[PATCH libdrm 0/9] amdgpu:

[PATCH libdrm 0/9] amdgpu:

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

The motivation for these patches is https://bugs.freedesktop.org/110903 .

Patches 1-3 are preparatory.

Patches 4 & 5 are the core patches allowing the issues discussed in the
bug report to be fixed.

Patches 6-8 are further optimizations / cleanups.

Patch 9 is the Mesa patch making use of the new
amdgpu_bo_handle_type_kms_user API to fix the user visible problem.

Note that the libdrm patches need to land first, a libdrm release needs
to be made, and the Mesa patch needs to bump the libdrm_amdgpu version
requirement to that release's version before it can land.

Michel Dänzer (9):
  amdgpu: Pass file descriptor directly to amdgpu_close_kms_handle
  amdgpu: Add BO handle to table in amdgpu_bo_create
  amdgpu: Rename fd_mutex/list to dev_mutex/list
  amdgpu: Add struct amdgpu_core_device and amdgpu_core_bo
  amdgpu: Add amdgpu_bo_handle_type_kms_user
  amdgpu: Re-use an existing amdgpu_device when possible
  amdgpu: Use normal integers for device / core BO reference counting
  amdgpu: Drop refcount member from struct amdgpu_core_bo/device
  winsys/amdgpu: Use amdgpu_bo_handle_type_kms_user for API KMS handles

 amdgpu/amdgpu.h               |  14 +-
 amdgpu/amdgpu_asic_id.c       |   4 +-
 amdgpu/amdgpu_bo.c            | 367 ++++++++++++++++++++++------------
 amdgpu/amdgpu_cs.c            |  64 +++---
 amdgpu/amdgpu_device.c        | 281 +++++++++++++-------------
 amdgpu/amdgpu_gpu_info.c      |  35 ++--
 amdgpu/amdgpu_internal.h      |  31 ++-
 amdgpu/amdgpu_vamgr.c         |   9 +-
 amdgpu/amdgpu_vm.c            |   4 +-
 tests/amdgpu/amdgpu_test.c    |   2 +-
 tests/amdgpu/bo_tests.c       |   2 +-
 tests/amdgpu/cs_tests.c       |   8 +-
 tests/amdgpu/deadlock_tests.c |   8 +-
 tests/amdgpu/uvd_enc_tests.c  |   2 +-
 tests/amdgpu/vce_tests.c      |  12 +-
 tests/amdgpu/vcn_tests.c      |   4 +-
 tests/amdgpu/vm_tests.c       |   2 +-
 17 files changed, 500 insertions(+), 349 deletions(-)

-- 
2.20.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH libdrm 1/9] amdgpu: Pass file descriptor directly to amdgpu_close_kms_handle

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

And propagate drmIoctl's return value.

This allows replacing all remaining open-coded DRM_IOCTL_GEM_CLOSE
ioctl calls with amdgpu_close_kms_handle calls.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 amdgpu/amdgpu_bo.c | 35 +++++++++++++++--------------------
 1 file changed, 15 insertions(+), 20 deletions(-)

diff --git a/amdgpu/amdgpu_bo.c b/amdgpu/amdgpu_bo.c
index 32ee358f..6c0b8517 100644
--- a/amdgpu/amdgpu_bo.c
+++ b/amdgpu/amdgpu_bo.c
@@ -39,13 +39,12 @@
 #include "amdgpu_internal.h"
 #include "util_math.h"
 
-static void amdgpu_close_kms_handle(amdgpu_device_handle dev,
-				     uint32_t handle)
+static int amdgpu_close_kms_handle(int fd, uint32_t handle)
 {
 	struct drm_gem_close args = {};
 
 	args.handle = handle;
-	drmIoctl(dev->fd, DRM_IOCTL_GEM_CLOSE, &args);
+	return drmIoctl(fd, DRM_IOCTL_GEM_CLOSE, &args);
 }
 
 static int amdgpu_bo_create(amdgpu_device_handle dev,
@@ -93,7 +92,7 @@ drm_public int amdgpu_bo_alloc(amdgpu_device_handle dev,
 	r = amdgpu_bo_create(dev, alloc_buffer->alloc_size, args.out.handle,
 			     buf_handle);
 	if (r) {
-		amdgpu_close_kms_handle(dev, args.out.handle);
+		amdgpu_close_kms_handle(dev->fd, args.out.handle);
 		goto out;
 	}
 
