [drm:intel_crtc_set_config], [CRTC:3] [NOFB] v2: drm_ioctls is now a constant (Ville Syrjälä) Signed-off-by: Chris Cummins <christopher.e.cummins@intel.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-05-09 21:20:40 +08:00
const char *name;
};
/**
* Creates a driver or general drm_ioctl_desc array entry for the given
* ioctl, for use by drm_ioctl().
*/
#define DRM_IOCTL_DEF_DRV(ioctl, _func, _flags) \
drm: Use names of ioctls in debug traces The intention here is to make the output of dmesg with full verbosity a bit easier for a human to parse. This commit transforms: [drm:drm_ioctl], pid=699, cmd=0x6458, nr=0x58, dev 0xe200, auth=1 [drm:drm_ioctl], pid=699, cmd=0xc010645b, nr=0x5b, dev 0xe200, auth=1 [drm:drm_ioctl], pid=699, cmd=0xc0106461, nr=0x61, dev 0xe200, auth=1 [drm:drm_ioctl], pid=699, cmd=0xc01c64ae, nr=0xae, dev 0xe200, auth=1 [drm:drm_mode_addfb], [FB:32] [drm:drm_ioctl], pid=699, cmd=0xc0106464, nr=0x64, dev 0xe200, auth=1 [drm:drm_vm_open_locked], 0x7fd9302fe000,0x00a00000 [drm:drm_ioctl], pid=699, cmd=0x400c645f, nr=0x5f, dev 0xe200, auth=1 [drm:drm_ioctl], pid=699, cmd=0xc00464af, nr=0xaf, dev 0xe200, auth=1 [drm:intel_crtc_set_config], [CRTC:3] [NOFB] into: [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_THROTTLE [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_CREATE [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_SET_TILING [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, IOCTL_MODE_ADDFB [drm:drm_mode_addfb], [FB:32] [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_MMAP_GTT [drm:drm_vm_open_locked], 0x7fd9302fe000,0x00a00000 [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, I915_GEM_SET_DOMAIN [drm:drm_ioctl], pid=699, dev=0xe200, auth=1, DRM_IOCTL_MODE_RMFB [drm:intel_crtc_set_config], [CRTC:3] [NOFB] v2: drm_ioctls is now a constant (Ville Syrjälä) Signed-off-by: Chris Cummins <christopher.e.cummins@intel.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-05-09 21:20:40 +08:00
[DRM_IOCTL_NR(DRM_##ioctl)] = {.cmd = DRM_##ioctl, .func = _func, .flags = _flags, .cmd_drv = DRM_IOCTL_##ioctl, .name = #ioctl}
/**
* DMA buffer.
*/
struct drm_buf {
int idx; /**< Index into master buflist */
int total; /**< Buffer size */
int order; /**< log-base-2(total) */
int used; /**< Amount of buffer in use (for DMA) */
unsigned long offset; /**< Byte offset (used internally) */
void *address; /**< Address of buffer */
unsigned long bus_address; /**< Bus address of buffer */
struct drm_buf *next; /**< Kernel-only: used for free list */
__volatile__ int waiting; /**< On kernel DMA queue */
__volatile__ int pending; /**< On hardware DMA queue */
struct drm_file *file_priv; /**< Private of holding file descr */
int context; /**< Kernel queue for this buffer */
int while_locked; /**< Dispatch this buffer while locked */
enum {
DRM_LIST_NONE = 0,
DRM_LIST_FREE = 1,
DRM_LIST_WAIT = 2,
DRM_LIST_PEND = 3,
DRM_LIST_PRIO = 4,
DRM_LIST_RECLAIM = 5
} list; /**< Which list we're on */
int dev_priv_size; /**< Size of buffer private storage */
void *dev_private; /**< Per-buffer private storage */
};
typedef struct drm_dma_handle {
dma_addr_t busaddr;
void *vaddr;
size_t size;
} drm_dma_handle_t;
/**
* Buffer entry. There is one of this for each buffer size order.
*/
struct drm_buf_entry {
int buf_size; /**< size */
int buf_count; /**< number of buffers */
struct drm_buf *buflist; /**< buffer list */
int seg_count;
int page_order;
struct drm_dma_handle **seglist;
int low_mark; /**< Low water mark */
int high_mark; /**< High water mark */
};
/* Event queued up for userspace to read */
struct drm_pending_event {
struct drm_event *event;
struct list_head link;
struct drm_file *file_priv;
pid_t pid; /* pid of requester, no guarantee it's valid by the time
we deliver the event, for tracing only */
void (*destroy)(struct drm_pending_event *event);
};
/* initial implementaton using a linked list - todo hashtab */
struct drm_prime_file_private {
struct list_head head;
struct mutex lock;
};
/** File private data */
struct drm_file {
unsigned authenticated :1;
/* Whether we're master for a minor. Protected by master_mutex */
unsigned is_master :1;
/* true when the client has asked us to expose stereo 3D mode flags */
unsigned stereo_allowed :1;
/*
* true if client understands CRTC primary planes and cursor planes
* in the plane list
*/
unsigned universal_planes:1;
struct pid *pid;
kuid_t uid;
drm_magic_t magic;
struct list_head lhead;
struct drm_minor *minor;
unsigned long lock_count;
/** Mapping of mm object handles to object pointers. */
struct idr object_idr;
/** Lock for synchronization of access to object_idr. */
spinlock_t table_lock;
struct file *filp;
void *driver_priv;
struct drm_master *master; /* master this node is currently associated with
N.B. not always minor->master */
drm: revamp locking around fb creation/destruction Well, at least step 1. The goal here is that framebuffer objects can survive outside of the mode_config lock, with just a reference held as protection. The first step to get there is to introduce a special fb_lock which protects fb lookup, creation and destruction, to make them appear atomic. This new fb_lock can nest within the mode_config lock. But the idea is (once the reference counting part is completed) that we only quickly take that fb_lock to lookup a framebuffer and grab a reference, without any other locks involved. vmwgfx is the only driver which does framebuffer lookups itself, also wrap those calls to drm_mode_object_find with the new lock. Also protect the fb_list walking in i915 and omapdrm with the new lock. As a slight complication there's also the list of user-created fbs attached to the file private. The problem now is that at fclose() time we need to walk that list, eventually do a modeset call to remove the fb from active usage (and are required to be able to take the mode_config lock), but in the end we need to grab the new fb_lock to remove the fb from the list. The easiest solution is to add another mutex to protect this per-file list. Currently that new fbs_lock nests within the modeset locks and so appears redudant. But later patches will switch around this sequence so that taking the modeset locks in the fb destruction path is optional in the fastpath. Ultimately the goal is that addfb and rmfb do not require the mode_config lock, since otherwise they have the potential to introduce stalls in the pageflip sequence of a compositor (if the compositor e.g. switches to a fullscreen client or if it enables a plane). But that requires a few more steps and hoops to jump through. Note that framebuffer creation/destruction is now double-protected - once by the fb_lock and in parts by the idr_lock. The later would be unnecessariy if framebuffers would have their own idr allocator. But that's material for another patch (series). v2: Properly initialize the fb->filp_head list in _init, otherwise the newly added WARN to check whether the fb isn't on a fpriv list any more will fail for driver-private objects. v3: Fixup two error-case unlock bugs spotted by Richard Wilbur. Reviewed-by: Rob Clark <rob@ti.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-11 04:19:18 +08:00
/**
* fbs - List of framebuffers associated with this file.
*
* Protected by fbs_lock. Note that the fbs list holds a reference on
* the fb object to prevent it from untimely disappearing.
*/
struct list_head fbs;
drm: revamp locking around fb creation/destruction Well, at least step 1. The goal here is that framebuffer objects can survive outside of the mode_config lock, with just a reference held as protection. The first step to get there is to introduce a special fb_lock which protects fb lookup, creation and destruction, to make them appear atomic. This new fb_lock can nest within the mode_config lock. But the idea is (once the reference counting part is completed) that we only quickly take that fb_lock to lookup a framebuffer and grab a reference, without any other locks involved. vmwgfx is the only driver which does framebuffer lookups itself, also wrap those calls to drm_mode_object_find with the new lock. Also protect the fb_list walking in i915 and omapdrm with the new lock. As a slight complication there's also the list of user-created fbs attached to the file private. The problem now is that at fclose() time we need to walk that list, eventually do a modeset call to remove the fb from active usage (and are required to be able to take the mode_config lock), but in the end we need to grab the new fb_lock to remove the fb from the list. The easiest solution is to add another mutex to protect this per-file list. Currently that new fbs_lock nests within the modeset locks and so appears redudant. But later patches will switch around this sequence so that taking the modeset locks in the fb destruction path is optional in the fastpath. Ultimately the goal is that addfb and rmfb do not require the mode_config lock, since otherwise they have the potential to introduce stalls in the pageflip sequence of a compositor (if the compositor e.g. switches to a fullscreen client or if it enables a plane). But that requires a few more steps and hoops to jump through. Note that framebuffer creation/destruction is now double-protected - once by the fb_lock and in parts by the idr_lock. The later would be unnecessariy if framebuffers would have their own idr allocator. But that's material for another patch (series). v2: Properly initialize the fb->filp_head list in _init, otherwise the newly added WARN to check whether the fb isn't on a fpriv list any more will fail for driver-private objects. v3: Fixup two error-case unlock bugs spotted by Richard Wilbur. Reviewed-by: Rob Clark <rob@ti.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-11 04:19:18 +08:00
struct mutex fbs_lock;
wait_queue_head_t event_wait;
struct list_head event_list;
int event_space;
struct drm_prime_file_private prime;
};
/**
* Lock data.
