drm/freedreno/freedreno_priv.h

260 lines
7.5 KiB
C
Raw Normal View History

freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
/* -*- mode: C; c-file-style: "k&r"; tab-width 4; indent-tabs-mode: t; -*- */
/*
* Copyright (C) 2012 Rob Clark <robclark@freedesktop.org>
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
* Authors:
* Rob Clark <robclark@freedesktop.org>
*/
#ifndef FREEDRENO_PRIV_H_
#define FREEDRENO_PRIV_H_
#include <stdlib.h>
#include <errno.h>
#include <string.h>
#include <unistd.h>
#include <errno.h>
#include <fcntl.h>
#include <sys/ioctl.h>
#include <pthread.h>
#include <stdio.h>
#include <assert.h>
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
#include "libdrm_macros.h"
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
#include "xf86drm.h"
#include "xf86atomic.h"
#include "util_double_list.h"
#include "util_math.h"
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
#include "freedreno_drmif.h"
#include "freedreno_ringbuffer.h"
#include "drm.h"
#ifndef TRUE
# define TRUE 1
#endif
#ifndef FALSE
# define FALSE 0
#endif
struct fd_device_funcs {
int (*bo_new_handle)(struct fd_device *dev, uint32_t size,
uint32_t flags, uint32_t *handle);
struct fd_bo * (*bo_from_handle)(struct fd_device *dev,
uint32_t size, uint32_t handle);
struct fd_pipe * (*pipe_new)(struct fd_device *dev, enum fd_pipe_id id,
unsigned prio);
void (*destroy)(struct fd_device *dev);
};
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
struct fd_bo_bucket {
uint32_t size;
struct list_head list;
};
struct fd_bo_cache {
struct fd_bo_bucket cache_bucket[14 * 4];
int num_buckets;
time_t time;
};
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
struct fd_device {
int fd;
enum fd_version version;
atomic_t refcnt;
/* tables to keep track of bo's, to avoid "evil-twin" fd_bo objects:
*
* handle_table: maps handle to fd_bo
* name_table: maps flink name to fd_bo
*
* We end up needing two tables, because DRM_IOCTL_GEM_OPEN always
* returns a new handle. So we need to figure out if the bo is already
* open in the process first, before calling gem-open.
*/
void *handle_table, *name_table;
const struct fd_device_funcs *funcs;
struct fd_bo_cache bo_cache;
struct fd_bo_cache ring_cache;
int closefd; /* call close(fd) upon destruction */
/* just for valgrind: */
int bo_size;
};
drm_private void fd_bo_cache_init(struct fd_bo_cache *cache, int coarse);
drm_private void fd_bo_cache_cleanup(struct fd_bo_cache *cache, time_t time);
drm_private struct fd_bo * fd_bo_cache_alloc(struct fd_bo_cache *cache,
uint32_t *size, uint32_t flags);
drm_private int fd_bo_cache_free(struct fd_bo_cache *cache, struct fd_bo *bo);
/* for where @table_lock is already held: */
drm_private void fd_device_del_locked(struct fd_device *dev);
struct fd_pipe_funcs {
struct fd_ringbuffer * (*ringbuffer_new)(struct fd_pipe *pipe, uint32_t size,
enum fd_ringbuffer_flags flags);
int (*get_param)(struct fd_pipe *pipe, enum fd_param_id param, uint64_t *value);
int (*wait)(struct fd_pipe *pipe, uint32_t timestamp, uint64_t timeout);
void (*destroy)(struct fd_pipe *pipe);
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
};
struct fd_pipe {
struct fd_device *dev;
enum fd_pipe_id id;
uint32_t gpu_id;
atomic_t refcnt;
const struct fd_pipe_funcs *funcs;
};
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
struct fd_ringbuffer_funcs {
void * (*hostptr)(struct fd_ringbuffer *ring);
int (*flush)(struct fd_ringbuffer *ring, uint32_t *last_start,
int in_fence_fd, int *out_fence_fd);