@@ -214,11 +213,8 @@ static int amdgpu_bo_export_flink(amdgpu_bo_handle bo)
 
 	bo->flink_name = flink.name;
 
-	if (bo->dev->flink_fd != bo->dev->fd) {
-		struct drm_gem_close args = {};
-		args.handle = handle;
-		drmIoctl(bo->dev->flink_fd, DRM_IOCTL_GEM_CLOSE, &args);
-	}
+	if (bo->dev->flink_fd != bo->dev->fd)
+		amdgpu_close_kms_handle(bo->dev->flink_fd, handle);
 
 	pthread_mutex_lock(&bo->dev->bo_table_mutex);
 	r = handle_table_insert(&bo->dev->bo_flink_names, bo->flink_name, bo);
@@ -261,7 +257,6 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle dev,
 		     struct amdgpu_bo_import_result *output)
 {
 	struct drm_gem_open open_arg = {};
-	struct drm_gem_close close_arg = {};
 	struct amdgpu_bo *bo = NULL;
 	uint32_t handle = 0, flink_name = 0;
 	uint64_t alloc_size = 0;
@@ -345,12 +340,12 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle dev,
 			close(dma_fd);
 			if (r)
 				goto free_bo_handle;
-			close_arg.handle = open_arg.handle;
-			r = drmIoctl(dev->flink_fd, DRM_IOCTL_GEM_CLOSE,
-				     &close_arg);
+			r = amdgpu_close_kms_handle(dev->flink_fd,
+						    open_arg.handle);
 			if (r)
 				goto free_bo_handle;
 		}
+		open_arg.handle = 0;
 		break;
 
 	case amdgpu_bo_handle_type_dma_buf_fd:
@@ -388,14 +383,13 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle dev,
 remove_handle:
 	handle_table_remove(&dev->bo_handles, bo->handle);
 free_bo_handle:
-	if (flink_name && !close_arg.handle && open_arg.handle) {
-		close_arg.handle = open_arg.handle;
-		drmIoctl(dev->flink_fd, DRM_IOCTL_GEM_CLOSE, &close_arg);
-	}
+	if (flink_name && open_arg.handle)
+		amdgpu_close_kms_handle(dev->flink_fd, open_arg.handle);
+
 	if (bo)
 		amdgpu_bo_free(bo);
 	else
-		amdgpu_close_kms_handle(dev, handle);
+		amdgpu_close_kms_handle(dev->fd, handle);
 unlock:
 	pthread_mutex_unlock(&dev->bo_table_mutex);
 	return r;
@@ -424,12 +418,13 @@ drm_public int amdgpu_bo_free(amdgpu_bo_handle buf_handle)
 			amdgpu_bo_cpu_unmap(bo);
 		}
 
-		amdgpu_close_kms_handle(dev, bo->handle);
+		amdgpu_close_kms_handle(dev->fd, bo->handle);
 		pthread_mutex_destroy(&bo->cpu_access_mutex);
 		free(bo);
 	}
 
 	pthread_mutex_unlock(&dev->bo_table_mutex);
+
 	return 0;
 }
 
@@ -604,7 +599,7 @@ drm_public int amdgpu_create_bo_from_user_mem(amdgpu_device_handle dev,
 
 	r = amdgpu_bo_create(dev, size, args.handle, buf_handle);
 	if (r) {
-		amdgpu_close_kms_handle(dev, args.handle);
+		amdgpu_close_kms_handle(dev->fd, args.handle);
 		goto out;
 	}
 
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH libdrm 1/9] amdgpu: Pass file descriptor directly to amdgpu_close_kms_handle

[Emil Velikov]

On Mon, 24 Jun 2019 at 17:54, Michel Dänzer <michel@daenzer.net> wrote:
>
> From: Michel Dänzer <michel.daenzer@amd.com>
>
> And propagate drmIoctl's return value.
>
> This allows replacing all remaining open-coded DRM_IOCTL_GEM_CLOSE
> ioctl calls with amdgpu_close_kms_handle calls.
>
> Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
> ---
>  amdgpu/amdgpu_bo.c | 35 +++++++++++++++--------------------
>  1 file changed, 15 insertions(+), 20 deletions(-)
>
Fwiw patches 1-3 are:
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

-Emil
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH libdrm 2/9] amdgpu: Add BO handle to table in amdgpu_bo_create

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

Simplifies its callers.

dev->bo_table_mutex is now always held when amdgpu_bo_create is called
(this was already the case in amdgpu_bo_import).