*/
struct drm_lock_data {
struct drm_hw_lock *hw_lock; /**< Hardware lock */
/** Private of lock holder's file (NULL=kernel) */
struct drm_file *file_priv;
wait_queue_head_t lock_queue; /**< Queue of blocked processes */
unsigned long lock_time; /**< Time of last lock in jiffies */
spinlock_t spinlock;
uint32_t kernel_waiters;
uint32_t user_waiters;
int idle_has_lock;
};
/**
* DMA data.
*/
struct drm_device_dma {
struct drm_buf_entry bufs[DRM_MAX_ORDER + 1]; /**< buffers, grouped by their size order */
int buf_count; /**< total number of buffers */
struct drm_buf **buflist; /**< Vector of pointers into drm_device_dma::bufs */
int seg_count;
int page_count; /**< number of pages */
unsigned long *pagelist; /**< page list */
unsigned long byte_count;
enum {
_DRM_DMA_USE_AGP = 0x01,
_DRM_DMA_USE_SG = 0x02,
_DRM_DMA_USE_FB = 0x04,
_DRM_DMA_USE_PCI_RO = 0x08
} flags;
};
/**
* Scatter-gather memory.
*/
struct drm_sg_mem {
unsigned long handle;
void *virtual;
int pages;
struct page **pagelist;
dma_addr_t *busaddr;
};
/**
* Kernel side of a mapping
*/
struct drm_local_map {
resource_size_t offset; /**< Requested physical address (0 for SAREA)*/
unsigned long size; /**< Requested physical size (bytes) */
enum drm_map_type type; /**< Type of memory to map */
enum drm_map_flags flags; /**< Flags */
void *handle; /**< User-space: "Handle" to pass to mmap() */
/**< Kernel-space: kernel-virtual address */
int mtrr; /**< MTRR slot used */
};
typedef struct drm_local_map drm_local_map_t;
/**
* Mappings list
*/
struct drm_map_list {
struct list_head head; /**< list head */
struct drm_hash_item hash;
struct drm_local_map *map; /**< mapping */
uint64_t user_token;
struct drm_master *master;
};
/* location of GART table */
#define DRM_ATI_GART_MAIN 1
#define DRM_ATI_GART_FB 2
#define DRM_ATI_GART_PCI 1
#define DRM_ATI_GART_PCIE 2
#define DRM_ATI_GART_IGP 3
struct drm_ati_pcigart_info {
int gart_table_location;
int gart_reg_if;
void *addr;
dma_addr_t bus_addr;
dma_addr_t table_mask;
struct drm_dma_handle *table_handle;
struct drm_local_map mapping;
int table_size;
};
/**
* This structure defines the drm_mm memory object, which will be used by the
* DRM for its buffer objects.
*/
struct drm_gem_object {
/** Reference count of this object */
struct kref refcount;
drm/gem: fix up flink name create race This is the 2nd attempt, I've always been a bit dissatisified with the tricky nature of the first one: http://lists.freedesktop.org/archives/dri-devel/2012-July/025451.html The issue is that the flink ioctl can race with calling gem_close on the last gem handle. In that case we'll end up with a zero handle count, but an flink name (and it's corresponding reference). Which results in a neat space leak. In my first attempt I've solved this by rechecking the handle count. But fundamentally the issue is that ->handle_count isn't your usual refcount - it can be resurrected from 0 among other things. For those special beasts atomic_t often suggest way more ordering that it actually guarantees. To prevent being tricked by those hairy semantics take the easy way out and simply protect the handle with the existing dev->object_name_lock. With that change implemented it's dead easy to fix the flink vs. gem close reace: When we try to create the name we simply have to check whether there's still officially a gem handle around and if not refuse to create the flink name. Since the handle count decrement and flink name destruction is now also protected by that lock the reace is gone and we can't ever leak the flink reference again. Outside of the drm core only the exynos driver looks at the handle count, and tbh I have no idea why (it's just for debug dmesg output luckily). I've considered inlining the drm_gem_object_handle_free, but I plan to add more name-like things (like the exported dma_buf) to this scheme, so it's clearer to leave the handle freeing in its own function. This is exercised by the new gem_flink_race i-g-t testcase, which on my snb leaks gem objects at a rate of roughly 1k objects/s. v2: Fix up the error path handling in handle_create and make it more robust by simply calling object_handle_unreference. v3: Fix up the handle_unreference logic bug - atomic_dec_and_test retursn 1 for 0. Oops. v4: Squash in inlining of drm_gem_object_handle_reference as suggested by Dave Airlie and add a note that we now have a testcase. Cc: Dave Airlie <airlied@gmail.com> Cc: Inki Dae <inki.dae@samsung.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-08-15 06:02:37 +08:00
/**
* handle_count - gem file_priv handle count of this object
*
* Each handle also holds a reference. Note that when the handle_count
* drops to 0 any global names (e.g. the id in the flink namespace) will
* be cleared.
*
* Protected by dev->object_name_lock.
* */
unsigned handle_count;
/** Related drm device */
struct drm_device *dev;
/** File representing the shmem storage */
struct file *filp;
/* Mapping info for this object */
struct drm_vma_offset_node vma_node;
/**
* Size of the object, in bytes. Immutable over the object's
* lifetime.
*/
size_t size;
/**
* Global name for this object, starts at 1. 0 means unnamed.
* Access is covered by the object_name_lock in the related drm_device
*/
int name;
/**
* Memory domains. These monitor which caches contain read/write data
* related to the object. When transitioning from one set of domains
* to another, the driver is called to ensure that caches are suitably
* flushed and invalidated
*/
uint32_t read_domains;
uint32_t write_domain;
/**
* While validating an exec operation, the
* new read/write domain values are computed here.
* They will be transferred to the above values
* at the point that any cache flushing occurs
*/
uint32_t pending_read_domains;
uint32_t pending_write_domain;
drm/prime: proper locking+refcounting for obj->dma_buf link The export dma-buf cache is semantically similar to an flink name. So semantically it makes sense to treat it the same and remove the name (i.e. the dma_buf pointer) and its references when the last gem handle disappears. Again we need to be careful, but double so: Not just could someone race and export with a gem close ioctl (so we need to recheck obj->handle_count again when assigning the new name), but multiple exports can also race against each another. This is prevented by holding the dev->object_name_lock across the entire section which touches obj->dma_buf. With the new scheme we also need to reinstate the obj->dma_buf link at import time (in case the only reference userspace has held in-between was through the dma-buf fd and not through any native gem handle). For simplicity we don't check whether it's a native object but unconditionally set up that link - with the new scheme of removing the obj->dma_buf reference when the last handle disappears we can do that. To make it clear that this is not just for exported buffers anymore als rename it from export_dma_buf to dma_buf. To make sure that now one can race a fd_to_handle or handle_to_fd with gem_close we use the same tricks as in flink of extending the dev->object_name_locking critical section. With this change we finally have a guaranteed 1:1 relationship (at least for native objects) between gem objects and dma-bufs, even accounting for races (which can happen since the dma-buf itself holds a reference while in-flight). This prevent igt/prime_self_import/export-vs-gem_close-race from Oopsing the kernel. There is still a leak though since the per-file priv dma-buf/handle cache handling is racy. That will be fixed in a later patch. v2: Remove the bogus dma_buf_put from the export_and_register_object failure path if we've raced with the handle count dropping to 0. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-08-15 06:02:46 +08:00
/**
* dma_buf - dma buf associated with this GEM object
*
* Pointer to the dma-buf associated with this gem object (either
* through importing or exporting). We break the resulting reference
* loop when the last gem handle for this object is released.
*
* Protected by obj->object_name_lock
*/
struct dma_buf *dma_buf;
/**
* import_attach - dma buf attachment backing this object
*
* Any foreign dma_buf imported as a gem object has this set to the
* attachment point for the device. This is invariant over the lifetime
* of a gem object.
*
* The driver's ->gem_free_object callback is responsible for cleaning
* up the dma_buf attachment and references acquired at import time.
*
* Note that the drm gem/prime core does not depend upon drivers setting
* this field any more. So for drivers where this doesn't make sense
* (e.g. virtual devices or a displaylink behind an usb bus) they can
* simply leave it as NULL.
*/
struct dma_buf_attachment *import_attach;
};
#include <drm/drm_crtc.h>
/**
* struct drm_master - drm master structure
*
* @refcount: Refcount for this master object.
* @minor: Link back to minor char device we are master for. Immutable.
* @unique: Unique identifier: e.g. busid. Protected by drm_global_mutex.
* @unique_len: Length of unique field. Protected by drm_global_mutex.
* @unique_size: Amount allocated. Protected by drm_global_mutex.
* @magiclist: Hash of used authentication tokens. Protected by struct_mutex.
