From cdbc78ba702650b02b9e3e957dbaa725423bdf1e Mon Sep 17 00:00:00 2001
From: Jordan Crouse <jcrouse@codeaurora.org>
Date: Thu, 27 Jul 2017 10:42:35 -0600
Subject: drm/msm: Remove __user from __u64 data types

__user should be used to identify user pointers and not __u64
variables containing pointers.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
---
 include/uapi/drm/msm_drm.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

(limited to 'include/uapi')

diff --git a/include/uapi/drm/msm_drm.h b/include/uapi/drm/msm_drm.h
index 26c54f6d595d..ad4eb2863e70 100644
--- a/include/uapi/drm/msm_drm.h
+++ b/include/uapi/drm/msm_drm.h
@@ -171,7 +171,7 @@ struct drm_msm_gem_submit_cmd {
 	__u32 size;           /* in, cmdstream size */
 	__u32 pad;
 	__u32 nr_relocs;      /* in, number of submit_reloc's */
-	__u64 __user relocs;  /* in, ptr to array of submit_reloc's */
+	__u64 relocs;         /* in, ptr to array of submit_reloc's */
 };
 
 /* Each buffer referenced elsewhere in the cmdstream submit (ie. the
@@ -215,8 +215,8 @@ struct drm_msm_gem_submit {
 	__u32 fence;          /* out */
 	__u32 nr_bos;         /* in, number of submit_bo's */
 	__u32 nr_cmds;        /* in, number of submit_cmd's */
-	__u64 __user bos;     /* in, ptr to array of submit_bo's */
-	__u64 __user cmds;    /* in, ptr to array of submit_cmd's */
+	__u64 bos;            /* in, ptr to array of submit_bo's */
+	__u64 cmds;           /* in, ptr to array of submit_cmd's */
 	__s32 fence_fd;       /* in/out fence fd (see MSM_SUBMIT_FENCE_FD_IN/OUT) */
 };
 
-- 
cgit v1.2.3


From 3be8eddd9d58a925b461b582fa5aa422a9c145ee Mon Sep 17 00:00:00 2001
From: Eric Anholt <eric@anholt.net>
Date: Tue, 25 Jul 2017 09:27:33 -0700
Subject: drm/vc4: Add exec flags to allow forcing a specific X/Y tile walk
 order.

This is useful to allow GL to provide defined results for overlapping
glBlitFramebuffer, which X11 in turn uses to accelerate uncomposited
window movement without first blitting to a temporary.  x11perf
-copywinwin100 goes from 1850/sec to 4850/sec.

v2: Default to the same behavior as before when the flags aren't
    passed. (suggested by Boris)

Signed-off-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20170725162733.28007-2-eric@anholt.net
Reviewed-by: Boris Brezillon <boris.brezillon@free-electrons.com>
---
 drivers/gpu/drm/vc4/vc4_drv.c       |  1 +
 drivers/gpu/drm/vc4/vc4_gem.c       |  5 ++++-
 drivers/gpu/drm/vc4/vc4_render_cl.c | 21 ++++++++++++++++-----
 include/uapi/drm/vc4_drm.h          | 11 +++++++++++
 4 files changed, 32 insertions(+), 6 deletions(-)

(limited to 'include/uapi')

diff --git a/drivers/gpu/drm/vc4/vc4_drv.c b/drivers/gpu/drm/vc4/vc4_drv.c
index e8f0e1790d5e..1c96edcb302b 100644
--- a/drivers/gpu/drm/vc4/vc4_drv.c
+++ b/drivers/gpu/drm/vc4/vc4_drv.c
@@ -99,6 +99,7 @@ static int vc4_get_param_ioctl(struct drm_device *dev, void *data,
 	case DRM_VC4_PARAM_SUPPORTS_BRANCHES:
 	case DRM_VC4_PARAM_SUPPORTS_ETC1:
 	case DRM_VC4_PARAM_SUPPORTS_THREADED_FS:
+	case DRM_VC4_PARAM_SUPPORTS_FIXED_RCL_ORDER:
 		args->value = true;
 		break;
 	default:
diff --git a/drivers/gpu/drm/vc4/vc4_gem.c b/drivers/gpu/drm/vc4/vc4_gem.c
index 22a55cff7e64..d0c6bfb68c4e 100644
--- a/drivers/gpu/drm/vc4/vc4_gem.c
+++ b/drivers/gpu/drm/vc4/vc4_gem.c
@@ -1007,7 +1007,10 @@ vc4_submit_cl_ioctl(struct drm_device *dev, void *data,
 	struct ww_acquire_ctx acquire_ctx;
 	int ret = 0;
 
-	if ((args->flags & ~VC4_SUBMIT_CL_USE_CLEAR_COLOR) != 0) {
+	if ((args->flags & ~(VC4_SUBMIT_CL_USE_CLEAR_COLOR |
+			     VC4_SUBMIT_CL_FIXED_RCL_ORDER |
+			     VC4_SUBMIT_CL_RCL_ORDER_INCREASING_X |
+			     VC4_SUBMIT_CL_RCL_ORDER_INCREASING_Y)) != 0) {
 		DRM_DEBUG("Unknown flags: 0x%02x\n", args->flags);
 		return -EINVAL;
 	}
diff --git a/drivers/gpu/drm/vc4/vc4_render_cl.c b/drivers/gpu/drm/vc4/vc4_render_cl.c
index e0539731130b..273984f71ae2 100644
--- a/drivers/gpu/drm/vc4/vc4_render_cl.c
+++ b/drivers/gpu/drm/vc4/vc4_render_cl.c
@@ -261,8 +261,17 @@ static int vc4_create_rcl_bo(struct drm_device *dev, struct vc4_exec_info *exec,
 	uint8_t max_y_tile = args->max_y_tile;
 	uint8_t xtiles = max_x_tile - min_x_tile + 1;
 	uint8_t ytiles = max_y_tile - min_y_tile + 1;
-	uint8_t x, y;
+	uint8_t xi, yi;
 	uint32_t size, loop_body_size;
+	bool positive_x = true;
+	bool positive_y = true;
+
+	if (args->flags & VC4_SUBMIT_CL_FIXED_RCL_ORDER) {
+		if (!(args->flags & VC4_SUBMIT_CL_RCL_ORDER_INCREASING_X))
+			positive_x = false;
+		if (!(args->flags & VC4_SUBMIT_CL_RCL_ORDER_INCREASING_Y))
+			positive_y = false;
+	}
 
 	size = VC4_PACKET_TILE_RENDERING_MODE_CONFIG_SIZE;
 	loop_body_size = VC4_PACKET_TILE_COORDINATES_SIZE;
@@ -354,10 +363,12 @@ static int vc4_create_rcl_bo(struct drm_device *dev, struct vc4_exec_info *exec,
 	rcl_u16(setup, args->height);
 	rcl_u16(setup, args->color_write.bits);
 
-	for (y = min_y_tile; y <= max_y_tile; y++) {
-		for (x = min_x_tile; x <= max_x_tile; x++) {
-			bool first = (x == min_x_tile && y == min_y_tile);
-			bool last = (x == max_x_tile && y == max_y_tile);
+	for (yi = 0; yi < ytiles; yi++) {
+		int y = positive_y ? min_y_tile + yi : max_y_tile - yi;
+		for (xi = 0; xi < xtiles; xi++) {
+			int x = positive_x ? min_x_tile + xi : max_x_tile - xi;
+			bool first = (xi == 0 && yi == 0);
+			bool last = (xi == xtiles - 1 && yi == ytiles - 1);
 
 			emit_tile(exec, setup, x, y, first, last);
 		}
diff --git a/include/uapi/drm/vc4_drm.h b/include/uapi/drm/vc4_drm.h
index 551628e571f9..afae87004963 100644
--- a/include/uapi/drm/vc4_drm.h
+++ b/include/uapi/drm/vc4_drm.h
@@ -155,6 +155,16 @@ struct drm_vc4_submit_cl {
 	__u32 pad:24;
 