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 amdgpu/amdgpu_bo.c | 32 ++++++++++++--------------------
 1 file changed, 12 insertions(+), 20 deletions(-)

diff --git a/amdgpu/amdgpu_bo.c b/amdgpu/amdgpu_bo.c
index 6c0b8517..5bdb8fe8 100644
--- a/amdgpu/amdgpu_bo.c
+++ b/amdgpu/amdgpu_bo.c
@@ -53,11 +53,18 @@ static int amdgpu_bo_create(amdgpu_device_handle dev,
 			    amdgpu_bo_handle *buf_handle)
 {
 	struct amdgpu_bo *bo;
+	int r;
 
 	bo = calloc(1, sizeof(struct amdgpu_bo));
 	if (!bo)
 		return -ENOMEM;
 
+	r = handle_table_insert(&dev->bo_handles, handle, bo);
+	if (r) {
+		free(bo);
+		return r;
+	}
+
 	atomic_set(&bo->refcount, 1);
 	bo->dev = dev;
 	bo->alloc_size = size;
@@ -89,19 +96,14 @@ drm_public int amdgpu_bo_alloc(amdgpu_device_handle dev,
 	if (r)
 		goto out;
 
+	pthread_mutex_lock(&dev->bo_table_mutex);
 	r = amdgpu_bo_create(dev, alloc_buffer->alloc_size, args.out.handle,
 			     buf_handle);
+	pthread_mutex_unlock(&dev->bo_table_mutex);
 	if (r) {
 		amdgpu_close_kms_handle(dev->fd, args.out.handle);
-		goto out;
 	}
 
-	pthread_mutex_lock(&dev->bo_table_mutex);
-	r = handle_table_insert(&dev->bo_handles, (*buf_handle)->handle,
-				*buf_handle);
-	pthread_mutex_unlock(&dev->bo_table_mutex);
-	if (r)
-		amdgpu_bo_free(*buf_handle);
 out:
 	return r;
 }
@@ -363,15 +365,12 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle dev,
 	if (r)
 		goto free_bo_handle;
 
-	r = handle_table_insert(&dev->bo_handles, bo->handle, bo);
-	if (r)
-		goto free_bo_handle;
 	if (flink_name) {
 		bo->flink_name = flink_name;
 		r = handle_table_insert(&dev->bo_flink_names, flink_name,
 					bo);
 		if (r)
-			goto remove_handle;
+			goto free_bo_handle;
 
 	}
 
@@ -380,8 +379,6 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle dev,
 	pthread_mutex_unlock(&dev->bo_table_mutex);
 	return 0;
 
-remove_handle:
-	handle_table_remove(&dev->bo_handles, bo->handle);
 free_bo_handle:
 	if (flink_name && open_arg.handle)
 		amdgpu_close_kms_handle(dev->flink_fd, open_arg.handle);
@@ -597,18 +594,13 @@ drm_public int amdgpu_create_bo_from_user_mem(amdgpu_device_handle dev,
 	if (r)
 		goto out;
 
+	pthread_mutex_lock(&dev->bo_table_mutex);
 	r = amdgpu_bo_create(dev, size, args.handle, buf_handle);
+	pthread_mutex_unlock(&dev->bo_table_mutex);
 	if (r) {
 		amdgpu_close_kms_handle(dev->fd, args.handle);
-		goto out;
 	}
 
-	pthread_mutex_lock(&dev->bo_table_mutex);
-	r = handle_table_insert(&dev->bo_handles, (*buf_handle)->handle,
-				*buf_handle);
-	pthread_mutex_unlock(&dev->bo_table_mutex);
-	if (r)
-		amdgpu_bo_free(*buf_handle);
 out:
 	return r;
 }
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH libdrm 3/9] amdgpu: Rename fd_mutex/list to dev_mutex/list