* @magicfree: List of used authentication tokens. Protected by struct_mutex.
* @lock: DRI lock information.
* @driver_priv: Pointer to driver-private information.
*/
struct drm_master {
struct kref refcount;
struct drm_minor *minor;
char *unique;
int unique_len;
int unique_size;
struct drm_open_hash magiclist;
struct list_head magicfree;
struct drm_lock_data lock;
void *driver_priv;
};
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
/* Size of ringbuffer for vblank timestamps. Just double-buffer
* in initial implementation.
*/
#define DRM_VBLANKTIME_RBSIZE 2
/* Flags and return codes for get_vblank_timestamp() driver function. */
#define DRM_CALLED_FROM_VBLIRQ 1
#define DRM_VBLANKTIME_SCANOUTPOS_METHOD (1 << 0)
#define DRM_VBLANKTIME_INVBL (1 << 1)
/* get_scanout_position() return flags */
#define DRM_SCANOUTPOS_VALID (1 << 0)
#define DRM_SCANOUTPOS_INVBL (1 << 1)
#define DRM_SCANOUTPOS_ACCURATE (1 << 2)
struct drm_bus {
int (*set_busid)(struct drm_device *dev, struct drm_master *master);
};
/**
* DRM driver structure. This structure represent the common code for
* a family of cards. There will one drm_device for each card present
* in this family
*/
struct drm_driver {
int (*load) (struct drm_device *, unsigned long flags);
int (*firstopen) (struct drm_device *);
int (*open) (struct drm_device *, struct drm_file *);
void (*preclose) (struct drm_device *, struct drm_file *file_priv);
void (*postclose) (struct drm_device *, struct drm_file *);
void (*lastclose) (struct drm_device *);
int (*unload) (struct drm_device *);
int (*suspend) (struct drm_device *, pm_message_t state);
int (*resume) (struct drm_device *);
int (*dma_ioctl) (struct drm_device *dev, void *data, struct drm_file *file_priv);
int (*dma_quiescent) (struct drm_device *);
int (*context_dtor) (struct drm_device *dev, int context);
/**
* get_vblank_counter - get raw hardware vblank counter
* @dev: DRM device
* @crtc: counter to fetch
*
* Driver callback for fetching a raw hardware vblank counter for @crtc.
* If a device doesn't have a hardware counter, the driver can simply
* return the value of drm_vblank_count. The DRM core will account for
* missed vblank events while interrupts where disabled based on system
* timestamps.
*
* Wraparound handling and loss of events due to modesetting is dealt
* with in the DRM core code.
*
* RETURNS
* Raw vblank counter value.
*/
u32 (*get_vblank_counter) (struct drm_device *dev, int crtc);
/**
* enable_vblank - enable vblank interrupt events
* @dev: DRM device
* @crtc: which irq to enable
*
* Enable vblank interrupts for @crtc. If the device doesn't have
* a hardware vblank counter, this routine should be a no-op, since
* interrupts will have to stay on to keep the count accurate.
*
* RETURNS
* Zero on success, appropriate errno if the given @crtc's vblank
* interrupt cannot be enabled.
*/
int (*enable_vblank) (struct drm_device *dev, int crtc);
/**
* disable_vblank - disable vblank interrupt events
* @dev: DRM device
* @crtc: which irq to enable
*
* Disable vblank interrupts for @crtc. If the device doesn't have
* a hardware vblank counter, this routine should be a no-op, since
* interrupts will have to stay on to keep the count accurate.
*/
void (*disable_vblank) (struct drm_device *dev, int crtc);
/**
* Called by \c drm_device_is_agp. Typically used to determine if a
* card is really attached to AGP or not.
*
* \param dev DRM device handle
*
* \returns
* One of three values is returned depending on whether or not the
* card is absolutely \b not AGP (return of 0), absolutely \b is AGP
* (return of 1), or may or may not be AGP (return of 2).
*/
int (*device_is_agp) (struct drm_device *dev);
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
/**
* Called by vblank timestamping code.
*
drm: Push latency sensitive bits of vblank scanoutpos timestamping into kms drivers. A change in locking of some kms drivers (currently intel-kms) make the old approach too inaccurate and also incompatible with the PREEMPT_RT realtime kernel patchset. The driver->get_scanout_position() method of intel-kms now needs to aquire a spinlock, which clashes badly with the former preempt_disable() calls in the drm, and it also introduces larger delays and timing uncertainty on a contended lock than acceptable. This patch changes the prototype of driver->get_scanout_position() to require/allow kms drivers to perform the ktime_get() system time queries which go along with actual scanout position readout in a way that provides maximum precision and to return those timestamps to the drm. kms drivers implementations of get_scanout_position() are asked to implement timestamping and scanoutpos readout in a way that is as precise as possible and compatible with preempt_disable() on a PREMPT_RT kernel. A driver should follow this pattern in get_scanout_position() for precision and compatibility: spin_lock...(...); preempt_disable_rt(); // On a PREEMPT_RT kernel, otherwise omit. if (stime) *stime = ktime_get(); ... Minimum amount of MMIO register reads to get scanout position ... ... no taking of locks allowed here! ... if (etime) *etime = ktime_get(); preempt_enable_rt(); // On PREEMPT_RT kernel, otherwise omit. spin_unlock...(...); v2: Fix formatting of new multi-line code comments. Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-10-30 12:13:06 +08:00
* Return the current display scanout position from a crtc, and an
* optional accurate ktime_get timestamp of when position was measured.
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
*
* \param dev DRM device.
* \param crtc Id of the crtc to query.
* \param flags Flags from the caller (DRM_CALLED_FROM_VBLIRQ or 0).
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
* \param *vpos Target location for current vertical scanout position.
* \param *hpos Target location for current horizontal scanout position.
drm: Push latency sensitive bits of vblank scanoutpos timestamping into kms drivers. A change in locking of some kms drivers (currently intel-kms) make the old approach too inaccurate and also incompatible with the PREEMPT_RT realtime kernel patchset. The driver->get_scanout_position() method of intel-kms now needs to aquire a spinlock, which clashes badly with the former preempt_disable() calls in the drm, and it also introduces larger delays and timing uncertainty on a contended lock than acceptable. This patch changes the prototype of driver->get_scanout_position() to require/allow kms drivers to perform the ktime_get() system time queries which go along with actual scanout position readout in a way that provides maximum precision and to return those timestamps to the drm. kms drivers implementations of get_scanout_position() are asked to implement timestamping and scanoutpos readout in a way that is as precise as possible and compatible with preempt_disable() on a PREMPT_RT kernel. A driver should follow this pattern in get_scanout_position() for precision and compatibility: spin_lock...(...); preempt_disable_rt(); // On a PREEMPT_RT kernel, otherwise omit. if (stime) *stime = ktime_get(); ... Minimum amount of MMIO register reads to get scanout position ... ... no taking of locks allowed here! ... if (etime) *etime = ktime_get(); preempt_enable_rt(); // On PREEMPT_RT kernel, otherwise omit. spin_unlock...(...); v2: Fix formatting of new multi-line code comments. Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-10-30 12:13:06 +08:00
* \param *stime Target location for timestamp taken immediately before
* scanout position query. Can be NULL to skip timestamp.
* \param *etime Target location for timestamp taken immediately after
* scanout position query. Can be NULL to skip timestamp.
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
*
* Returns vpos as a positive number while in active scanout area.
* Returns vpos as a negative number inside vblank, counting the number
* of scanlines to go until end of vblank, e.g., -1 means "one scanline
* until start of active scanout / end of vblank."
*
* \return Flags, or'ed together as follows:
*
* DRM_SCANOUTPOS_VALID = Query successful.
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
* DRM_SCANOUTPOS_INVBL = Inside vblank.
* DRM_SCANOUTPOS_ACCURATE = Returned position is accurate. A lack of
* this flag means that returned position may be offset by a constant
* but unknown small number of scanlines wrt. real scanout position.
*
*/
int (*get_scanout_position) (struct drm_device *dev, int crtc,
unsigned int flags,
drm: Push latency sensitive bits of vblank scanoutpos timestamping into kms drivers. A change in locking of some kms drivers (currently intel-kms) make the old approach too inaccurate and also incompatible with the PREEMPT_RT realtime kernel patchset. The driver->get_scanout_position() method of intel-kms now needs to aquire a spinlock, which clashes badly with the former preempt_disable() calls in the drm, and it also introduces larger delays and timing uncertainty on a contended lock than acceptable. This patch changes the prototype of driver->get_scanout_position() to require/allow kms drivers to perform the ktime_get() system time queries which go along with actual scanout position readout in a way that provides maximum precision and to return those timestamps to the drm. kms drivers implementations of get_scanout_position() are asked to implement timestamping and scanoutpos readout in a way that is as precise as possible and compatible with preempt_disable() on a PREMPT_RT kernel. A driver should follow this pattern in get_scanout_position() for precision and compatibility: spin_lock...(...); preempt_disable_rt(); // On a PREEMPT_RT kernel, otherwise omit. if (stime) *stime = ktime_get(); ... Minimum amount of MMIO register reads to get scanout position ... ... no taking of locks allowed here! ... if (etime) *etime = ktime_get(); preempt_enable_rt(); // On PREEMPT_RT kernel, otherwise omit. spin_unlock...(...); v2: Fix formatting of new multi-line code comments. Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-10-30 12:13:06 +08:00
int *vpos, int *hpos, ktime_t *stime,
ktime_t *etime);
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
/**
* Called by \c drm_get_last_vbltimestamp. Should return a precise
* timestamp when the most recent VBLANK interval ended or will end.