 #define VC4_SUBMIT_CL_USE_CLEAR_COLOR			(1 << 0)
+/* By default, the kernel gets to choose the order that the tiles are
+ * rendered in.  If this is set, then the tiles will be rendered in a
+ * raster order, with the right-to-left vs left-to-right and
+ * top-to-bottom vs bottom-to-top dictated by
+ * VC4_SUBMIT_CL_RCL_ORDER_INCREASING_*.  This allows overlapping
+ * blits to be implemented using the 3D engine.
+ */
+#define VC4_SUBMIT_CL_FIXED_RCL_ORDER			(1 << 1)
+#define VC4_SUBMIT_CL_RCL_ORDER_INCREASING_X		(1 << 2)
+#define VC4_SUBMIT_CL_RCL_ORDER_INCREASING_Y		(1 << 3)
 	__u32 flags;
 
 	/* Returned value of the seqno of this render job (for the
@@ -294,6 +304,7 @@ struct drm_vc4_get_hang_state {
 #define DRM_VC4_PARAM_SUPPORTS_BRANCHES		3
 #define DRM_VC4_PARAM_SUPPORTS_ETC1		4
 #define DRM_VC4_PARAM_SUPPORTS_THREADED_FS	5
+#define DRM_VC4_PARAM_SUPPORTS_FIXED_RCL_ORDER	6
 
 struct drm_vc4_get_param {
 	__u32 param;
-- 
cgit v1.2.3


From 6a1c9510694fe1e901a3b5b53386eac069adcea6 Mon Sep 17 00:00:00 2001
From: Moses Reuben <moses.reuben@amd.com>
Date: Tue, 15 Aug 2017 23:00:20 -0400
Subject: drm/amdkfd: Adding new IOCTL for scratch memory v2

v2:
* Renamed ALLOC_MEMORY_OF_SCRATCH to SET_SCRATCH_BACKING_VA
* Removed size parameter from the ioctl, it was unused
* Removed hole in ioctl number space
* No more call to write_config_static_mem
* Return correct error code from ioctl

Signed-off-by: Moses Reuben <moses.reuben@amd.com>
Signed-off-by: Ben Goz <ben.goz@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_chardev.c           | 37 ++++++++++++++++++++++
 .../gpu/drm/amd/amdkfd/kfd_device_queue_manager.c  |  3 ++
 .../drm/amd/amdkfd/kfd_device_queue_manager_cik.c  |  2 ++
 .../drm/amd/amdkfd/kfd_device_queue_manager_vi.c   |  2 ++
 drivers/gpu/drm/amd/amdkfd/kfd_priv.h              |  1 +
 include/uapi/linux/kfd_ioctl.h                     | 11 ++++++-
 6 files changed, 55 insertions(+), 1 deletion(-)

(limited to 'include/uapi')

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index 65b506f19b46..7436d34b77ab 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -848,6 +848,40 @@ static int kfd_ioctl_wait_events(struct file *filp, struct kfd_process *p,
 
 	return err;
 }
+static int kfd_ioctl_set_scratch_backing_va(struct file *filep,
+					struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_set_scratch_backing_va_args *args = data;
+	struct kfd_process_device *pdd;
+	struct kfd_dev *dev;
+	long err;
+
+	dev = kfd_device_by_id(args->gpu_id);
+	if (!dev)
+		return -EINVAL;
+
+	mutex_lock(&p->mutex);
+
+	pdd = kfd_bind_process_to_device(dev, p);
+	if (IS_ERR(pdd)) {
+		err = PTR_ERR(pdd);
+		goto bind_process_to_device_fail;
+	}
+
+	pdd->qpd.sh_hidden_private_base = args->va_addr;
+
+	mutex_unlock(&p->mutex);
+
+	if (sched_policy == KFD_SCHED_POLICY_NO_HWS && pdd->qpd.vmid != 0)
+		dev->kfd2kgd->set_scratch_backing_va(
+			dev->kgd, args->va_addr, pdd->qpd.vmid);
+
+	return 0;
+
+bind_process_to_device_fail:
+	mutex_unlock(&p->mutex);
+	return err;
+}
 
 #define AMDKFD_IOCTL_DEF(ioctl, _func, _flags) \
 	[_IOC_NR(ioctl)] = {.cmd = ioctl, .func = _func, .flags = _flags, \
@@ -902,6 +936,9 @@ static const struct amdkfd_ioctl_desc amdkfd_ioctls[] = {
 
 	AMDKFD_IOCTL_DEF(AMDKFD_IOC_DBG_WAVE_CONTROL,
 			kfd_ioctl_dbg_wave_control, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_SCRATCH_BACKING_VA,
+			kfd_ioctl_set_scratch_backing_va, 0),
 };
 
 #define AMDKFD_CORE_IOCTL_COUNT	ARRAY_SIZE(amdkfd_ioctls)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 618ac65b6136..53a66e821624 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -270,6 +270,9 @@ static int create_compute_queue_nocpsch(struct device_queue_manager *dqm,
 	pr_debug("Loading mqd to hqd on pipe %d, queue %d\n",
 			q->pipe, q->queue);
 
+	dqm->dev->kfd2kgd->set_scratch_backing_va(
+			dqm->dev->kgd, qpd->sh_hidden_private_base, qpd->vmid);
+
 	retval = mqd->load_mqd(mqd, q->mqd, q->pipe, q->queue, &q->properties,
 			       q->process->mm);
 	if (retval)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c
index fadc56a8be71..72c3cbabc0a7 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c
@@ -24,6 +24,7 @@
 #include "kfd_device_queue_manager.h"
 #include "cik_regs.h"
 #include "oss/oss_2_4_sh_mask.h"
+#include "gca/gfx_7_2_sh_mask.h"
 
 static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm,
 				   struct qcm_process_device *qpd,
@@ -123,6 +124,7 @@ static int register_process_cik(struct device_queue_manager *dqm,
 	} else {
 		temp = get_sh_mem_bases_nybble_64(pdd);
 		qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp);
+		qpd->sh_mem_config |= 1  << SH_MEM_CONFIG__PRIVATE_ATC__SHIFT;
 	}
 
 	pr_debug("is32bit process: %d sh_mem_bases nybble: 0x%X and register 0x%X\n",
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c
index 15e81ae9d2f4..40e9ddd096cd 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c
@@ -135,6 +135,8 @@ static int register_process_vi(struct device_queue_manager *dqm,
 		qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp);
 		qpd->sh_mem_config |= SH_MEM_ADDRESS_MODE_HSA64 <<
 			SH_MEM_CONFIG__ADDRESS_MODE__SHIFT;
+		qpd->sh_mem_config |= 1  <<
+			SH_MEM_CONFIG__PRIVATE_ATC__SHIFT;
 	}
 
 	pr_debug("is32bit process: %d sh_mem_bases nybble: 0x%X and register 0x%X\n",
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index 30ce92c6e6a1..b397ec726400 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -432,6 +432,7 @@ struct qcm_process_device {
 	uint32_t gds_size;
 	uint32_t num_gws;
 	uint32_t num_oac;
+	uint32_t sh_hidden_private_base;
 };
 
 /* Data that is per-process-per device. */
diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux/kfd_ioctl.h
index d6833426fdef..1b9c5609d523 100644
--- a/include/uapi/linux/kfd_ioctl.h
+++ b/include/uapi/linux/kfd_ioctl.h
@@ -232,6 +232,12 @@ struct kfd_ioctl_wait_events_args {
 	uint32_t wait_result;		/* from KFD */
 };
 
+struct kfd_ioctl_set_scratch_backing_va_args {
+	uint64_t va_addr;	/* to KFD */
+	uint32_t gpu_id;	/* to KFD */
+	uint32_t pad;
+};
+
 #define AMDKFD_IOCTL_BASE 'K'
 #define AMDKFD_IO(nr)			_IO(AMDKFD_IOCTL_BASE, nr)
 #define AMDKFD_IOR(nr, type)		_IOR(AMDKFD_IOCTL_BASE, nr, type)
@@ -286,7 +292,10 @@ struct kfd_ioctl_wait_events_args {
 #define AMDKFD_IOC_DBG_WAVE_CONTROL		\
 		AMDKFD_IOW(0x10, struct kfd_ioctl_dbg_wave_control_args)
 
+#define AMDKFD_IOC_SET_SCRATCH_BACKING_VA	\
+		AMDKFD_IOWR(0x11, struct kfd_ioctl_set_scratch_backing_va_args)
+
 #define AMDKFD_COMMAND_START		0x01
-#define AMDKFD_COMMAND_END		0x11
+#define AMDKFD_COMMAND_END		0x12
 