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

Seems to better reflect what they're for.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 amdgpu/amdgpu_device.c | 30 +++++++++++++++---------------
 1 file changed, 15 insertions(+), 15 deletions(-)

diff --git a/amdgpu/amdgpu_device.c b/amdgpu/amdgpu_device.c
index 362494b1..76b4e5eb 100644
--- a/amdgpu/amdgpu_device.c
+++ b/amdgpu/amdgpu_device.c
@@ -43,8 +43,8 @@
 
 #define PTR_TO_UINT(x) ((unsigned)((intptr_t)(x)))
 
-static pthread_mutex_t fd_mutex = PTHREAD_MUTEX_INITIALIZER;
-static amdgpu_device_handle fd_list;
+static pthread_mutex_t dev_mutex = PTHREAD_MUTEX_INITIALIZER;
+static amdgpu_device_handle dev_list;
 
 static int fd_compare(int fd1, int fd2)
 {
@@ -95,13 +95,13 @@ static int amdgpu_get_auth(int fd, int *auth)
 
 static void amdgpu_device_free_internal(amdgpu_device_handle dev)
 {
-	amdgpu_device_handle *node = &fd_list;
+	amdgpu_device_handle *node = &dev_list;
 
-	pthread_mutex_lock(&fd_mutex);
+	pthread_mutex_lock(&dev_mutex);
 	while (*node != dev && (*node)->next)
 		node = &(*node)->next;
 	*node = (*node)->next;
-	pthread_mutex_unlock(&fd_mutex);
+	pthread_mutex_unlock(&dev_mutex);
 
 	close(dev->fd);
 	if ((dev->flink_fd >= 0) && (dev->fd != dev->flink_fd))
@@ -155,16 +155,16 @@ drm_public int amdgpu_device_initialize(int fd,
 
 	*device_handle = NULL;
 
-	pthread_mutex_lock(&fd_mutex);
+	pthread_mutex_lock(&dev_mutex);
 	r = amdgpu_get_auth(fd, &flag_auth);
 	if (r) {
 		fprintf(stderr, "%s: amdgpu_get_auth (1) failed (%i)\n",
 			__func__, r);
-		pthread_mutex_unlock(&fd_mutex);
+		pthread_mutex_unlock(&dev_mutex);
 		return r;
 	}
 
-	for (dev = fd_list; dev; dev = dev->next)
+	for (dev = dev_list; dev; dev = dev->next)
 		if (fd_compare(dev->fd, fd) == 0)
 			break;
 
@@ -173,7 +173,7 @@ drm_public int amdgpu_device_initialize(int fd,
 		if (r) {
 			fprintf(stderr, "%s: amdgpu_get_auth (2) failed (%i)\n",
 				__func__, r);
-			pthread_mutex_unlock(&fd_mutex);
+			pthread_mutex_unlock(&dev_mutex);
 			return r;
 		}
 		if ((flag_auth) && (!flag_authexist)) {
@@ -182,14 +182,14 @@ drm_public int amdgpu_device_initialize(int fd,
 		*major_version = dev->major_version;
 		*minor_version = dev->minor_version;
 		amdgpu_device_reference(device_handle, dev);
-		pthread_mutex_unlock(&fd_mutex);
+		pthread_mutex_unlock(&dev_mutex);
 		return 0;
 	}
 
 	dev = calloc(1, sizeof(struct amdgpu_device));
 	if (!dev) {
 		fprintf(stderr, "%s: calloc failed\n", __func__);
-		pthread_mutex_unlock(&fd_mutex);
+		pthread_mutex_unlock(&dev_mutex);
 		return -ENOMEM;
 	}
 
@@ -265,9 +265,9 @@ drm_public int amdgpu_device_initialize(int fd,
 	*major_version = dev->major_version;
 	*minor_version = dev->minor_version;
 	*device_handle = dev;
-	dev->next = fd_list;
-	fd_list = dev;
-	pthread_mutex_unlock(&fd_mutex);
+	dev->next = dev_list;
+	dev_list = dev;
+	pthread_mutex_unlock(&dev_mutex);
 
 	return 0;
 
@@ -275,7 +275,7 @@ cleanup:
 	if (dev->fd >= 0)
 		close(dev->fd);
 	free(dev);
-	pthread_mutex_unlock(&fd_mutex);
+	pthread_mutex_unlock(&dev_mutex);
 	return r;
 }
 
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH libdrm 8/9] amdgpu: Drop refcount member from struct amdgpu_core_bo/device