*
* Specifically, the timestamp in @vblank_time should correspond as
* closely as possible to the time when the first video scanline of
* the video frame after the end of VBLANK will start scanning out,
* the time immediately after end of the VBLANK interval. If the
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
* @crtc is currently inside VBLANK, this will be a time in the future.
* If the @crtc is currently scanning out a frame, this will be the
* past start time of the current scanout. This is meant to adhere
* to the OpenML OML_sync_control extension specification.
*
* \param dev dev DRM device handle.
* \param crtc crtc for which timestamp should be returned.
* \param *max_error Maximum allowable timestamp error in nanoseconds.
* Implementation should strive to provide timestamp
* with an error of at most *max_error nanoseconds.
* Returns true upper bound on error for timestamp.
* \param *vblank_time Target location for returned vblank timestamp.
* \param flags 0 = Defaults, no special treatment needed.
* \param DRM_CALLED_FROM_VBLIRQ = Function is called from vblank
* irq handler. Some drivers need to apply some workarounds
* for gpu-specific vblank irq quirks if flag is set.
*
* \returns
* Zero if timestamping isn't supported in current display mode or a
* negative number on failure. A positive status code on success,
* which describes how the vblank_time timestamp was computed.
*/
int (*get_vblank_timestamp) (struct drm_device *dev, int crtc,
int *max_error,
struct timeval *vblank_time,
unsigned flags);
/* these have to be filled in */
irqreturn_t(*irq_handler) (int irq, void *arg);
void (*irq_preinstall) (struct drm_device *dev);
int (*irq_postinstall) (struct drm_device *dev);
void (*irq_uninstall) (struct drm_device *dev);
/* Master routines */
int (*master_create)(struct drm_device *dev, struct drm_master *master);
void (*master_destroy)(struct drm_device *dev, struct drm_master *master);
/**
* master_set is called whenever the minor master is set.
* master_drop is called whenever the minor master is dropped.
*/
int (*master_set)(struct drm_device *dev, struct drm_file *file_priv,
bool from_open);
void (*master_drop)(struct drm_device *dev, struct drm_file *file_priv,
bool from_release);
int (*debugfs_init)(struct drm_minor *minor);
void (*debugfs_cleanup)(struct drm_minor *minor);
/**
* Driver-specific constructor for drm_gem_objects, to set up
* obj->driver_private.
*
* Returns 0 on success.
*/
void (*gem_free_object) (struct drm_gem_object *obj);
int (*gem_open_object) (struct drm_gem_object *, struct drm_file *);
void (*gem_close_object) (struct drm_gem_object *, struct drm_file *);
/* prime: */
/* export handle -> fd (see drm_gem_prime_handle_to_fd() helper) */
int (*prime_handle_to_fd)(struct drm_device *dev, struct drm_file *file_priv,
uint32_t handle, uint32_t flags, int *prime_fd);
/* import fd -> handle (see drm_gem_prime_fd_to_handle() helper) */
int (*prime_fd_to_handle)(struct drm_device *dev, struct drm_file *file_priv,
int prime_fd, uint32_t *handle);
/* export GEM -> dmabuf */
struct dma_buf * (*gem_prime_export)(struct drm_device *dev,
struct drm_gem_object *obj, int flags);
/* import dmabuf -> GEM */
struct drm_gem_object * (*gem_prime_import)(struct drm_device *dev,
struct dma_buf *dma_buf);
drm: add prime helpers Instead of reimplementing all of the dma_buf functionality in every driver, create helpers drm_prime_import and drm_prime_export that implement them in terms of new, lower-level hook functions: gem_prime_pin: callback when a buffer is created, used to pin buffers into GTT gem_prime_get_sg_table: convert a drm_gem_object to an sg_table for export gem_prime_import_sg_table: convert an sg_table into a drm_gem_object gem_prime_vmap, gem_prime_vunmap: map and unmap an object These hooks are optional; drivers can opt in by using drm_gem_prime_import and drm_gem_prime_export as the .gem_prime_import and .gem_prime_export fields of struct drm_driver. v2: - Drop .begin_cpu_access. None of the drivers this code replaces implemented it. Having it here was a leftover from when I was trying to include i915 in this rework. - Use mutex_lock instead of mutex_lock_interruptible, as these three drivers did. This patch series shouldn't change that behavior. - Rename helpers to gem_prime_get_sg_table and gem_prime_import_sg_table. Rename struct sg_table* variables to 'sgt' for clarity. - Update drm.tmpl for these new hooks. v3: - Pass the vaddr down to the driver. This lets drivers that just call vunmap on the pointer avoid having to store the pointer in their GEM private structures. - Move documentation into a /** DOC */ comment in drm_prime.c and include it in drm.tmpl with a !P line. I tried to use !F lines to include documentation of the individual functions from drmP.h, but the docproc / kernel-doc scripts barf on that file, so hopefully this is good enough for now. - apply refcount fix from commit be8a42ae60addd8b6092535c11b42d099d6470ec ("drm/prime: drop reference on imported dma-buf come from gem") Signed-off-by: Aaron Plattner <aplattner@nvidia.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: David Airlie <airlied@linux.ie> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-01-16 04:47:42 +08:00
/* low-level interface used by drm_gem_prime_{import,export} */
int (*gem_prime_pin)(struct drm_gem_object *obj);
void (*gem_prime_unpin)(struct drm_gem_object *obj);
struct reservation_object * (*gem_prime_res_obj)(
struct drm_gem_object *obj);
drm: add prime helpers Instead of reimplementing all of the dma_buf functionality in every driver, create helpers drm_prime_import and drm_prime_export that implement them in terms of new, lower-level hook functions: gem_prime_pin: callback when a buffer is created, used to pin buffers into GTT gem_prime_get_sg_table: convert a drm_gem_object to an sg_table for export gem_prime_import_sg_table: convert an sg_table into a drm_gem_object gem_prime_vmap, gem_prime_vunmap: map and unmap an object These hooks are optional; drivers can opt in by using drm_gem_prime_import and drm_gem_prime_export as the .gem_prime_import and .gem_prime_export fields of struct drm_driver. v2: - Drop .begin_cpu_access. None of the drivers this code replaces implemented it. Having it here was a leftover from when I was trying to include i915 in this rework. - Use mutex_lock instead of mutex_lock_interruptible, as these three drivers did. This patch series shouldn't change that behavior. - Rename helpers to gem_prime_get_sg_table and gem_prime_import_sg_table. Rename struct sg_table* variables to 'sgt' for clarity. - Update drm.tmpl for these new hooks. v3: - Pass the vaddr down to the driver. This lets drivers that just call vunmap on the pointer avoid having to store the pointer in their GEM private structures. - Move documentation into a /** DOC */ comment in drm_prime.c and include it in drm.tmpl with a !P line. I tried to use !F lines to include documentation of the individual functions from drmP.h, but the docproc / kernel-doc scripts barf on that file, so hopefully this is good enough for now. - apply refcount fix from commit be8a42ae60addd8b6092535c11b42d099d6470ec ("drm/prime: drop reference on imported dma-buf come from gem") Signed-off-by: Aaron Plattner <aplattner@nvidia.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: David Airlie <airlied@linux.ie> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-01-16 04:47:42 +08:00
struct sg_table *(*gem_prime_get_sg_table)(struct drm_gem_object *obj);
struct drm_gem_object *(*gem_prime_import_sg_table)(
struct drm_device *dev, size_t size,
struct sg_table *sgt);
void *(*gem_prime_vmap)(struct drm_gem_object *obj);
void (*gem_prime_vunmap)(struct drm_gem_object *obj, void *vaddr);
int (*gem_prime_mmap)(struct drm_gem_object *obj,
struct vm_area_struct *vma);
/* vga arb irq handler */
void (*vgaarb_irq)(struct drm_device *dev, bool state);
/* dumb alloc support */
int (*dumb_create)(struct drm_file *file_priv,
struct drm_device *dev,
struct drm_mode_create_dumb *args);
int (*dumb_map_offset)(struct drm_file *file_priv,
struct drm_device *dev, uint32_t handle,
uint64_t *offset);
int (*dumb_destroy)(struct drm_file *file_priv,
struct drm_device *dev,
uint32_t handle);
/* Driver private ops for this object */
const struct vm_operations_struct *gem_vm_ops;
int major;
int minor;
int patchlevel;
char *name;
char *desc;
char *date;
u32 driver_features;
int dev_priv_size;
const struct drm_ioctl_desc *ioctls;
int num_ioctls;
const struct file_operations *fops;
struct drm_bus *bus;
/* List of devices hanging off this driver with stealth attach. */
struct list_head legacy_dev_list;
};
enum drm_minor_type {
DRM_MINOR_LEGACY,
DRM_MINOR_CONTROL,
DRM_MINOR_RENDER,
DRM_MINOR_CNT,
};
/**
* Info file list entry. This structure represents a debugfs or proc file to
* be created by the drm core
*/
struct drm_info_list {
const char *name; /** file name */
int (*show)(struct seq_file*, void*); /** show callback */
u32 driver_features; /**< Required driver features for this entry */
void *data;
};
/**
* debugfs node structure. This structure represents a debugfs file.