 #endif
-- 
cgit v1.2.3


From 5d71dbc3a588690c3d66d76db8cd29973425ce6d Mon Sep 17 00:00:00 2001
From: Yong Zhao <yong.zhao@amd.com>
Date: Tue, 15 Aug 2017 23:00:22 -0400
Subject: drm/amdkfd: Implement image tiling mode support v2

v2: Removed hole in ioctl number space

Signed-off-by: Yong Zhao <yong.zhao@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_chardev.c | 43 ++++++++++++++++++++++++++++++++
 include/uapi/linux/kfd_ioctl.h           | 28 ++++++++++++++++++++-
 2 files changed, 70 insertions(+), 1 deletion(-)

(limited to 'include/uapi')

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index 7436d34b77ab..e4a8c2e52cb2 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -883,6 +883,46 @@ bind_process_to_device_fail:
 	return err;
 }
 
+static int kfd_ioctl_get_tile_config(struct file *filep,
+		struct kfd_process *p, void *data)
+{
+	struct kfd_ioctl_get_tile_config_args *args = data;
+	struct kfd_dev *dev;
+	struct tile_config config;
+	int err = 0;
+
+	dev = kfd_device_by_id(args->gpu_id);
+
+	dev->kfd2kgd->get_tile_config(dev->kgd, &config);
+
+	args->gb_addr_config = config.gb_addr_config;
+	args->num_banks = config.num_banks;
+	args->num_ranks = config.num_ranks;
+
+	if (args->num_tile_configs > config.num_tile_configs)
+		args->num_tile_configs = config.num_tile_configs;
+	err = copy_to_user((void __user *)args->tile_config_ptr,
+			config.tile_config_ptr,
+			args->num_tile_configs * sizeof(uint32_t));
+	if (err) {
+		args->num_tile_configs = 0;
+		return -EFAULT;
+	}
+
+	if (args->num_macro_tile_configs > config.num_macro_tile_configs)
+		args->num_macro_tile_configs =
+				config.num_macro_tile_configs;
+	err = copy_to_user((void __user *)args->macro_tile_config_ptr,
+			config.macro_tile_config_ptr,
+			args->num_macro_tile_configs * sizeof(uint32_t));
+	if (err) {
+		args->num_macro_tile_configs = 0;
+		return -EFAULT;
+	}
+
+	return 0;
+}
+
 #define AMDKFD_IOCTL_DEF(ioctl, _func, _flags) \
 	[_IOC_NR(ioctl)] = {.cmd = ioctl, .func = _func, .flags = _flags, \
 			    .cmd_drv = 0, .name = #ioctl}
@@ -939,6 +979,9 @@ static const struct amdkfd_ioctl_desc amdkfd_ioctls[] = {
 
 	AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_SCRATCH_BACKING_VA,
 			kfd_ioctl_set_scratch_backing_va, 0),
+
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_TILE_CONFIG,
+			kfd_ioctl_get_tile_config, 0)
 };
 
 #define AMDKFD_CORE_IOCTL_COUNT	ARRAY_SIZE(amdkfd_ioctls)
diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux/kfd_ioctl.h
index 1b9c5609d523..7b4567bacfc2 100644
--- a/include/uapi/linux/kfd_ioctl.h
+++ b/include/uapi/linux/kfd_ioctl.h
@@ -238,6 +238,29 @@ struct kfd_ioctl_set_scratch_backing_va_args {
 	uint32_t pad;
 };
 
+struct kfd_ioctl_get_tile_config_args {
+	/* to KFD: pointer to tile array */
+	uint64_t tile_config_ptr;
+	/* to KFD: pointer to macro tile array */
+	uint64_t macro_tile_config_ptr;
+	/* to KFD: array size allocated by user mode
+	 * from KFD: array size filled by kernel
+	 */
+	uint32_t num_tile_configs;
+	/* to KFD: array size allocated by user mode
+	 * from KFD: array size filled by kernel
+	 */
+	uint32_t num_macro_tile_configs;
+
+	uint32_t gpu_id;		/* to KFD */
+	uint32_t gb_addr_config;	/* from KFD */
+	uint32_t num_banks;		/* from KFD */
+	uint32_t num_ranks;		/* from KFD */
+	/* struct size can be extended later if needed
+	 * without breaking ABI compatibility
+	 */
+};
+
 #define AMDKFD_IOCTL_BASE 'K'
 #define AMDKFD_IO(nr)			_IO(AMDKFD_IOCTL_BASE, nr)
 #define AMDKFD_IOR(nr, type)		_IOR(AMDKFD_IOCTL_BASE, nr, type)
@@ -295,7 +318,10 @@ struct kfd_ioctl_set_scratch_backing_va_args {
 #define AMDKFD_IOC_SET_SCRATCH_BACKING_VA	\
 		AMDKFD_IOWR(0x11, struct kfd_ioctl_set_scratch_backing_va_args)
 
+#define AMDKFD_IOC_GET_TILE_CONFIG                                      \
+		AMDKFD_IOWR(0x12, struct kfd_ioctl_get_tile_config_args)
+
 #define AMDKFD_COMMAND_START		0x01
-#define AMDKFD_COMMAND_END		0x12
+#define AMDKFD_COMMAND_END		0x13
 
 #endif
-- 
cgit v1.2.3


From f44d85389e17b2e960620c1c6d89bda978a11f2b Mon Sep 17 00:00:00 2001
From: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Date: Thu, 24 Aug 2017 16:08:14 +0100
Subject: drm: rename u32 in __u32 in uapi

All other fields use __

Cc: Ben Widawsky <ben@bwidawsk.net>
Fixes: db1689aa61b ("drm: Create a format/modifier blob")
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170824150814.5878-1-lionel.g.landwerlin@intel.com
---
 include/uapi/drm/drm_mode.h | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

(limited to 'include/uapi')

diff --git a/include/uapi/drm/drm_mode.h b/include/uapi/drm/drm_mode.h
index a2bb7161f020..54fc38c3c3f1 100644
--- a/include/uapi/drm/drm_mode.h
+++ b/include/uapi/drm/drm_mode.h
@@ -715,24 +715,24 @@ struct drm_mode_atomic {
 struct drm_format_modifier_blob {
 #define FORMAT_BLOB_CURRENT 1
 	/* Version of this blob format */
-	u32 version;
+	__u32 version;
 
 	/* Flags */
-	u32 flags;
+	__u32 flags;
 
 	/* Number of fourcc formats supported */
-	u32 count_formats;
+	__u32 count_formats;
 
 	/* Where in this blob the formats exist (in bytes) */
-	u32 formats_offset;
+	__u32 formats_offset;
 
 	/* Number of drm_format_modifiers */
-	u32 count_modifiers;
+	__u32 count_modifiers;
 
 	/* Where in this blob the modifiers exist (in bytes) */
-	u32 modifiers_offset;
+	__u32 modifiers_offset;
 
-	/* u32 formats[] */
+	/* __u32 formats[] */
 	/* struct drm_format_modifier modifiers[] */
 };
 
-- 
cgit v1.2.3


From 2cfa0bb25d25aa183ea29f1f9c2bc65f3f2c2264 Mon Sep 17 00:00:00 2001
From: Sinclair Yeh <syeh@vmware.com>
Date: Wed, 5 Jul 2017 01:37:55 -0700
Subject: drm/vmwgfx: Prepare to support fence fd

Make the fields and flags available.