freedreno: support growable cmdstream buffers The issue that userspace needed to solve is that there is ~two orders of magnitude size difference in cmdstream buffers (both for gmem commands and for draw commands), and that the previous practice of allocating worst-case sizes is quite wasteful. Previously a submit would be constructed (for example) like: CMD TARGET DESCRIPTION g0 N gmem/tiling commands b0 Y binning commands d0 Y draw commands Which, after the one non-IB-target cmd buffer is inserted into the kernel controlled ringbuffer, looks like (not to scale): b0: d0: +-----+ +-----+ IB1 | ... | | ... | +-----+ +-----+ ^ ^ | | +-----+ +-+---------+ g0: | | | +----+----+----+----+----+----+---- IB0 | .. | IB | .. | IB | .. | IB | ... +----+----+----+----+----+----+---- ^ tile0 tile1 | +-----------+ userspace | ~~~~~~~~~~~~~~~~~~~|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ kernel | ----+----+---- ringbuffer ... | IB | ... ----+----+---- Now, multiple physical cmdstream buffers per fd_ringbuffer are supported, so this becomes: CMD TARGET DESCRIPTION g0 N ... N gmem/tiling commands gN N b0 Y ... Y binning commands bN Y d0 Y ... Y draw commands dN Y Which, after the non-IB-target cmd buffers (g0..gN) are inserted into the kernel controlled ringbuffer, looks like: b0: b1 d0: d1 +-----+ +-----+ +-----+ +-----+ IB1 | ... | | ... | ... | ... | | ... | ... +-----+ +-----+ +-----+ +-----+ ^ ^ ^ ^ | | | | | +-+ | +-----+------+ +-----+ | | | | | | +--+----------+ | g0: | | | | | | +----+----+----+----+----+----+---+----+----+---- IB0 | .. | IB | IB | .. | IB | IB |.. | IB | IB |... +----+----+----+----+----+----+---+----+----+---- ^ tile0 tile1 | to b0 to b1 | | | to|d0 to|d1 | | +----+ | +-+-----------+ | | | | | | | +------+ | +-+-------------+ | | g1: | | | | | | | +----+----+----+----+----+----+---+----+----+---- IB0 | | .. | IB | IB | .. | IB | IB |.. | IB | IB |... | +----+----+----+----+----+----+---+----+----+---- | ^ tileX tileY | | | +-----------+ +-----------+ | userspace | | ~~~~~~~~~~~~~~~~~~~|~~~~|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ kernel | | ----+----+----+---- ringbuffer ... | IB | IB | ... ----+----+----+---- Signed-off-by: Rob Clark <robclark@freedesktop.org>
2016-06-20 12:06:24 -06:00
void (*grow)(struct fd_ringbuffer *ring, uint32_t size);
void (*reset)(struct fd_ringbuffer *ring);
void (*emit_reloc)(struct fd_ringbuffer *ring,
const struct fd_reloc *reloc);
freedreno: support growable cmdstream buffers The issue that userspace needed to solve is that there is ~two orders of magnitude size difference in cmdstream buffers (both for gmem commands and for draw commands), and that the previous practice of allocating worst-case sizes is quite wasteful. Previously a submit would be constructed (for example) like: CMD TARGET DESCRIPTION g0 N gmem/tiling commands b0 Y binning commands d0 Y draw commands Which, after the one non-IB-target cmd buffer is inserted into the kernel controlled ringbuffer, looks like (not to scale): b0: d0: +-----+ +-----+ IB1 | ... | | ... | +-----+ +-----+ ^ ^ | | +-----+ +-+---------+ g0: | | | +----+----+----+----+----+----+---- IB0 | .. | IB | .. | IB | .. | IB | ... +----+----+----+----+----+----+---- ^ tile0 tile1 | +-----------+ userspace | ~~~~~~~~~~~~~~~~~~~|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ kernel | ----+----+---- ringbuffer ... | IB | ... ----+----+---- Now, multiple physical cmdstream buffers per fd_ringbuffer are supported, so this becomes: CMD TARGET DESCRIPTION g0 N ... N gmem/tiling commands gN N b0 Y ... Y binning commands bN Y d0 Y ... Y draw commands dN Y Which, after the non-IB-target cmd buffers (g0..gN) are inserted into the kernel controlled ringbuffer, looks like: b0: b1 d0: d1 +-----+ +-----+ +-----+ +-----+ IB1 | ... | | ... | ... | ... | | ... | ... +-----+ +-----+ +-----+ +-----+ ^ ^ ^ ^ | | | | | +-+ | +-----+------+ +-----+ | | | | | | +--+----------+ | g0: | | | | | | +----+----+----+----+----+----+---+----+----+---- IB0 | .. | IB | IB | .. | IB | IB |.. | IB | IB |... +----+----+----+----+----+----+---+----+----+---- ^ tile0 tile1 | to b0 to b1 | | | to|d0 to|d1 | | +----+ | +-+-----------+ | | | | | | | +------+ | +-+-------------+ | | g1: | | | | | | | +----+----+----+----+----+----+---+----+----+---- IB0 | | .. | IB | IB | .. | IB | IB |.. | IB | IB |... | +----+----+----+----+----+----+---+----+----+---- | ^ tileX tileY | | | +-----------+ +-----------+ | userspace | | ~~~~~~~~~~~~~~~~~~~|~~~~|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ kernel | | ----+----+----+---- ringbuffer ... | IB | IB | ... ----+----+----+---- Signed-off-by: Rob Clark <robclark@freedesktop.org>