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

We can just walk the lists of struct amdgpu_bo/device to find out if
there are any references left.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 amdgpu/amdgpu_bo.c       | 14 +++++++++++---
 amdgpu/amdgpu_device.c   | 30 ++++++++++++++++++++++--------
 amdgpu/amdgpu_internal.h |  2 --
 3 files changed, 33 insertions(+), 13 deletions(-)

diff --git a/amdgpu/amdgpu_bo.c b/amdgpu/amdgpu_bo.c
index f40df484..332ce4be 100644
--- a/amdgpu/amdgpu_bo.c
+++ b/amdgpu/amdgpu_bo.c
@@ -65,7 +65,6 @@ static int amdgpu_core_bo_create(struct amdgpu_core_device *dev,
 		return r;
 	}
 
-	bo->refcount = 1;
 	bo->alloc_size = size;
 	bo->handle = handle;
 	pthread_mutex_init(&bo->cpu_access_mutex, NULL);
@@ -95,7 +94,6 @@ static int amdgpu_bo_create(amdgpu_device_handle user_dev,
 				goto out;
 			}
 		}
-		bo->refcount++;
 	} else {
 		r = amdgpu_core_bo_create(dev, size, handle, &bo);
 		if (r)
@@ -466,8 +464,18 @@ static void amdgpu_core_bo_free(struct amdgpu_bo *user_bo)
 {
 	struct amdgpu_core_device *dev = user_bo->dev->core;
 	struct amdgpu_core_bo *bo = user_bo->core;
+	struct amdgpu_bo *iter_bo;
 
-	if (--bo->refcount == 0) {
+	/* Are there any other user BOs left referencing the same core BO? */
+	for (iter_bo = bo->user_bos; iter_bo; iter_bo = iter_bo->next) {
+		if (iter_bo == user_bo)
+			continue;
+
+		if (iter_bo->core == bo)
+			break;
+	}
+
+	if (!iter_bo) {
 		/* Remove the buffer from the hash tables. */
 		handle_table_remove(&dev->bo_handles, bo->handle);
 
diff --git a/amdgpu/amdgpu_device.c b/amdgpu/amdgpu_device.c
index 8a99b8bb..1709099e 100644
--- a/amdgpu/amdgpu_device.c
+++ b/amdgpu/amdgpu_device.c
@@ -72,9 +72,6 @@ static int fd_compare(int fd1, int fd2)
 
 static void amdgpu_device_free(struct amdgpu_core_device *dev)
 {
-	if (--dev->refcount > 0)
-		return;
-
 	close(dev->fd);
 
 	amdgpu_vamgr_deinit(&dev->vamgr_32);
@@ -102,6 +99,26 @@ static bool same_file_description(int fd1, int fd2)
 	return fd1 == fd2;
 }
 
+static void amdgpu_device_unreference(amdgpu_device_handle user_dev)
+{
+	struct amdgpu_device *dev;
+
+	if (!user_dev->core)
+		return;
+
+	/* Are there any other user devices left referencing the core device? */
+	for (dev = dev_list; dev; dev = dev->next) {
+		if (dev == user_dev)
+			continue;
+
+		if (dev->core == user_dev->core)
+			break;
+	}
+
+	if (!dev)
+		amdgpu_device_free(user_dev->core);
+}
+
 static int amdgpu_device_init(amdgpu_device_handle user_dev)
 {
 	struct amdgpu_device *dev_iter;
@@ -115,7 +132,6 @@ static int amdgpu_device_init(amdgpu_device_handle user_dev)
 			break;
 
 	if (dev_iter) {
-		dev_iter->core->refcount++;
 		user_dev->core = dev_iter->core;
 		return 0;
 	}
@@ -126,7 +142,6 @@ static int amdgpu_device_init(amdgpu_device_handle user_dev)
 		return -ENOMEM;
 	}
 
-	dev->refcount = 1;
 	pthread_mutex_init(&dev->bo_table_mutex, NULL);
 
 	dev->fd = user_dev->user_fd;
@@ -251,8 +266,7 @@ out:
 cleanup:
 	if (!user_dev->core || user_dev->user_fd != user_dev->core->fd)
 		close(user_dev->user_fd);
-	if (user_dev->core)
-		amdgpu_device_free(user_dev->core);
+	amdgpu_device_unreference(user_dev);
 	free(user_dev);
 	pthread_mutex_unlock(&dev_mutex);
 	return r;
@@ -274,7 +288,7 @@ drm_public int amdgpu_device_deinitialize(amdgpu_device_handle user_dev)
 		if (user_dev->user_fd != dev->fd)
 			close(user_dev->user_fd);
 