Signed-off-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Deepak Singh Rawat <drawat@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
---
 drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c |  5 -----
 include/uapi/drm/vmwgfx_drm.h           | 11 ++++++++---
 2 files changed, 8 insertions(+), 8 deletions(-)

(limited to 'include/uapi')

diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
index 8c583fd16c79..178dabcdb198 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
@@ -4449,11 +4449,6 @@ int vmw_execbuf_ioctl(struct drm_device *dev, unsigned long data,
 		arg.context_handle = (uint32_t) -1;
 		break;
 	case 2:
-		if (arg.pad64 != 0) {
-			DRM_ERROR("Unused IOCTL data not set to zero.\n");
-			return -EINVAL;
-		}
-		break;
 	default:
 		break;
 	}
diff --git a/include/uapi/drm/vmwgfx_drm.h b/include/uapi/drm/vmwgfx_drm.h
index d9dfde9aa757..0bc784f5e0db 100644
--- a/include/uapi/drm/vmwgfx_drm.h
+++ b/include/uapi/drm/vmwgfx_drm.h
@@ -297,13 +297,17 @@ union drm_vmw_surface_reference_arg {
  * @version: Allows expanding the execbuf ioctl parameters without breaking
  * backwards compatibility, since user-space will always tell the kernel
  * which version it uses.
- * @flags: Execbuf flags. None currently.
+ * @flags: Execbuf flags.
+ * @imported_fence_fd:  FD for a fence imported from another device
  *
  * Argument to the DRM_VMW_EXECBUF Ioctl.
  */
 
 #define DRM_VMW_EXECBUF_VERSION 2
 
+#define DRM_VMW_EXECBUF_FLAG_IMPORT_FENCE_FD (1 << 0)
+#define DRM_VMW_EXECBUF_FLAG_EXPORT_FENCE_FD (1 << 1)
+
 struct drm_vmw_execbuf_arg {
 	__u64 commands;
 	__u32 command_size;
@@ -312,7 +316,7 @@ struct drm_vmw_execbuf_arg {
 	__u32 version;
 	__u32 flags;
 	__u32 context_handle;
-	__u32 pad64;
+	__s32 imported_fence_fd;
 };
 
 /**
@@ -328,6 +332,7 @@ struct drm_vmw_execbuf_arg {
  * @passed_seqno: The highest seqno number processed by the hardware
  * so far. This can be used to mark user-space fence objects as signaled, and
  * to determine whether a fence seqno might be stale.
+ * @fd: FD associated with the fence, -1 if not exported
  * @error: This member should've been set to -EFAULT on submission.
  * The following actions should be take on completion:
  * error == -EFAULT: Fence communication failed. The host is synchronized.
@@ -345,7 +350,7 @@ struct drm_vmw_fence_rep {
 	__u32 mask;
 	__u32 seqno;
 	__u32 passed_seqno;
-	__u32 pad64;
+	__s32 fd;
 	__s32 error;
 };
 
-- 
cgit v1.2.3


From 5e60a10eaebab93f823295cd7ec3848ba3b6e553 Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@redhat.com>
Date: Fri, 25 Aug 2017 10:52:22 -0700
Subject: drm/syncobj: add sync obj wait interface. (v8)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This interface will allow sync object to be used to back
Vulkan fences. This API is pretty much the vulkan fence waiting
API, and I've ported the code from amdgpu.

v2: accept relative timeout, pass remaining time back
to userspace.
v3: return to absolute timeouts.
v4: absolute zero = poll,
    rewrite any/all code to have same operation for arrays
    return -EINVAL for 0 fences.
v4.1: fixup fences allocation check, use u64_to_user_ptr
v5: move to sec/nsec, and use timespec64 for calcs.
v6: use -ETIME and drop the out status flag. (-ETIME
is suggested by ickle, I can feel a shed painting)
v7: talked to Daniel/Arnd, use ktime and ns everywhere.
v8: be more careful in the timeout calculations
    use uint32_t for counter variables so we don't overflow
    graciously handle -ENOINT being returned from dma_fence_wait_timeout

Signed-off-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/drm_internal.h |   2 +
 drivers/gpu/drm/drm_ioctl.c    |   2 +
 drivers/gpu/drm/drm_syncobj.c  | 142 +++++++++++++++++++++++++++++++++++++++++
 include/uapi/drm/drm.h         |  12 ++++
 4 files changed, 158 insertions(+)

(limited to 'include/uapi')

diff --git a/drivers/gpu/drm/drm_internal.h b/drivers/gpu/drm/drm_internal.h
index 4e906b82a170..534e5ac43bf8 100644
--- a/drivers/gpu/drm/drm_internal.h
+++ b/drivers/gpu/drm/drm_internal.h
@@ -167,3 +167,5 @@ int drm_syncobj_handle_to_fd_ioctl(struct drm_device *dev, void *data,
 				   struct drm_file *file_private);
 int drm_syncobj_fd_to_handle_ioctl(struct drm_device *dev, void *data,
 				   struct drm_file *file_private);
+int drm_syncobj_wait_ioctl(struct drm_device *dev, void *data,
+			   struct drm_file *file_private);
diff --git a/drivers/gpu/drm/drm_ioctl.c b/drivers/gpu/drm/drm_ioctl.c
index d920b2118a39..b4f443417a28 100644
--- a/drivers/gpu/drm/drm_ioctl.c
+++ b/drivers/gpu/drm/drm_ioctl.c
@@ -657,6 +657,8 @@ static const struct drm_ioctl_desc drm_ioctls[] = {
 		      DRM_UNLOCKED|DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF(DRM_IOCTL_SYNCOBJ_FD_TO_HANDLE, drm_syncobj_fd_to_handle_ioctl,
 		      DRM_UNLOCKED|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF(DRM_IOCTL_SYNCOBJ_WAIT, drm_syncobj_wait_ioctl,
+		      DRM_UNLOCKED|DRM_RENDER_ALLOW),
 };
 
 #define DRM_CORE_IOCTL_COUNT	ARRAY_SIZE( drm_ioctls )
diff --git a/drivers/gpu/drm/drm_syncobj.c b/drivers/gpu/drm/drm_syncobj.c
index eea38d82645c..4e8563c36d6e 100644
--- a/drivers/gpu/drm/drm_syncobj.c
+++ b/drivers/gpu/drm/drm_syncobj.c
@@ -1,5 +1,7 @@
 /*
  * Copyright 2017 Red Hat
+ * Parts ported from amdgpu (fence wait code).
+ * Copyright 2016 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -31,6 +33,9 @@
  * that contain an optional fence. The fence can be updated with a new
  * fence, or be NULL.
  *
+ * syncobj's can be waited upon, where it will wait for the underlying
+ * fence.
+ *
  * syncobj's can be export to fd's and back, these fd's are opaque and
  * have no other use case, except passing the syncobj between processes.
  *
@@ -447,3 +452,140 @@ drm_syncobj_fd_to_handle_ioctl(struct drm_device *dev, void *data,
 	return drm_syncobj_fd_to_handle(file_private, args->fd,
 					&args->handle);
 }
+
+/**
+ * drm_timeout_abs_to_jiffies - calculate jiffies timeout from absolute value
+ *
+ * @timeout_nsec: timeout nsec component in ns, 0 for poll
+ *
+ * Calculate the timeout in jiffies from an absolute time in sec/nsec.
+ */
+static signed long drm_timeout_abs_to_jiffies(int64_t timeout_nsec)
+{
+	ktime_t abs_timeout, now;
+	u64 timeout_ns, timeout_jiffies64;
+
+	/* make 0 timeout means poll - absolute 0 doesn't seem valid */
+	if (timeout_nsec == 0)
+		return 0;
+
+	abs_timeout = ns_to_ktime(timeout_nsec);
+	now = ktime_get();
+
+	if (!ktime_after(abs_timeout, now))
+		return 0;
+
+	timeout_ns = ktime_to_ns(ktime_sub(abs_timeout, now));
+
+	timeout_jiffies64 = nsecs_to_jiffies64(timeout_ns);
+	/*  clamp timeout to avoid infinite timeout */
+	if (timeout_jiffies64 >= MAX_SCHEDULE_TIMEOUT - 1)
+		return MAX_SCHEDULE_TIMEOUT - 1;
+
+	return timeout_jiffies64 + 1;
+}
+
+static int drm_syncobj_wait_fences(struct drm_device *dev,
+				   struct drm_file *file_private,
+				   struct drm_syncobj_wait *wait,
+				   struct dma_fence **fences)
+{
+	signed long timeout = drm_timeout_abs_to_jiffies(wait->timeout_nsec);
+	signed long ret = 0;
+	uint32_t first = ~0;
+
+	if (wait->flags & DRM_SYNCOBJ_WAIT_FLAGS_WAIT_ALL) {
+		uint32_t i;
+		for (i = 0; i < wait->count_handles; i++) {
+			ret = dma_fence_wait_timeout(fences[i], true, timeout);
+
+			/* Various dma_fence wait callbacks will return
+			 * ENOENT to indicate that the fence has already
+			 * been signaled.  We need to sanitize this to 0 so
+			 * we don't return early and the client doesn't see
+			 * an unexpected error.
+			 */
+			if (ret == -ENOENT)
+				ret = 0;
+
+			if (ret < 0)
+				return ret;
+			if (ret == 0)
+				break;
+			timeout = ret;
+		}
+		first = 0;
+	} else {
+		ret = dma_fence_wait_any_timeout(fences,
+						 wait->count_handles,
+						 true, timeout,
+						 &first);
+	}
+
+	if (ret < 0)
+		return ret;
+
+	wait->first_signaled = first;
+	if (ret == 0)
+		return -ETIME;
+	return 0;
+}
+
+int
+drm_syncobj_wait_ioctl(struct drm_device *dev, void *data,
+		       struct drm_file *file_private)
+{
+	struct drm_syncobj_wait *args = data;
+	uint32_t *handles;
+	struct dma_fence **fences;
+	int ret = 0;
+	uint32_t i;
+
+	if (!drm_core_check_feature(dev, DRIVER_SYNCOBJ))
+		return -ENODEV;
+
+	if (args->flags != 0 && args->flags != DRM_SYNCOBJ_WAIT_FLAGS_WAIT_ALL)
+		return -EINVAL;
+
+	if (args->count_handles == 0)
+		return -EINVAL;
+
+	/* Get the handles from userspace */
+	handles = kmalloc_array(args->count_handles, sizeof(uint32_t),
+				GFP_KERNEL);
+	if (handles == NULL)
+		return -ENOMEM;
+
+	if (copy_from_user(handles,
+			   u64_to_user_ptr(args->handles),
+			   sizeof(uint32_t) * args->count_handles)) {
+		ret = -EFAULT;
+		goto err_free_handles;
+	}
+
+	fences = kcalloc(args->count_handles,
+			 sizeof(struct dma_fence *), GFP_KERNEL);
+	if (!fences) {
+		ret = -ENOMEM;
+		goto err_free_handles;
+	}
+
+	for (i = 0; i < args->count_handles; i++) {
+		ret = drm_syncobj_find_fence(file_private, handles[i],
+					     &fences[i]);
+		if (ret)
+			goto err_free_fence_array;
+	}
+
+	ret = drm_syncobj_wait_fences(dev, file_private,
+				      args, fences);
+
+err_free_fence_array:
+	for (i = 0; i < args->count_handles; i++)
+		dma_fence_put(fences[i]);
+	kfree(fences);
+err_free_handles:
+	kfree(handles);
+
+	return ret;
+}
diff --git a/include/uapi/drm/drm.h b/include/uapi/drm/drm.h
index 101593ab10ac..0757c1a41821 100644
--- a/include/uapi/drm/drm.h
+++ b/include/uapi/drm/drm.h
@@ -718,6 +718,17 @@ struct drm_syncobj_handle {
 	__u32 pad;
 };
 