*/
struct drm_info_node {
struct list_head list;
struct drm_minor *minor;
const struct drm_info_list *info_ent;
struct dentry *dent;
};
/**
* DRM minor structure. This structure represents a drm minor number.
*/
struct drm_minor {
int index; /**< Minor device number */
int type; /**< Control or render */
struct device *kdev; /**< Linux device */
struct drm_device *dev;
struct dentry *debugfs_root;
struct list_head debugfs_list;
struct mutex debugfs_lock; /* Protects debugfs_list. */
/* currently active master for this node. Protected by master_mutex */
struct drm_master *master;
struct drm_mode_group mode_group;
};
struct drm_pending_vblank_event {
struct drm_pending_event base;
int pipe;
struct drm_event_vblank event;
};
struct drm_vblank_crtc {
struct drm_device *dev; /* pointer to the drm_device */
wait_queue_head_t queue; /**< VBLANK wait queue */
struct timeval time[DRM_VBLANKTIME_RBSIZE]; /**< timestamp of current count */
struct timer_list disable_timer; /* delayed disable timer */
atomic_t count; /**< number of VBLANK interrupts */
atomic_t refcount; /* number of users of vblank interruptsper crtc */
u32 last; /* protected by dev->vbl_lock, used */
/* for wraparound handling */
u32 last_wait; /* Last vblank seqno waited per CRTC */
unsigned int inmodeset; /* Display driver is setting mode */
int crtc; /* crtc index */
bool enabled; /* so we don't call enable more than
once per disable */
};
/**
* DRM device structure. This structure represent a complete card that
* may contain multiple heads.
*/
struct drm_device {
struct list_head legacy_dev_list;/**< list of devices per driver for stealth attach cleanup */
int if_version; /**< Highest interface version set */
/** \name Lifetime Management */
/*@{ */
struct kref ref; /**< Object ref-count */
struct device *dev; /**< Device structure of bus-device */
struct drm_driver *driver; /**< DRM driver managing the device */
void *dev_private; /**< DRM driver private data */
struct drm_minor *control; /**< Control node */
struct drm_minor *primary; /**< Primary node */
struct drm_minor *render; /**< Render node */
atomic_t unplugged; /**< Flag whether dev is dead */
struct inode *anon_inode; /**< inode for private address-space */
char *unique; /**< unique name of the device */
/*@} */
/** \name Locks */
/*@{ */
struct mutex struct_mutex; /**< For others */
struct mutex master_mutex; /**< For drm_minor::master and drm_file::is_master */
/*@} */
/** \name Usage Counters */
/*@{ */
drm: Rip out totally bogus vga_switcheroo->can_switch locking So I just wanted to add a new field to struct drm_device and accidentally stumbled over something. According to comments dev->open_count is protected by dev->count_lock, but that's totally not the case. It's protected by drm_global_mutex. Unfortunately the vga switcheroo callbacks took this comment at face value. The problem is that we can't just take the drm_global_mutex because: - It would lead to a locking inversion with the driver load/unload paths. - It wouldn't actually protect anything, for that we'd need to wrap the entire vga switcheroo code in the drm_global_mutex. And I'm not sure whether that would actually solve anything. What we probably want is a try_to_grab_switcheroo reference kind of thing which is used in the driver's ->open callback. Then we could move all that ->can_switch madness into the vga switcheroo core where it really belongs. But since that would amount to real work take the easy way out and just add a comment. It's definitely not going to make anything worse since doing switcheroo state changes while restarting X just isn't recommended. Even though the delayed switching code does exactly that. v2: - Simplify the ->can_switch implementations more (Thierry) - Fix comment about the dev->open_count locking (Thierry) Cc: Thierry Reding <treding@nvidia.com> Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> (v1) Reviewed-by: Thierry Reding <treding@nvidia.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-11-04 03:46:34 +08:00
int open_count; /**< Outstanding files open, protected by drm_global_mutex. */
spinlock_t buf_lock; /**< For drm_device::buf_use and a few other things. */
int buf_use; /**< Buffers in use -- cannot alloc */
atomic_t buf_alloc; /**< Buffer allocation in progress */
/*@} */
struct list_head filelist;
/** \name Memory management */
/*@{ */
struct list_head maplist; /**< Linked list of regions */
struct drm_open_hash map_hash; /**< User token hash table for maps */
/** \name Context handle management */
/*@{ */
struct list_head ctxlist; /**< Linked list of context handles */
struct mutex ctxlist_mutex; /**< For ctxlist */
struct idr ctx_idr;
struct list_head vmalist; /**< List of vmas (for debugging) */
/*@} */
/** \name DMA support */
/*@{ */
struct drm_device_dma *dma; /**< Optional pointer for DMA support */
/*@} */
/** \name Context support */
/*@{ */
bool irq_enabled; /**< True if irq handler is enabled */
int irq;
__volatile__ long context_flag; /**< Context swapping flag */
int last_context; /**< Last current context */
/*@} */
/** \name VBLANK IRQ support */
/*@{ */
/*
* At load time, disabling the vblank interrupt won't be allowed since
* old clients may not call the modeset ioctl and therefore misbehave.
* Once the modeset ioctl *has* been called though, we can safely
* disable them when unused.
*/
bool vblank_disable_allowed;
/* array of size num_crtcs */
struct drm_vblank_crtc *vblank;
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
spinlock_t vblank_time_lock; /**< Protects vblank count and time updates during vblank enable/disable */
spinlock_t vbl_lock;
u32 max_vblank_count; /**< size of vblank counter register */
/**
* List of events
*/
struct list_head vblank_event_list;
spinlock_t event_lock;
/*@} */
struct drm_agp_head *agp; /**< AGP data */
struct pci_dev *pdev; /**< PCI device structure */
#ifdef __alpha__
struct pci_controller *hose;
#endif
struct platform_device *platformdev; /**< Platform device struture */
struct usb_device *usbdev;
struct drm_sg_mem *sg; /**< Scatter gather memory */
unsigned int num_crtcs; /**< Number of CRTCs on this device */
sigset_t sigmask;
struct {
int context;
struct drm_hw_lock *lock;
} sigdata;
struct drm_local_map *agp_buffer_map;
unsigned int agp_buffer_token;
struct drm_mode_config mode_config; /**< Current mode config */
/** \name GEM information */
/*@{ */
struct mutex object_name_lock;
struct idr object_name_idr;
struct drm_vma_offset_manager *vma_offset_manager;
/*@} */
int switch_power_state;
};
#define DRM_SWITCH_POWER_ON 0
#define DRM_SWITCH_POWER_OFF 1
#define DRM_SWITCH_POWER_CHANGING 2
#define DRM_SWITCH_POWER_DYNAMIC_OFF 3
static __inline__ int drm_core_check_feature(struct drm_device *dev,
int feature)
{
return ((dev->driver->driver_features & feature) ? 1 : 0);
}
static inline void drm_device_set_unplugged(struct drm_device *dev)
{
smp_wmb();
atomic_set(&dev->unplugged, 1);
}
static inline int drm_device_is_unplugged(struct drm_device *dev)
{
int ret = atomic_read(&dev->unplugged);
smp_rmb();
return ret;
}
static inline bool drm_is_render_client(const struct drm_file *file_priv)
drm: implement experimental render nodes Render nodes provide an API for userspace to use non-privileged GPU commands without any running DRM-Master. It is useful for offscreen rendering, GPGPU clients, and normal render clients which do not perform modesetting. Compared to legacy clients, render clients no longer need any authentication to perform client ioctls. Instead, user-space controls render/client access to GPUs via filesystem access-modes on the render-node. Once a render-node was opened, a client has full access to the client/render operations on the GPU. However, no modesetting or ioctls that affect global state are allowed on render nodes. To prevent privilege-escalation, drivers must explicitly state that they support render nodes. They must mark their render-only ioctls as DRM_RENDER_ALLOW so render clients can use them. Furthermore, they must support clients without any attached master. If filesystem access-modes are not enough for fine-grained access control to render nodes (very unlikely, considering the versaitlity of FS-ACLs), you may still fall-back to fd-passing from server to client (which allows arbitrary access-control). However, note that revoking access is currently impossible and unlikely to get implemented. Note: Render clients no longer have any associated DRM-Master as they are supposed to be independent of any server state. DRM core highly depends on file_priv->master to be non-NULL for modesetting/ctx/etc. commands. Therefore, drivers must be very careful to not require DRM-Master if they support DRIVER_RENDER. So far render-nodes are protected by "drm_rnodes". As long as this module-parameter is not set to 1, a driver will not create render nodes. This allows us to experiment with the API a bit before we stabilize it. v2: drop insecure GEM_FLINK to force use of dmabuf Signed-off-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-08-26 00:29:00 +08:00
{
return file_priv->minor->type == DRM_MINOR_RENDER;
}
static inline bool drm_is_control_client(const struct drm_file *file_priv)
{
return file_priv->minor->type == DRM_MINOR_CONTROL;
}
static inline bool drm_is_primary_client(const struct drm_file *file_priv)
{
return file_priv->minor->type == DRM_MINOR_LEGACY;
}
/******************************************************************/
/** \name Internal function definitions */
/*@{*/
/* Driver support (drm_drv.h) */
extern long drm_ioctl(struct file *filp,
unsigned int cmd, unsigned long arg);
extern long drm_compat_ioctl(struct file *filp,
unsigned int cmd, unsigned long arg);
extern int drm_lastclose(struct drm_device *dev);
extern bool drm_ioctl_flags(unsigned int nr, unsigned int *flags);
/* Device support (drm_fops.h) */
extern struct mutex drm_global_mutex;
extern int drm_open(struct inode *inode, struct file *filp);
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) */
extern int drm_mmap(struct file *filp, struct vm_area_struct *vma);
extern int drm_mmap_locked(struct file *filp, struct vm_area_struct *vma);
extern void drm_vm_open_locked(struct drm_device *dev, struct vm_area_struct *vma);
extern void drm_vm_close_locked(struct drm_device *dev, struct vm_area_struct *vma);
extern unsigned int drm_poll(struct file *filp, struct poll_table_struct *wait);
/* Misc. IOCTL support (drm_ioctl.h) */
extern int drm_irq_by_busid(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_getunique(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_setunique(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_getmap(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_getclient(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_getstats(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_getcap(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_setclientcap(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_setversion(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_noop(struct drm_device *dev, void *data,
struct drm_file *file_priv);
/* Authentication IOCTL support (drm_auth.h) */
extern int drm_getmagic(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_authmagic(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_remove_magic(struct drm_master *master, drm_magic_t magic);
/* Cache management (drm_cache.c) */
void drm_clflush_pages(struct page *pages[], unsigned long num_pages);
void drm_clflush_sg(struct sg_table *st);
void drm_clflush_virt_range(void *addr, unsigned long length);
/* Locking IOCTL support (drm_lock.h) */
extern int drm_lock(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_unlock(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_lock_free(struct drm_lock_data *lock_data, unsigned int context);
extern void drm_idlelock_take(struct drm_lock_data *lock_data);
extern void drm_idlelock_release(struct drm_lock_data *lock_data);
/*
* These are exported to drivers so that they can implement fencing using
* DMA quiscent + idle. DMA quiescent usually requires the hardware lock.