-		amdgpu_device_free(dev);
+		amdgpu_device_unreference(user_dev);
 
 		free(user_dev);
 	}
diff --git a/amdgpu/amdgpu_internal.h b/amdgpu/amdgpu_internal.h
index 70566f98..7110e30c 100644
--- a/amdgpu/amdgpu_internal.h
+++ b/amdgpu/amdgpu_internal.h
@@ -65,7 +65,6 @@ struct amdgpu_va {
 };
 
 struct amdgpu_core_device {
-	int refcount;
 	int fd;
 	unsigned major_version;
 	unsigned minor_version;
@@ -97,7 +96,6 @@ struct amdgpu_device {
 };
 
 struct amdgpu_core_bo {
-	int refcount;
 	amdgpu_bo_handle user_bos;
 
 	uint64_t alloc_size;
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH mesa 9/9] winsys/amdgpu: Use amdgpu_bo_handle_type_kms_user for API KMS handles

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

Gallium API callers expect the returned handles to be valid for the
DRM file descriptor passed to the driver during initialization, which
may not be the case with amdgpu_bo_handle_type_kms.

Bugzilla: https://bugs.freedesktop.org/110903
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 src/gallium/winsys/amdgpu/drm/amdgpu_bo.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_bo.c b/src/gallium/winsys/amdgpu/drm/amdgpu_bo.c
index 37098ab305f..d4f7e1c7a95 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_bo.c
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_bo.c
@@ -1544,7 +1544,7 @@ static bool amdgpu_bo_get_handle(struct pb_buffer *buffer,
       type = amdgpu_bo_handle_type_dma_buf_fd;
       break;
    case WINSYS_HANDLE_TYPE_KMS:
-      type = amdgpu_bo_handle_type_kms;
+      type = amdgpu_bo_handle_type_kms_user;
       break;
    default:
       return false;
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH libdrm 4/9] amdgpu: Add struct amdgpu_core_device and amdgpu_core_bo

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

They can be referenced by any number of struct amdgpu_device/bo, which
are used for amdgpu_device/bo_handle in the public API.

This allows keeping track of the DRM file descriptor passed to
amdgpu_device_initialize and the one used for CS submission etc.
separately. The core structs hold the information relevant for the
latter.

Because we now always keep a duplicate of the file descriptor passed to
amdgpu_device_initialize, we can use that for flink, and we no longer
need to check its authentication status (flink could never be expected
to work after passing an unauthenticated file descriptor to
amdgpu_device_initialize).

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 amdgpu/amdgpu_asic_id.c       |   4 +-
 amdgpu/amdgpu_bo.c            | 251 ++++++++++++++++++++++------------
 amdgpu/amdgpu_cs.c            |  64 +++++----
 amdgpu/amdgpu_device.c        | 219 ++++++++++++-----------------
 amdgpu/amdgpu_gpu_info.c      |  35 ++---
 amdgpu/amdgpu_internal.h      |  27 ++--
 amdgpu/amdgpu_vamgr.c         |   9 +-
 amdgpu/amdgpu_vm.c            |   4 +-
 tests/amdgpu/amdgpu_test.c    |   2 +-
 tests/amdgpu/bo_tests.c       |   2 +-
 tests/amdgpu/cs_tests.c       |   8 +-
 tests/amdgpu/deadlock_tests.c |   8 +-
 tests/amdgpu/uvd_enc_tests.c  |   2 +-
 tests/amdgpu/vce_tests.c      |  12 +-
 tests/amdgpu/vcn_tests.c      |   4 +-
 tests/amdgpu/vm_tests.c       |   2 +-
 16 files changed, 360 insertions(+), 293 deletions(-)

diff --git a/amdgpu/amdgpu_asic_id.c b/amdgpu/amdgpu_asic_id.c
index a5007ffc..356c8a59 100644
--- a/amdgpu/amdgpu_asic_id.c
+++ b/amdgpu/amdgpu_asic_id.c
@@ -34,7 +34,7 @@
 #include "amdgpu_drm.h"
 #include "amdgpu_internal.h"
 