+#define DRM_SYNCOBJ_WAIT_FLAGS_WAIT_ALL (1 << 0)
+struct drm_syncobj_wait {
+	__u64 handles;
+	/* absolute timeout */
+	__s64 timeout_nsec;
+	__u32 count_handles;
+	__u32 flags;
+	__u32 first_signaled; /* only valid when not waiting all */
+	__u32 pad;
+};
+
 #if defined(__cplusplus)
 }
 #endif
@@ -840,6 +851,7 @@ extern "C" {
 #define DRM_IOCTL_SYNCOBJ_DESTROY	DRM_IOWR(0xC0, struct drm_syncobj_destroy)
 #define DRM_IOCTL_SYNCOBJ_HANDLE_TO_FD	DRM_IOWR(0xC1, struct drm_syncobj_handle)
 #define DRM_IOCTL_SYNCOBJ_FD_TO_HANDLE	DRM_IOWR(0xC2, struct drm_syncobj_handle)
+#define DRM_IOCTL_SYNCOBJ_WAIT		DRM_IOWR(0xC3, struct drm_syncobj_wait)
 
 /**
  * Device specific ioctls should only be in their respective headers
-- 
cgit v1.2.3


From 1fc08218ed2a42c86af5c905fe4c00885376a07e Mon Sep 17 00:00:00 2001
From: Jason Ekstrand <jason@jlekstrand.net>
Date: Fri, 25 Aug 2017 10:52:25 -0700
Subject: drm/syncobj: Add a CREATE_SIGNALED flag

This requests that the driver create the sync object such that it
already has a signaled dma_fence attached.  Because we don't need
anything in particular (just something signaled), we use a dummy null
fence.  This is useful for Vulkan which has a similar flag that can be
passed to vkCreateFence.

Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/drm_syncobj.c | 57 ++++++++++++++++++++++++++++++++++++++++---
 include/uapi/drm/drm.h        |  1 +
 2 files changed, 55 insertions(+), 3 deletions(-)

(limited to 'include/uapi')

diff --git a/drivers/gpu/drm/drm_syncobj.c b/drivers/gpu/drm/drm_syncobj.c
index bade497b3f1d..12db8c9564cd 100644
--- a/drivers/gpu/drm/drm_syncobj.c
+++ b/drivers/gpu/drm/drm_syncobj.c
@@ -154,6 +154,49 @@ void drm_syncobj_replace_fence(struct drm_syncobj *syncobj,
 }
 EXPORT_SYMBOL(drm_syncobj_replace_fence);
 
+struct drm_syncobj_null_fence {
+	struct dma_fence base;
+	spinlock_t lock;
+};
+
+static const char *drm_syncobj_null_fence_get_name(struct dma_fence *fence)
+{
+        return "syncobjnull";
+}
+
+static bool drm_syncobj_null_fence_enable_signaling(struct dma_fence *fence)
+{
+    dma_fence_enable_sw_signaling(fence);
+    return !dma_fence_is_signaled(fence);
+}
+
+static const struct dma_fence_ops drm_syncobj_null_fence_ops = {
+	.get_driver_name = drm_syncobj_null_fence_get_name,
+	.get_timeline_name = drm_syncobj_null_fence_get_name,
+	.enable_signaling = drm_syncobj_null_fence_enable_signaling,
+	.wait = dma_fence_default_wait,
+	.release = NULL,
+};
+
+static int drm_syncobj_assign_null_handle(struct drm_syncobj *syncobj)
+{
+	struct drm_syncobj_null_fence *fence;
+	fence = kzalloc(sizeof(*fence), GFP_KERNEL);
+	if (fence == NULL)
+		return -ENOMEM;
+
+	spin_lock_init(&fence->lock);
+	dma_fence_init(&fence->base, &drm_syncobj_null_fence_ops,
+		       &fence->lock, 0, 0);
+	dma_fence_signal(&fence->base);
+
+	drm_syncobj_replace_fence(syncobj, &fence->base);
+
+	dma_fence_put(&fence->base);
+
+	return 0;
+}
+
 int drm_syncobj_find_fence(struct drm_file *file_private,
 			   u32 handle,
 			   struct dma_fence **fence)
@@ -190,7 +233,7 @@ void drm_syncobj_free(struct kref *kref)
 EXPORT_SYMBOL(drm_syncobj_free);
 
 static int drm_syncobj_create(struct drm_file *file_private,
-			      u32 *handle)
+			      u32 *handle, uint32_t flags)
 {
 	int ret;
 	struct drm_syncobj *syncobj;
@@ -203,6 +246,14 @@ static int drm_syncobj_create(struct drm_file *file_private,
 	INIT_LIST_HEAD(&syncobj->cb_list);
 	spin_lock_init(&syncobj->lock);
 
+	if (flags & DRM_SYNCOBJ_CREATE_SIGNALED) {
+		ret = drm_syncobj_assign_null_handle(syncobj);
+		if (ret < 0) {
+			drm_syncobj_put(syncobj);
+			return ret;
+		}
+	}
+
 	idr_preload(GFP_KERNEL);
 	spin_lock(&file_private->syncobj_table_lock);
 	ret = idr_alloc(&file_private->syncobj_idr, syncobj, 1, 0, GFP_NOWAIT);
@@ -438,11 +489,11 @@ drm_syncobj_create_ioctl(struct drm_device *dev, void *data,
 		return -ENODEV;
 