*/
extern int drm_i_have_hw_lock(struct drm_device *dev, struct drm_file *file_priv);
/* DMA support (drm_dma.h) */
extern int drm_legacy_dma_setup(struct drm_device *dev);
extern void drm_legacy_dma_takedown(struct drm_device *dev);
extern void drm_free_buffer(struct drm_device *dev, struct drm_buf * buf);
extern void drm_core_reclaim_buffers(struct drm_device *dev,
struct drm_file *filp);
/* IRQ support (drm_irq.h) */
extern int drm_control(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_irq_install(struct drm_device *dev, int irq);
extern int drm_irq_uninstall(struct drm_device *dev);
extern int drm_vblank_init(struct drm_device *dev, int num_crtcs);
extern int drm_wait_vblank(struct drm_device *dev, void *data,
struct drm_file *filp);
extern u32 drm_vblank_count(struct drm_device *dev, int crtc);
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
extern u32 drm_vblank_count_and_time(struct drm_device *dev, int crtc,
struct timeval *vblanktime);
extern void drm_send_vblank_event(struct drm_device *dev, int crtc,
struct drm_pending_vblank_event *e);
extern bool 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 int drm_crtc_vblank_get(struct drm_crtc *crtc);
extern void drm_crtc_vblank_put(struct drm_crtc *crtc);
extern void drm_wait_one_vblank(struct drm_device *dev, int crtc);
extern void drm_crtc_wait_one_vblank(struct drm_crtc *crtc);
extern void drm_vblank_off(struct drm_device *dev, int crtc);
extern void drm_vblank_on(struct drm_device *dev, int crtc);
extern void drm_crtc_vblank_off(struct drm_crtc *crtc);
extern void drm_crtc_vblank_on(struct drm_crtc *crtc);
extern void drm_vblank_cleanup(struct drm_device *dev);
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
extern u32 drm_get_last_vbltimestamp(struct drm_device *dev, int crtc,
struct timeval *tvblank, unsigned flags);
extern int drm_calc_vbltimestamp_from_scanoutpos(struct drm_device *dev,
int crtc, int *max_error,
struct timeval *vblank_time,
unsigned flags,
const struct drm_crtc *refcrtc,
const struct drm_display_mode *mode);
extern void drm_calc_timestamping_constants(struct drm_crtc *crtc,
const struct drm_display_mode *mode);
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
/**
* drm_crtc_vblank_waitqueue - get vblank waitqueue for the CRTC
* @crtc: which CRTC's vblank waitqueue to retrieve
*
* This function returns a pointer to the vblank waitqueue for the CRTC.
* Drivers can use this to implement vblank waits using wait_event() & co.
*/
static inline wait_queue_head_t *drm_crtc_vblank_waitqueue(struct drm_crtc *crtc)
{
return &crtc->dev->vblank[drm_crtc_index(crtc)].queue;
}
/* Modesetting support */
extern void drm_vblank_pre_modeset(struct drm_device *dev, int crtc);
extern void drm_vblank_post_modeset(struct drm_device *dev, int crtc);
extern int drm_modeset_ctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
/* AGP/GART support (drm_agpsupport.h) */
#include <drm/drm_agpsupport.h>
/* Stub support (drm_stub.h) */
extern int drm_setmaster_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_dropmaster_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
struct drm_master *drm_master_create(struct drm_minor *minor);
extern struct drm_master *drm_master_get(struct drm_master *master);
extern void drm_master_put(struct drm_master **master);
extern void drm_put_dev(struct drm_device *dev);
extern void drm_unplug_dev(struct drm_device *dev);
extern unsigned int drm_debug;
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
extern unsigned int drm_vblank_offdelay;
extern unsigned int drm_timestamp_precision;
extern unsigned int drm_timestamp_monotonic;
drm/vblank: Add support for precise vblank timestamping. The DRI2 swap & sync implementation needs precise vblank counts and precise timestamps corresponding to those vblank counts. For conformance to the OpenML OML_sync_control extension specification the DRM timestamp associated with a vblank count should correspond to the start of video scanout of the first scanline of the video frame following the vblank interval for that vblank count. Therefore we need to carry around precise timestamps for vblanks. Currently the DRM and KMS drivers generate timestamps ad-hoc via do_gettimeofday() in some places. The resulting timestamps are sometimes not very precise due to interrupt handling delays, they don't conform to OML_sync_control and some are wrong, as they aren't taken synchronized to the vblank. This patch implements support inside the drm core for precise and robust timestamping. It consists of the following interrelated pieces. 1. Vblank timestamp caching: A per-crtc ringbuffer stores the most recent vblank timestamps corresponding to vblank counts. The ringbuffer can be read out lock-free via the accessor function: struct timeval timestamp; vblankcount = drm_vblank_count_and_time(dev, crtcid, &timestamp). The function returns the current vblank count and the corresponding timestamp for start of video scanout following the vblank interval. It can be used anywhere between enclosing drm_vblank_get(dev, crtcid) and drm_vblank_put(dev,crtcid) statements. It is used inside the drmWaitVblank ioctl and in the vblank event queueing and handling. It should be used by kms drivers for timestamping of bufferswap completion. The timestamp ringbuffer is reinitialized each time vblank irq's get reenabled in drm_vblank_get()/ drm_update_vblank_count(). It is invalidated when vblank irq's get disabled. The ringbuffer is updated inside drm_handle_vblank() at each vblank irq. 2. Calculation of precise vblank timestamps: drm_get_last_vbltimestamp() is used to compute the timestamp for the end of the most recent vblank (if inside active scanout), or the expected end of the current vblank interval (if called inside a vblank interval). The function calls into a new optional kms driver entry point dev->driver->get_vblank_timestamp() which is supposed to provide the precise timestamp. If a kms driver doesn't implement the entry point or if the call fails, a simple do_gettimeofday() timestamp is returned as crude approximation of the true vblank time. A new drm module parameter drm.timestamp_precision_usec allows to disable high precision timestamps (if set to zero) or to specify the maximum acceptable error in the timestamps in microseconds. Kms drivers could implement their get_vblank_timestamp() function in a gpu specific way, as long as returned timestamps conform to OML_sync_control, e.g., by use of gpu specific hardware timestamps. Optionally, kms drivers can simply wrap and use the new utility function drm_calc_vbltimestamp_from_scanoutpos(). This function calls a new optional kms driver function dev->driver->get_scanout_position() which returns the current horizontal and vertical video scanout position of the crtc. The scanout position together with the drm_display_timing of the current video mode is used to calculate elapsed time relative to start of active scanout for the current video frame. This elapsed time is subtracted from the current do_gettimeofday() time to get the timestamp corresponding to start of video scanout. Currently non-interlaced, non-doublescan video modes, with or without panel scaling are handled correctly. Interlaced/ doublescan modes are tbd in a future patch. 3. Filtering of redundant vblank irq's and removal of some race-conditions in the vblank irq enable/disable path: Some gpu's (e.g., Radeon R500/R600) send spurious vblank irq's outside the vblank if vblank irq's get reenabled. These get detected by use of the vblank timestamps and filtered out to avoid miscounting of vblanks. Some race-conditions between the vblank irq enable/disable functions, the vblank irq handler and the gpu itself (updating its hardware vblank counter in the "wrong" moment) are fixed inside vblank_disable_and_save() and drm_update_vblank_count() by use of the vblank timestamps and a new spinlock dev->vblank_time_lock. The time until vblank irq disable is now configurable via a new drm module parameter drm.vblankoffdelay to allow experimentation with timeouts that are much shorter than the current 5 seconds and should allow longer vblank off periods for better power savings. Followup patches will use these new functions to implement precise timestamping for the intel and radeon kms drivers. Signed-off-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de> Signed-off-by: Dave Airlie <airlied@redhat.com>
2010-10-23 10:20:23 +08:00
extern struct class *drm_class;
/* Debugfs support */
#if defined(CONFIG_DEBUG_FS)