2016-06-20 12:06:24 -06:00
uint32_t (*emit_reloc_ring)(struct fd_ringbuffer *ring,
struct fd_ringbuffer *target, uint32_t cmd_idx,
uint32_t submit_offset, uint32_t size);
freedreno: support growable cmdstream buffers The issue that userspace needed to solve is that there is ~two orders of magnitude size difference in cmdstream buffers (both for gmem commands and for draw commands), and that the previous practice of allocating worst-case sizes is quite wasteful. Previously a submit would be constructed (for example) like: CMD TARGET DESCRIPTION g0 N gmem/tiling commands b0 Y binning commands d0 Y draw commands Which, after the one non-IB-target cmd buffer is inserted into the kernel controlled ringbuffer, looks like (not to scale): b0: d0: +-----+ +-----+ IB1 | ... | | ... | +-----+ +-----+ ^ ^ | | +-----+ +-+---------+ g0: | | | +----+----+----+----+----+----+---- IB0 | .. | IB | .. | IB | .. | IB | ... +----+----+----+----+----+----+---- ^ tile0 tile1 | +-----------+ userspace | ~~~~~~~~~~~~~~~~~~~|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ kernel | ----+----+---- ringbuffer ... | IB | ... ----+----+---- Now, multiple physical cmdstream buffers per fd_ringbuffer are supported, so this becomes: CMD TARGET DESCRIPTION g0 N ... N gmem/tiling commands gN N b0 Y ... Y binning commands bN Y d0 Y ... Y draw commands dN Y Which, after the non-IB-target cmd buffers (g0..gN) are inserted into the kernel controlled ringbuffer, looks like: b0: b1 d0: d1 +-----+ +-----+ +-----+ +-----+ IB1 | ... | | ... | ... | ... | | ... | ... +-----+ +-----+ +-----+ +-----+ ^ ^ ^ ^ | | | | | +-+ | +-----+------+ +-----+ | | | | | | +--+----------+ | g0: | | | | | | +----+----+----+----+----+----+---+----+----+---- IB0 | .. | IB | IB | .. | IB | IB |.. | IB | IB |... +----+----+----+----+----+----+---+----+----+---- ^ tile0 tile1 | to b0 to b1 | | | to|d0 to|d1 | | +----+ | +-+-----------+ | | | | | | | +------+ | +-+-------------+ | | g1: | | | | | | | +----+----+----+----+----+----+---+----+----+---- IB0 | | .. | IB | IB | .. | IB | IB |.. | IB | IB |... | +----+----+----+----+----+----+---+----+----+---- | ^ tileX tileY | | | +-----------+ +-----------+ | userspace | | ~~~~~~~~~~~~~~~~~~~|~~~~|~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ kernel | | ----+----+----+---- ringbuffer ... | IB | IB | ... ----+----+----+---- Signed-off-by: Rob Clark <robclark@freedesktop.org>
2016-06-20 12:06:24 -06:00
uint32_t (*cmd_count)(struct fd_ringbuffer *ring);
void (*destroy)(struct fd_ringbuffer *ring);
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
};
struct fd_bo_funcs {
int (*offset)(struct fd_bo *bo, uint64_t *offset);
int (*cpu_prep)(struct fd_bo *bo, struct fd_pipe *pipe, uint32_t op);
void (*cpu_fini)(struct fd_bo *bo);
int (*madvise)(struct fd_bo *bo, int willneed);
uint64_t (*iova)(struct fd_bo *bo);
void (*destroy)(struct fd_bo *bo);
};
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
struct fd_bo {
struct fd_device *dev;
uint32_t size;
uint32_t handle;
uint32_t name;
void *map;
atomic_t refcnt;
const struct fd_bo_funcs *funcs;
enum {
NO_CACHE = 0,
BO_CACHE = 1,
RING_CACHE = 2,
} bo_reuse;
struct list_head list; /* bucket-list entry */
time_t free_time; /* time when added to bucket-list */
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
};
drm_private struct fd_bo *fd_bo_new_ring(struct fd_device *dev,
uint32_t size, uint32_t flags);
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
#define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]))