 	/* no valid flags yet */
-	if (args->flags)
+	if (args->flags & ~DRM_SYNCOBJ_CREATE_SIGNALED)
 		return -EINVAL;
 
 	return drm_syncobj_create(file_private,
-				  &args->handle);
+				  &args->handle, args->flags);
 }
 
 int
diff --git a/include/uapi/drm/drm.h b/include/uapi/drm/drm.h
index 0757c1a41821..ade7f68d32b5 100644
--- a/include/uapi/drm/drm.h
+++ b/include/uapi/drm/drm.h
@@ -700,6 +700,7 @@ struct drm_prime_handle {
 
 struct drm_syncobj_create {
 	__u32 handle;
+#define DRM_SYNCOBJ_CREATE_SIGNALED (1 << 0)
 	__u32 flags;
 };
 
-- 
cgit v1.2.3


From e7aca5031a2fb51b6120864d0eff5478c95e6651 Mon Sep 17 00:00:00 2001
From: Jason Ekstrand <jason@jlekstrand.net>
Date: Fri, 25 Aug 2017 10:52:24 -0700
Subject: drm/syncobj: Allow wait for submit and signal behavior (v5)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Vulkan VkFence semantics require that the application be able to perform
a CPU wait on work which may not yet have been submitted.  This is
perfectly safe because the CPU wait has a timeout which will get
triggered eventually if no work is ever submitted.  This behavior is
advantageous for multi-threaded workloads because, so long as all of the
threads agree on what fences to use up-front, you don't have the extra
cross-thread synchronization cost of thread A telling thread B that it
has submitted its dependent work and thread B is now free to wait.

Within a single process, this can be implemented in the userspace driver
by doing exactly the same kind of tracking the app would have to do
using posix condition variables or similar.  However, in order for this
to work cross-process (as is required by VK_KHR_external_fence), we need
to handle this in the kernel.

This commit adds a WAIT_FOR_SUBMIT flag to DRM_IOCTL_SYNCOBJ_WAIT which
instructs the IOCTL to wait for the syncobj to have a non-null fence and
then wait on the fence.  Combined with DRM_IOCTL_SYNCOBJ_RESET, you can
easily get the Vulkan behavior.

v2:
 - Fix a bug in the invalid syncobj error path
 - Unify the wait-all and wait-any cases
v3:
 - Unify the timeout == 0 case a bit with the timeout > 0 case
 - Use wait_event_interruptible_timeout
v4:
 - Use proxy fence
v5:
 - Revert to a combination of v2 and v3
 - Don't use proxy fences
 - Don't use wait_event_interruptible_timeout because it just adds an
   extra layer of callbacks

Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Christian König <christian.koenig@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/drm_syncobj.c | 252 ++++++++++++++++++++++++++++++++++--------
 include/uapi/drm/drm.h        |   1 +
 2 files changed, 208 insertions(+), 45 deletions(-)

(limited to 'include/uapi')

diff --git a/drivers/gpu/drm/drm_syncobj.c b/drivers/gpu/drm/drm_syncobj.c
index 12db8c9564cd..cccd3bd194c6 100644
--- a/drivers/gpu/drm/drm_syncobj.c
+++ b/drivers/gpu/drm/drm_syncobj.c
@@ -51,6 +51,7 @@
 #include <linux/fs.h>
 #include <linux/anon_inodes.h>
 #include <linux/sync_file.h>
+#include <linux/sched/signal.h>
 
 #include "drm_internal.h"
 #include <drm/drm_syncobj.h>
@@ -88,6 +89,35 @@ static void drm_syncobj_add_callback_locked(struct drm_syncobj *syncobj,
 	list_add_tail(&cb->node, &syncobj->cb_list);
 }
 
+static int drm_syncobj_fence_get_or_add_callback(struct drm_syncobj *syncobj,
+						 struct dma_fence **fence,
+						 struct drm_syncobj_cb *cb,
+						 drm_syncobj_func_t func)
+{
+	int ret;
+
+	*fence = drm_syncobj_fence_get(syncobj);
+	if (*fence)
+		return 1;
+
+	spin_lock(&syncobj->lock);
+	/* We've already tried once to get a fence and failed.  Now that we
+	 * have the lock, try one more time just to be sure we don't add a
+	 * callback when a fence has already been set.
+	 */
+	if (syncobj->fence) {
+		*fence = dma_fence_get(syncobj->fence);
+		ret = 1;
+	} else {
+		*fence = NULL;
+		drm_syncobj_add_callback_locked(syncobj, cb, func);
+		ret = 0;
+	}
+	spin_unlock(&syncobj->lock);
+
+	return ret;
+}
+
 /**
  * drm_syncobj_add_callback - adds a callback to syncobj::cb_list
  * @syncobj: Sync object to which to add the callback
@@ -560,6 +590,160 @@ drm_syncobj_fd_to_handle_ioctl(struct drm_device *dev, void *data,
 					&args->handle);
 }
 
+struct syncobj_wait_entry {
+	struct task_struct *task;
+	struct dma_fence *fence;
+	struct dma_fence_cb fence_cb;
+	struct drm_syncobj_cb syncobj_cb;
+};
+
+static void syncobj_wait_fence_func(struct dma_fence *fence,
+				    struct dma_fence_cb *cb)
+{
+	struct syncobj_wait_entry *wait =
+		container_of(cb, struct syncobj_wait_entry, fence_cb);
+
+	wake_up_process(wait->task);
+}
+
+static void syncobj_wait_syncobj_func(struct drm_syncobj *syncobj,
+				      struct drm_syncobj_cb *cb)
+{
+	struct syncobj_wait_entry *wait =
+		container_of(cb, struct syncobj_wait_entry, syncobj_cb);
+
+	/* This happens inside the syncobj lock */
+	wait->fence = dma_fence_get(syncobj->fence);
+	wake_up_process(wait->task);
+}
+
+static signed long drm_syncobj_array_wait_timeout(struct drm_syncobj **syncobjs,
+						  uint32_t count,
+						  uint32_t flags,
+						  signed long timeout,
+						  uint32_t *idx)
+{
+	struct syncobj_wait_entry *entries;
+	struct dma_fence *fence;
+	signed long ret;
+	uint32_t signaled_count, i;
+
+	entries = kcalloc(count, sizeof(*entries), GFP_KERNEL);
+	if (!entries)
+		return -ENOMEM;
+
+	/* Walk the list of sync objects and initialize entries.  We do
+	 * this up-front so that we can properly return -EINVAL if there is
+	 * a syncobj with a missing fence and then never have the chance of
+	 * returning -EINVAL again.
+	 */
+	signaled_count = 0;
+	for (i = 0; i < count; ++i) {
+		entries[i].task = current;
+		entries[i].fence = drm_syncobj_fence_get(syncobjs[i]);
+		if (!entries[i].fence) {
+			if (flags & DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT) {
+				continue;
+			} else {
+				ret = -EINVAL;
+				goto cleanup_entries;
+			}
+		}
+
+		if (dma_fence_is_signaled(entries[i].fence)) {
+			if (signaled_count == 0 && idx)
+				*idx = i;
+			signaled_count++;
+		}
+	}
+
+	/* Initialize ret to the max of timeout and 1.  That way, the
+	 * default return value indicates a successful wait and not a
+	 * timeout.
+	 */
+	ret = max_t(signed long, timeout, 1);
+
+	if (signaled_count == count ||
+	    (signaled_count > 0 &&
+	     !(flags & DRM_SYNCOBJ_WAIT_FLAGS_WAIT_ALL)))
+		goto cleanup_entries;
+
+	/* There's a very annoying laxness in the dma_fence API here, in
+	 * that backends are not required to automatically report when a
+	 * fence is signaled prior to fence->ops->enable_signaling() being
+	 * called.  So here if we fail to match signaled_count, we need to
+	 * fallthough and try a 0 timeout wait!
+	 */
+
+	if (flags & DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT) {
+		for (i = 0; i < count; ++i) {
+			drm_syncobj_fence_get_or_add_callback(syncobjs[i],
+							      &entries[i].fence,
+							      &entries[i].syncobj_cb,
+							      syncobj_wait_syncobj_func);
+		}
+	}
+
+	do {
+		set_current_state(TASK_INTERRUPTIBLE);
+
+		signaled_count = 0;
+		for (i = 0; i < count; ++i) {
+			fence = entries[i].fence;
+			if (!fence)
+				continue;
+
+			if (dma_fence_is_signaled(fence) ||
+			    (!entries[i].fence_cb.func &&
+			     dma_fence_add_callback(fence,
+						    &entries[i].fence_cb,
+						    syncobj_wait_fence_func))) {
+				/* The fence has been signaled */
+				if (flags & DRM_SYNCOBJ_WAIT_FLAGS_WAIT_ALL) {
+					signaled_count++;
+				} else {
+					if (idx)
+						*idx = i;
+					goto done_waiting;
+				}
+			}
+		}
+
+		if (signaled_count == count)
+			goto done_waiting;
+
+		if (timeout == 0) {
+			/* If we are doing a 0 timeout wait and we got
+			 * here, then we just timed out.
+			 */
+			ret = 0;
+			goto done_waiting;
+		}
+
+		ret = schedule_timeout(ret);
+
+		if (ret > 0 && signal_pending(current))
+			ret = -ERESTARTSYS;
+	} while (ret > 0);
+
+done_waiting:
+	__set_current_state(TASK_RUNNING);
+
+cleanup_entries:
+	for (i = 0; i < count; ++i) {
+		if (entries[i].syncobj_cb.func)
+			drm_syncobj_remove_callback(syncobjs[i],
+						    &entries[i].syncobj_cb);
+		if (entries[i].fence_cb.func)
+			dma_fence_remove_callback(entries[i].fence,
+						  &entries[i].fence_cb);
+		dma_fence_put(entries[i].fence);
+	}
+	kfree(entries);
+
+	return ret;
+}
+
 /**
  * drm_timeout_abs_to_jiffies - calculate jiffies timeout from absolute value
  *
@@ -592,43 +776,19 @@ static signed long drm_timeout_abs_to_jiffies(int64_t timeout_nsec)
 	return timeout_jiffies64 + 1;
 }
 