extern int drm_debugfs_init(struct drm_minor *minor, int minor_id,
struct dentry *root);
extern int drm_debugfs_create_files(const struct drm_info_list *files,
int count, struct dentry *root,
struct drm_minor *minor);
extern int drm_debugfs_remove_files(const struct drm_info_list *files,
int count, struct drm_minor *minor);
extern int drm_debugfs_cleanup(struct drm_minor *minor);
extern int drm_debugfs_connector_add(struct drm_connector *connector);
extern void drm_debugfs_connector_remove(struct drm_connector *connector);
#else
static inline int drm_debugfs_init(struct drm_minor *minor, int minor_id,
struct dentry *root)
{
return 0;
}
static inline int drm_debugfs_create_files(const struct drm_info_list *files,
int count, struct dentry *root,
struct drm_minor *minor)
{
return 0;
}
static inline int drm_debugfs_remove_files(const struct drm_info_list *files,
int count, struct drm_minor *minor)
{
return 0;
}
static inline int drm_debugfs_cleanup(struct drm_minor *minor)
{
return 0;
}
static inline int drm_debugfs_connector_add(struct drm_connector *connector)
{
return 0;
}
static inline void drm_debugfs_connector_remove(struct drm_connector *connector)
{
}
#endif
/* Info file support */
extern int drm_name_info(struct seq_file *m, void *data);
extern int drm_vm_info(struct seq_file *m, void *data);
extern int drm_bufs_info(struct seq_file *m, void *data);
extern int drm_vblank_info(struct seq_file *m, void *data);
extern int drm_clients_info(struct seq_file *m, void* data);
extern int drm_gem_name_info(struct seq_file *m, void *data);
drm: add prime helpers Instead of reimplementing all of the dma_buf functionality in every driver, create helpers drm_prime_import and drm_prime_export that implement them in terms of new, lower-level hook functions: gem_prime_pin: callback when a buffer is created, used to pin buffers into GTT gem_prime_get_sg_table: convert a drm_gem_object to an sg_table for export gem_prime_import_sg_table: convert an sg_table into a drm_gem_object gem_prime_vmap, gem_prime_vunmap: map and unmap an object These hooks are optional; drivers can opt in by using drm_gem_prime_import and drm_gem_prime_export as the .gem_prime_import and .gem_prime_export fields of struct drm_driver. v2: - Drop .begin_cpu_access. None of the drivers this code replaces implemented it. Having it here was a leftover from when I was trying to include i915 in this rework. - Use mutex_lock instead of mutex_lock_interruptible, as these three drivers did. This patch series shouldn't change that behavior. - Rename helpers to gem_prime_get_sg_table and gem_prime_import_sg_table. Rename struct sg_table* variables to 'sgt' for clarity. - Update drm.tmpl for these new hooks. v3: - Pass the vaddr down to the driver. This lets drivers that just call vunmap on the pointer avoid having to store the pointer in their GEM private structures. - Move documentation into a /** DOC */ comment in drm_prime.c and include it in drm.tmpl with a !P line. I tried to use !F lines to include documentation of the individual functions from drmP.h, but the docproc / kernel-doc scripts barf on that file, so hopefully this is good enough for now. - apply refcount fix from commit be8a42ae60addd8b6092535c11b42d099d6470ec ("drm/prime: drop reference on imported dma-buf come from gem") Signed-off-by: Aaron Plattner <aplattner@nvidia.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: David Airlie <airlied@linux.ie> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-01-16 04:47:42 +08:00
extern struct dma_buf *drm_gem_prime_export(struct drm_device *dev,
struct drm_gem_object *obj, int flags);
extern int drm_gem_prime_handle_to_fd(struct drm_device *dev,
struct drm_file *file_priv, uint32_t handle, uint32_t flags,
int *prime_fd);
drm: add prime helpers Instead of reimplementing all of the dma_buf functionality in every driver, create helpers drm_prime_import and drm_prime_export that implement them in terms of new, lower-level hook functions: gem_prime_pin: callback when a buffer is created, used to pin buffers into GTT gem_prime_get_sg_table: convert a drm_gem_object to an sg_table for export gem_prime_import_sg_table: convert an sg_table into a drm_gem_object gem_prime_vmap, gem_prime_vunmap: map and unmap an object These hooks are optional; drivers can opt in by using drm_gem_prime_import and drm_gem_prime_export as the .gem_prime_import and .gem_prime_export fields of struct drm_driver. v2: - Drop .begin_cpu_access. None of the drivers this code replaces implemented it. Having it here was a leftover from when I was trying to include i915 in this rework. - Use mutex_lock instead of mutex_lock_interruptible, as these three drivers did. This patch series shouldn't change that behavior. - Rename helpers to gem_prime_get_sg_table and gem_prime_import_sg_table. Rename struct sg_table* variables to 'sgt' for clarity. - Update drm.tmpl for these new hooks. v3: - Pass the vaddr down to the driver. This lets drivers that just call vunmap on the pointer avoid having to store the pointer in their GEM private structures. - Move documentation into a /** DOC */ comment in drm_prime.c and include it in drm.tmpl with a !P line. I tried to use !F lines to include documentation of the individual functions from drmP.h, but the docproc / kernel-doc scripts barf on that file, so hopefully this is good enough for now. - apply refcount fix from commit be8a42ae60addd8b6092535c11b42d099d6470ec ("drm/prime: drop reference on imported dma-buf come from gem") Signed-off-by: Aaron Plattner <aplattner@nvidia.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: David Airlie <airlied@linux.ie> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-01-16 04:47:42 +08:00
extern struct drm_gem_object *drm_gem_prime_import(struct drm_device *dev,
struct dma_buf *dma_buf);
extern int drm_gem_prime_fd_to_handle(struct drm_device *dev,
struct drm_file *file_priv, int prime_fd, uint32_t *handle);
extern void drm_gem_dmabuf_release(struct dma_buf *dma_buf);
extern int drm_prime_handle_to_fd_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_prime_fd_to_handle_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_prime_sg_to_page_addr_arrays(struct sg_table *sgt, struct page **pages,
dma_addr_t *addrs, int max_pages);
extern struct sg_table *drm_prime_pages_to_sg(struct page **pages, int nr_pages);
extern void drm_prime_gem_destroy(struct drm_gem_object *obj, struct sg_table *sg);
int drm_gem_dumb_destroy(struct drm_file *file,
struct drm_device *dev,
uint32_t handle);
void drm_prime_init_file_private(struct drm_prime_file_private *prime_fpriv);
void drm_prime_destroy_file_private(struct drm_prime_file_private *prime_fpriv);
drm/prime: Always add exported buffers to the handle cache ... not only when the dma-buf is freshly created. In contrived examples someone else could have exported/imported the dma-buf already and handed us the gem object with a flink name. If such on object gets reexported as a dma_buf we won't have it in the handle cache already, which breaks the guarantee that for dma-buf imports we always hand back an existing handle if there is one. This is exercised by igt/prime_self_import/with_one_bo_two_files Now if we extend the locked sections just a notch more we can also plug th racy buf/handle cache setup in handle_to_fd: If evil userspace races a concurrent gem close against a prime export operation we can end up tearing down the gem handle before the dma buf handle cache is set up. When handle_to_fd gets around to adding the handle to the cache there will be no one left to clean it up, effectily leaking the bo (and the dma-buf, since the handle cache holds a ref on the dma-buf): Thread A Thread B handle_to_fd: lookup gem object from handle creates new dma_buf gem_close on the same handle obj->dma_buf is set, but file priv buf handle cache has no entry obj->handle_count drops to 0 drm_prime_add_buf_handle sets up the handle cache -> We have a dma-buf reference in the handle cache, but since the handle_count of the gem object already dropped to 0 no on will clean it up. When closing the drm device fd we'll hit the WARN_ON in drm_prime_destroy_file_private. The important change is to extend the critical section of the filp->prime.lock to cover the gem handle lookup. This serializes with a concurrent gem handle close. This leak is exercised by igt/prime_self_import/export-vs-gem_close-race Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-08-15 06:02:49 +08:00