#define enable_debug 0 /* TODO make dynamic */
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
#define INFO_MSG(fmt, ...) \
do { drmMsg("[I] "fmt " (%s:%d)\n", \
##__VA_ARGS__, __FUNCTION__, __LINE__); } while (0)
#define DEBUG_MSG(fmt, ...) \
do if (enable_debug) { drmMsg("[D] "fmt " (%s:%d)\n", \
##__VA_ARGS__, __FUNCTION__, __LINE__); } while (0)
#define WARN_MSG(fmt, ...) \
do { drmMsg("[W] "fmt " (%s:%d)\n", \
##__VA_ARGS__, __FUNCTION__, __LINE__); } while (0)
#define ERROR_MSG(fmt, ...) \
do { drmMsg("[E] " fmt " (%s:%d)\n", \
##__VA_ARGS__, __FUNCTION__, __LINE__); } while (0)
#define U642VOID(x) ((void *)(unsigned long)(x))
#define VOID2U64(x) ((uint64_t)(unsigned long)(x))
static inline uint32_t
offset_bytes(void *end, void *start)
{
return ((char *)end) - ((char *)start);
}
#if HAVE_VALGRIND
# include <memcheck.h>
/*
* For tracking the backing memory (if valgrind enabled, we force a mmap
* for the purposes of tracking)
*/
static inline void VG_BO_ALLOC(struct fd_bo *bo)
{
if (bo && RUNNING_ON_VALGRIND) {
VALGRIND_MALLOCLIKE_BLOCK(fd_bo_map(bo), bo->size, 0, 1);
}
}
static inline void VG_BO_FREE(struct fd_bo *bo)
{
VALGRIND_FREELIKE_BLOCK(bo->map, 0);
}
/*
* For tracking bo structs that are in the buffer-cache, so that valgrind
* doesn't attribute ownership to the first one to allocate the recycled
* bo.
*
* Note that the list_head in fd_bo is used to track the buffers in cache
* so disable error reporting on the range while they are in cache so
* valgrind doesn't squawk about list traversal.
*
*/
static inline void VG_BO_RELEASE(struct fd_bo *bo)
{
if (RUNNING_ON_VALGRIND) {
VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(bo, bo->dev->bo_size);
VALGRIND_MAKE_MEM_NOACCESS(bo, bo->dev->bo_size);
VALGRIND_FREELIKE_BLOCK(bo->map, 0);
}
}
static inline void VG_BO_OBTAIN(struct fd_bo *bo)
{
if (RUNNING_ON_VALGRIND) {
VALGRIND_MAKE_MEM_DEFINED(bo, bo->dev->bo_size);
VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(bo, bo->dev->bo_size);
VALGRIND_MALLOCLIKE_BLOCK(bo->map, bo->size, 0, 1);
}
}
#else
static inline void VG_BO_ALLOC(struct fd_bo *bo) {}
static inline void VG_BO_FREE(struct fd_bo *bo) {}
static inline void VG_BO_RELEASE(struct fd_bo *bo) {}
static inline void VG_BO_OBTAIN(struct fd_bo *bo) {}
#endif
freedreno: add freedreno DRM The libdrm_freedreno helper layer for use by xf86-video-freedreno, fdre (freedreno r/e library and tests for driving gpu), and eventual gallium driver for the Adreno GPU. This uses the msm gpu driver from QCOM's android kernel tree. Note that current msm kernel driver is a bit strange. It provides a DRM interface for GEM, which is basically sufficient to have DRI2 working. But it does not provide KMS. And interface to 2d and 3d cores is via different other devices (/dev/kgsl-*). This is not quite how I'd write a DRM driver, but at this stage it is useful for xf86-video-freedreno and fdre (and eventual gallium driver) to be able to work on existing kernel driver from QCOM, to allow to capture cmdstream dumps from the binary blob drivers without having to reboot. So libdrm_freedreno attempts to hide most of the crazy. The intention is that when there is a proper kernel driver, it will be mostly just changes in libdrm_freedreno to adapt the gallium driver and xf86-video-freedreno (ignoring the fbdev->KMS changes). So don't look at freedreno as an example of how to write a libdrm module or a DRM driver.. it is just an attempt to paper over a non- standard kernel driver architecture. v1: original v2: hold ref's to pending bo's (because qcom's kernel driver doesn't), various bug fixes, add ringbuffer markers so we can emit IB's to portion of ringbuffer (so that gallium driver can use a single ringbuffer for both tile cmds and draw cmds. Signed-off-by: Rob Clark <robclark@freedesktop.org>
2012-10-07 17:57:31 -06:00
#endif /* FREEDRENO_PRIV_H_ */