-static int drm_syncobj_wait_fences(struct drm_device *dev,
-				   struct drm_file *file_private,
-				   struct drm_syncobj_wait *wait,
-				   struct dma_fence **fences)
+static int drm_syncobj_array_wait(struct drm_device *dev,
+				  struct drm_file *file_private,
+				  struct drm_syncobj_wait *wait,
+				  struct drm_syncobj **syncobjs)
 {
 	signed long timeout = drm_timeout_abs_to_jiffies(wait->timeout_nsec);
 	signed long ret = 0;
 	uint32_t first = ~0;
 
-	if (wait->flags & DRM_SYNCOBJ_WAIT_FLAGS_WAIT_ALL) {
-		uint32_t i;
-		for (i = 0; i < wait->count_handles; i++) {
-			ret = dma_fence_wait_timeout(fences[i], true, timeout);
-
-			/* Various dma_fence wait callbacks will return
-			 * ENOENT to indicate that the fence has already
-			 * been signaled.  We need to sanitize this to 0 so
-			 * we don't return early and the client doesn't see
-			 * an unexpected error.
-			 */
-			if (ret == -ENOENT)
-				ret = 0;
-
-			if (ret < 0)
-				return ret;
-			if (ret == 0)
-				break;
-			timeout = ret;
-		}
-		first = 0;
-	} else {
-		ret = dma_fence_wait_any_timeout(fences,
-						 wait->count_handles,
-						 true, timeout,
-						 &first);
-	}
-
+	ret = drm_syncobj_array_wait_timeout(syncobjs,
+					     wait->count_handles,
+					     wait->flags,
+					     timeout, &first);
 	if (ret < 0)
 		return ret;
 
@@ -644,14 +804,15 @@ drm_syncobj_wait_ioctl(struct drm_device *dev, void *data,
 {
 	struct drm_syncobj_wait *args = data;
 	uint32_t *handles;
-	struct dma_fence **fences;
+	struct drm_syncobj **syncobjs;
 	int ret = 0;
 	uint32_t i;
 
 	if (!drm_core_check_feature(dev, DRIVER_SYNCOBJ))
 		return -ENODEV;
 
-	if (args->flags != 0 && args->flags != DRM_SYNCOBJ_WAIT_FLAGS_WAIT_ALL)
+	if (args->flags & ~(DRM_SYNCOBJ_WAIT_FLAGS_WAIT_ALL |
+			    DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT))
 		return -EINVAL;
 
 	if (args->count_handles == 0)
@@ -670,27 +831,28 @@ drm_syncobj_wait_ioctl(struct drm_device *dev, void *data,
 		goto err_free_handles;
 	}
 
-	fences = kcalloc(args->count_handles,
-			 sizeof(struct dma_fence *), GFP_KERNEL);
-	if (!fences) {
+	syncobjs = kcalloc(args->count_handles,
+			   sizeof(struct drm_syncobj *), GFP_KERNEL);
+	if (!syncobjs) {
 		ret = -ENOMEM;
 		goto err_free_handles;
 	}
 
 	for (i = 0; i < args->count_handles; i++) {
-		ret = drm_syncobj_find_fence(file_private, handles[i],
-					     &fences[i]);
-		if (ret)
+		syncobjs[i] = drm_syncobj_find(file_private, handles[i]);
+		if (!syncobjs[i]) {
+			ret = -ENOENT;
 			goto err_free_fence_array;
+		}
 	}
 
-	ret = drm_syncobj_wait_fences(dev, file_private,
-				      args, fences);
+	ret = drm_syncobj_array_wait(dev, file_private,
+				     args, syncobjs);
 
 err_free_fence_array:
-	for (i = 0; i < args->count_handles; i++)
-		dma_fence_put(fences[i]);
-	kfree(fences);
+	while (i-- > 0)
+		drm_syncobj_put(syncobjs[i]);
+	kfree(syncobjs);
 err_free_handles:
 	kfree(handles);
 
diff --git a/include/uapi/drm/drm.h b/include/uapi/drm/drm.h
index ade7f68d32b5..4c746597225e 100644
--- a/include/uapi/drm/drm.h
+++ b/include/uapi/drm/drm.h
@@ -720,6 +720,7 @@ struct drm_syncobj_handle {
 };
 
 #define DRM_SYNCOBJ_WAIT_FLAGS_WAIT_ALL (1 << 0)
+#define DRM_SYNCOBJ_WAIT_FLAGS_WAIT_FOR_SUBMIT (1 << 1)
 struct drm_syncobj_wait {
 	__u64 handles;
 	/* absolute timeout */
-- 
cgit v1.2.3


From aa4035d2c7683d2f2fb0ffe8087abd9eabf6d54a Mon Sep 17 00:00:00 2001
From: Jason Ekstrand <jason@jlekstrand.net>
Date: Mon, 28 Aug 2017 14:10:27 -0700
Subject: drm/syncobj: Add a reset ioctl (v3)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This just resets the dma_fence to NULL so it looks like it's never been
signaled.  This will be useful once we add the new wait API for allowing
wait on "submit and signal" behavior.

v2:
 - Take an array of sync objects (Dave Airlie)
v3:
 - Throw -EINVAL if pad != 0

Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Christian König <christian.koenig@amd.com> (v1)
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/drm_internal.h |  2 ++
 drivers/gpu/drm/drm_ioctl.c    |  2 ++
 drivers/gpu/drm/drm_syncobj.c  | 33 +++++++++++++++++++++++++++++++++
 include/uapi/drm/drm.h         |  7 +++++++
 4 files changed, 44 insertions(+)

(limited to 'include/uapi')

diff --git a/drivers/gpu/drm/drm_internal.h b/drivers/gpu/drm/drm_internal.h
index 534e5ac43bf8..83f1615eb1ec 100644
--- a/drivers/gpu/drm/drm_internal.h
+++ b/drivers/gpu/drm/drm_internal.h
@@ -169,3 +169,5 @@ int drm_syncobj_fd_to_handle_ioctl(struct drm_device *dev, void *data,
 				   struct drm_file *file_private);
 int drm_syncobj_wait_ioctl(struct drm_device *dev, void *data,
 			   struct drm_file *file_private);
+int drm_syncobj_reset_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *file_private);
diff --git a/drivers/gpu/drm/drm_ioctl.c b/drivers/gpu/drm/drm_ioctl.c
index b4f443417a28..16c5d51a43aa 100644
--- a/drivers/gpu/drm/drm_ioctl.c
+++ b/drivers/gpu/drm/drm_ioctl.c
@@ -659,6 +659,8 @@ static const struct drm_ioctl_desc drm_ioctls[] = {
 		      DRM_UNLOCKED|DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF(DRM_IOCTL_SYNCOBJ_WAIT, drm_syncobj_wait_ioctl,
 		      DRM_UNLOCKED|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF(DRM_IOCTL_SYNCOBJ_RESET, drm_syncobj_reset_ioctl,
+		      DRM_UNLOCKED|DRM_RENDER_ALLOW),
 };
 