void drm_prime_remove_buf_handle_locked(struct drm_prime_file_private *prime_fpriv, struct dma_buf *dma_buf);
extern int drm_vma_info(struct seq_file *m, void *data);
/* Scatter Gather Support (drm_scatter.h) */
extern void drm_legacy_sg_cleanup(struct drm_device *dev);
extern int drm_sg_alloc(struct drm_device *dev, void *data,
struct drm_file *file_priv);
extern int drm_sg_free(struct drm_device *dev, void *data,
struct drm_file *file_priv);
/* ATI PCIGART support (ati_pcigart.h) */
extern int drm_ati_pcigart_init(struct drm_device *dev,
struct drm_ati_pcigart_info * gart_info);
extern int drm_ati_pcigart_cleanup(struct drm_device *dev,
struct drm_ati_pcigart_info * gart_info);
extern drm_dma_handle_t *drm_pci_alloc(struct drm_device *dev, size_t size,
size_t align);
extern void __drm_pci_free(struct drm_device *dev, drm_dma_handle_t * dmah);
extern void drm_pci_free(struct drm_device *dev, drm_dma_handle_t * dmah);
extern int drm_pci_set_unique(struct drm_device *dev,
struct drm_master *master,
struct drm_unique *u);
/* Legacy Support */
int drm_legacy_addmap(struct drm_device *d, resource_size_t offset,
unsigned int size, enum drm_map_type type,
enum drm_map_flags flags, struct drm_local_map **map_p);
int drm_legacy_rmmap(struct drm_device *d, struct drm_local_map *map);
int drm_legacy_rmmap_locked(struct drm_device *d, struct drm_local_map *map);
struct drm_local_map *drm_legacy_getsarea(struct drm_device *dev);
int drm_legacy_addbufs_agp(struct drm_device *d, struct drm_buf_desc *req);
int drm_legacy_addbufs_pci(struct drm_device *d, struct drm_buf_desc *req);
void drm_legacy_vma_flush(struct drm_device *d);
/* sysfs support (drm_sysfs.c) */
struct drm_sysfs_class;
extern struct class *drm_sysfs_create(struct module *owner, char *name);
extern void drm_sysfs_destroy(void);
extern struct device *drm_sysfs_minor_alloc(struct drm_minor *minor);
extern void drm_sysfs_hotplug_event(struct drm_device *dev);
extern int drm_sysfs_connector_add(struct drm_connector *connector);
extern void drm_sysfs_connector_remove(struct drm_connector *connector);
/* Graphics Execution Manager library functions (drm_gem.c) */
int drm_gem_init(struct drm_device *dev);
void drm_gem_destroy(struct drm_device *dev);
void drm_gem_object_release(struct drm_gem_object *obj);
void drm_gem_object_free(struct kref *kref);
int drm_gem_object_init(struct drm_device *dev,
struct drm_gem_object *obj, size_t size);
void drm_gem_private_object_init(struct drm_device *dev,
struct drm_gem_object *obj, size_t size);
void drm_gem_vm_open(struct vm_area_struct *vma);
void drm_gem_vm_close(struct vm_area_struct *vma);
int drm_gem_mmap_obj(struct drm_gem_object *obj, unsigned long obj_size,
struct vm_area_struct *vma);
int drm_gem_mmap(struct file *filp, struct vm_area_struct *vma);
#include <drm/drm_global.h>
static inline void
drm_gem_object_reference(struct drm_gem_object *obj)
{
kref_get(&obj->refcount);
}
static inline void
drm_gem_object_unreference(struct drm_gem_object *obj)
{
if (obj != NULL)
kref_put(&obj->refcount, drm_gem_object_free);
}
static inline void
drm_gem_object_unreference_unlocked(struct drm_gem_object *obj)
{
if (obj && !atomic_add_unless(&obj->refcount.refcount, -1, 1)) {
struct drm_device *dev = obj->dev;
mutex_lock(&dev->struct_mutex);
if (likely(atomic_dec_and_test(&obj->refcount.refcount)))
drm_gem_object_free(&obj->refcount);
mutex_unlock(&dev->struct_mutex);
}
}
drm/gem: completely close gem_open vs. gem_close races The gem flink name holds a reference onto the object itself, and this self-reference would prevent an flink'ed object from every being freed. To break that loop we remove the flink name when the last userspace handle disappears, i.e. when obj->handle_count reaches 0. Now in gem_open we drop the dev->object_name_lock between the flink name lookup and actually adding the handle. This means a concurrent gem_close of the last handle could result in the flink name getting reaped right inbetween, i.e. Thread 1 Thread 2 gem_open gem_close flink -> obj lookup handle_count drops to 0 remove flink name create_handle handle_count++ If someone now flinks this object again, we'll get a new flink name. We can close this race by removing the lock dropping and making the entire lookup+handle_create sequence atomic. Unfortunately to still be able to share the handle_create logic this requires a handle_create_tail function which drops the lock - we can't hold the object_name_lock while calling into a driver's ->gem_open callback. Note that for flink fixing this race isn't really important, since racing gem_open against gem_close is clearly a userspace bug. And no matter how the race ends, we won't leak any references. But with dma-buf where the userspace dma-buf fd itself is refcounted this is a valid sequence and hence we should fix it. Therefore this patch here is just a warm-up exercise (and for consistency between flink buffer sharing and dma-buf buffer sharing with self-imports). Also note that this extension of the critical section in gem_open protected by dev->object_name_lock only works because it's now a mutex: A spinlock would conflict with the potential memory allocation in idr_preload(). This is exercises by igt/gem_flink_race/flink_name. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Dave Airlie <airlied@redhat.com>
2013-08-15 06:02:45 +08:00
int drm_gem_handle_create_tail(struct drm_file *file_priv,
struct drm_gem_object *obj,
u32 *handlep);
int drm_gem_handle_create(struct drm_file *file_priv,
struct drm_gem_object *obj,
u32 *handlep);
int drm_gem_handle_delete(struct drm_file *filp, u32 handle);
void drm_gem_free_mmap_offset(struct drm_gem_object *obj);
int drm_gem_create_mmap_offset(struct drm_gem_object *obj);
int drm_gem_create_mmap_offset_size(struct drm_gem_object *obj, size_t size);
struct page **drm_gem_get_pages(struct drm_gem_object *obj);
void drm_gem_put_pages(struct drm_gem_object *obj, struct page **pages,
bool dirty, bool accessed);
struct drm_gem_object *drm_gem_object_lookup(struct drm_device *dev,
struct drm_file *filp,
u32 handle);
int drm_gem_close_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int drm_gem_flink_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int drm_gem_open_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
void drm_gem_open(struct drm_device *dev, struct drm_file *file_private);
void drm_gem_release(struct drm_device *dev, struct drm_file *file_private);
extern void drm_core_ioremap(struct drm_local_map *map, struct drm_device *dev);
extern void drm_core_ioremap_wc(struct drm_local_map *map, struct drm_device *dev);
extern void drm_core_ioremapfree(struct drm_local_map *map, struct drm_device *dev);
static __inline__ struct drm_local_map *drm_core_findmap(struct drm_device *dev,
unsigned int token)
{
struct drm_map_list *_entry;
list_for_each_entry(_entry, &dev->maplist, head)
if (_entry->user_token == token)
return _entry->map;
return NULL;
}
static __inline__ void drm_core_dropmap(struct drm_local_map *map)
{
}
#include <drm/drm_mem_util.h>
struct drm_device *drm_dev_alloc(struct drm_driver *driver,
struct device *parent);
void drm_dev_ref(struct drm_device *dev);
void drm_dev_unref(struct drm_device *dev);
int drm_dev_register(struct drm_device *dev, unsigned long flags);
void drm_dev_unregister(struct drm_device *dev);
int drm_dev_set_unique(struct drm_device *dev, const char *fmt, ...);
struct drm_minor *drm_minor_acquire(unsigned int minor_id);
void drm_minor_release(struct drm_minor *minor);
/*@}*/
/* PCI section */
static __inline__ int drm_pci_device_is_agp(struct drm_device *dev)
{
if (dev->driver->device_is_agp != NULL) {
int err = (*dev->driver->device_is_agp) (dev);
if (err != 2) {
return err;
}
}
return pci_find_capability(dev->pdev, PCI_CAP_ID_AGP);
}
void drm_pci_agp_destroy(struct drm_device *dev);
extern int drm_pci_init(struct drm_driver *driver, struct pci_driver *pdriver);
extern void drm_pci_exit(struct drm_driver *driver, struct pci_driver *pdriver);
extern int drm_get_pci_dev(struct pci_dev *pdev,
const struct pci_device_id *ent,
struct drm_driver *driver);
#define DRM_PCIE_SPEED_25 1
#define DRM_PCIE_SPEED_50 2
#define DRM_PCIE_SPEED_80 4
extern int drm_pcie_get_speed_cap_mask(struct drm_device *dev, u32 *speed_mask);
/* platform section */
extern int drm_platform_init(struct drm_driver *driver, struct platform_device *platform_device);
/* returns true if currently okay to sleep */
static __inline__ bool drm_can_sleep(void)
{
if (in_atomic() || in_dbg_master() || irqs_disabled())
return false;
return true;
}
#endif