 #define DRM_CORE_IOCTL_COUNT	ARRAY_SIZE( drm_ioctls )
diff --git a/drivers/gpu/drm/drm_syncobj.c b/drivers/gpu/drm/drm_syncobj.c
index 15e74ca61760..40d2ad293661 100644
--- a/drivers/gpu/drm/drm_syncobj.c
+++ b/drivers/gpu/drm/drm_syncobj.c
@@ -885,3 +885,36 @@ drm_syncobj_wait_ioctl(struct drm_device *dev, void *data,
 
 	return ret;
 }
+
+int
+drm_syncobj_reset_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *file_private)
+{
+	struct drm_syncobj_array *args = data;
+	struct drm_syncobj **syncobjs;
+	uint32_t i;
+	int ret;
+
+	if (!drm_core_check_feature(dev, DRIVER_SYNCOBJ))
+		return -ENODEV;
+
+	if (args->pad != 0)
+		return -EINVAL;
+
+	if (args->count_handles == 0)
+		return -EINVAL;
+
+	ret = drm_syncobj_array_find(file_private,
+				     u64_to_user_ptr(args->handles),
+				     args->count_handles,
+				     &syncobjs);
+	if (ret < 0)
+		return ret;
+
+	for (i = 0; i < args->count_handles; i++)
+		drm_syncobj_replace_fence(syncobjs[i], NULL);
+
+	drm_syncobj_array_free(syncobjs, args->count_handles);
+
+	return 0;
+}
diff --git a/include/uapi/drm/drm.h b/include/uapi/drm/drm.h
index 4c746597225e..b037fdf9e43b 100644
--- a/include/uapi/drm/drm.h
+++ b/include/uapi/drm/drm.h
@@ -731,6 +731,12 @@ struct drm_syncobj_wait {
 	__u32 pad;
 };
 
+struct drm_syncobj_array {
+	__u64 handles;
+	__u32 count_handles;
+	__u32 pad;
+};
+
 #if defined(__cplusplus)
 }
 #endif
@@ -854,6 +860,7 @@ extern "C" {
 #define DRM_IOCTL_SYNCOBJ_HANDLE_TO_FD	DRM_IOWR(0xC1, struct drm_syncobj_handle)
 #define DRM_IOCTL_SYNCOBJ_FD_TO_HANDLE	DRM_IOWR(0xC2, struct drm_syncobj_handle)
 #define DRM_IOCTL_SYNCOBJ_WAIT		DRM_IOWR(0xC3, struct drm_syncobj_wait)
+#define DRM_IOCTL_SYNCOBJ_RESET		DRM_IOWR(0xC4, struct drm_syncobj_array)
 
 /**
  * Device specific ioctls should only be in their respective headers
-- 
cgit v1.2.3


From ffa9443fb3d3eddf0fdf6ac473dc8b5c87f08f15 Mon Sep 17 00:00:00 2001
From: Jason Ekstrand <jason@jlekstrand.net>
Date: Mon, 28 Aug 2017 14:10:28 -0700
Subject: drm/syncobj: Add a signal ioctl (v3)

This IOCTL provides a mechanism for userspace to trigger a sync object
directly.  There are other ways that userspace can trigger a syncobj
such as submitting a dummy batch somewhere or hanging on to a triggered
sync_file and doing an import.  This just provides an easy way to
manually trigger the sync object without weird hacks.

The motivation for this IOCTL is Vulkan fences.  Vulkan lets you create
a fence already in the signaled state so that you can wait on it
immediatly without stalling.  We could also handle this with a new
create flag to ask the driver to create a syncobj that is already
signaled but the IOCTL seemed a bit cleaner and more generic.

v2:
 - Take an array of sync objects (Dave Airlie)
v3:
 - Throw -EINVAL if pad != 0

Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/drm_internal.h |  2 ++
 drivers/gpu/drm/drm_ioctl.c    |  2 ++
 drivers/gpu/drm/drm_syncobj.c  | 36 ++++++++++++++++++++++++++++++++++++
 include/uapi/drm/drm.h         |  1 +
 4 files changed, 41 insertions(+)

(limited to 'include/uapi')

diff --git a/drivers/gpu/drm/drm_internal.h b/drivers/gpu/drm/drm_internal.h
index 83f1615eb1ec..fbc3f308fa19 100644
--- a/drivers/gpu/drm/drm_internal.h
+++ b/drivers/gpu/drm/drm_internal.h
@@ -171,3 +171,5 @@ int drm_syncobj_wait_ioctl(struct drm_device *dev, void *data,
 			   struct drm_file *file_private);
 int drm_syncobj_reset_ioctl(struct drm_device *dev, void *data,
 			    struct drm_file *file_private);
+int drm_syncobj_signal_ioctl(struct drm_device *dev, void *data,
+			     struct drm_file *file_private);
diff --git a/drivers/gpu/drm/drm_ioctl.c b/drivers/gpu/drm/drm_ioctl.c
index 16c5d51a43aa..a9ae6dd2d593 100644
--- a/drivers/gpu/drm/drm_ioctl.c
+++ b/drivers/gpu/drm/drm_ioctl.c
@@ -661,6 +661,8 @@ static const struct drm_ioctl_desc drm_ioctls[] = {
 		      DRM_UNLOCKED|DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF(DRM_IOCTL_SYNCOBJ_RESET, drm_syncobj_reset_ioctl,
 		      DRM_UNLOCKED|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF(DRM_IOCTL_SYNCOBJ_SIGNAL, drm_syncobj_signal_ioctl,
+		      DRM_UNLOCKED|DRM_RENDER_ALLOW),
 };
 
 #define DRM_CORE_IOCTL_COUNT	ARRAY_SIZE( drm_ioctls )
diff --git a/drivers/gpu/drm/drm_syncobj.c b/drivers/gpu/drm/drm_syncobj.c
index 40d2ad293661..0422b8c2c2e7 100644
--- a/drivers/gpu/drm/drm_syncobj.c
+++ b/drivers/gpu/drm/drm_syncobj.c
@@ -918,3 +918,39 @@ drm_syncobj_reset_ioctl(struct drm_device *dev, void *data,
 
 	return 0;
 }
+
+int
+drm_syncobj_signal_ioctl(struct drm_device *dev, void *data,
+			 struct drm_file *file_private)
+{
+	struct drm_syncobj_array *args = data;
+	struct drm_syncobj **syncobjs;
+	uint32_t i;
+	int ret;
+
+	if (!drm_core_check_feature(dev, DRIVER_SYNCOBJ))
+		return -ENODEV;
+
+	if (args->pad != 0)
+		return -EINVAL;
+
+	if (args->count_handles == 0)
+		return -EINVAL;
+
+	ret = drm_syncobj_array_find(file_private,
+				     u64_to_user_ptr(args->handles),
+				     args->count_handles,
+				     &syncobjs);
+	if (ret < 0)
+		return ret;
+
+	for (i = 0; i < args->count_handles; i++) {
+		ret = drm_syncobj_assign_null_handle(syncobjs[i]);
+		if (ret < 0)
+			break;
+	}
+
+	drm_syncobj_array_free(syncobjs, args->count_handles);
+
+	return ret;
+}
diff --git a/include/uapi/drm/drm.h b/include/uapi/drm/drm.h
index b037fdf9e43b..97677cd6964d 100644
--- a/include/uapi/drm/drm.h
+++ b/include/uapi/drm/drm.h
@@ -861,6 +861,7 @@ extern "C" {
 #define DRM_IOCTL_SYNCOBJ_FD_TO_HANDLE	DRM_IOWR(0xC2, struct drm_syncobj_handle)
 #define DRM_IOCTL_SYNCOBJ_WAIT		DRM_IOWR(0xC3, struct drm_syncobj_wait)
 #define DRM_IOCTL_SYNCOBJ_RESET		DRM_IOWR(0xC4, struct drm_syncobj_array)
+#define DRM_IOCTL_SYNCOBJ_SIGNAL	DRM_IOWR(0xC5, struct drm_syncobj_array)
 
 /**
  * Device specific ioctls should only be in their respective headers
-- 
cgit v1.2.3