-static int parse_one_line(struct amdgpu_device *dev, const char *line)
+static int parse_one_line(struct amdgpu_core_device *dev, const char *line)
 {
 	char *buf, *saveptr;
 	char *s_did;
@@ -104,7 +104,7 @@ out:
 	return r;
 }
 
-void amdgpu_parse_asic_ids(struct amdgpu_device *dev)
+void amdgpu_parse_asic_ids(struct amdgpu_core_device *dev)
 {
 	FILE *fp;
 	char *line = NULL;
diff --git a/amdgpu/amdgpu_bo.c b/amdgpu/amdgpu_bo.c
index 5bdb8fe8..7fec1f15 100644
--- a/amdgpu/amdgpu_bo.c
+++ b/amdgpu/amdgpu_bo.c
@@ -47,15 +47,15 @@ static int amdgpu_close_kms_handle(int fd, uint32_t handle)
 	return drmIoctl(fd, DRM_IOCTL_GEM_CLOSE, &args);
 }
 
-static int amdgpu_bo_create(amdgpu_device_handle dev,
-			    uint64_t size,
-			    uint32_t handle,
-			    amdgpu_bo_handle *buf_handle)
+static int amdgpu_core_bo_create(struct amdgpu_core_device *dev,
+				 uint64_t size,
+				 uint32_t handle,
+				 struct amdgpu_core_bo **out_bo)
 {
-	struct amdgpu_bo *bo;
+	struct amdgpu_core_bo *bo;
 	int r;
 
-	bo = calloc(1, sizeof(struct amdgpu_bo));
+	bo = calloc(1, sizeof(struct amdgpu_core_bo));
 	if (!bo)
 		return -ENOMEM;
 
@@ -66,19 +66,64 @@ static int amdgpu_bo_create(amdgpu_device_handle dev,
 	}
 
 	atomic_set(&bo->refcount, 1);
-	bo->dev = dev;
 	bo->alloc_size = size;
 	bo->handle = handle;
 	pthread_mutex_init(&bo->cpu_access_mutex, NULL);
 
-	*buf_handle = bo;
+	*out_bo = bo;
 	return 0;
 }
 
-drm_public int amdgpu_bo_alloc(amdgpu_device_handle dev,
+static int amdgpu_bo_create(amdgpu_device_handle user_dev,
+			    uint64_t size,
+			    uint32_t handle,
+			    amdgpu_bo_handle *buf_handle)
+{
+	struct amdgpu_core_device *dev = user_dev->core;
+	struct amdgpu_bo *user_bo = NULL;
+	struct amdgpu_core_bo *bo;
+	int r;
+
+	bo = handle_table_lookup(&dev->bo_handles, handle);
+
+	if (bo) {
+		for (user_bo = bo->user_bos; user_bo; user_bo = user_bo->next) {
+			if (user_bo->dev == user_dev) {
+				/* Re-use existing buffer */
+				atomic_inc(&user_bo->refcount);
+				r = 0;
+				goto out;
+			}
+		}
+		atomic_inc(&bo->refcount);
+	} else {
+		r = amdgpu_core_bo_create(dev, size, handle, &bo);
+		if (r)
+			goto out;
+	}
+
+	user_bo = calloc(1, sizeof(struct amdgpu_bo));
+	if (!user_bo) {
+		r = -ENOMEM;
+		goto out;
+	}
+
+	atomic_set(&user_bo->refcount, 1);
+	user_bo->next = bo->user_bos;
+	bo->user_bos = user_bo;
+	user_bo->core = bo;
+	user_bo->dev = user_dev;
+
+out:
+	*buf_handle = user_bo;
+	return r;
+}
+
+drm_public int amdgpu_bo_alloc(amdgpu_device_handle user_dev,
 			       struct amdgpu_bo_alloc_request *alloc_buffer,
 			       amdgpu_bo_handle *buf_handle)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	union drm_amdgpu_gem_create args;
 	int r;
 
@@ -97,8 +142,8 @@ drm_public int amdgpu_bo_alloc(amdgpu_device_handle dev,
 		goto out;
 
 	pthread_mutex_lock(&dev->bo_table_mutex);
-	r = amdgpu_bo_create(dev, alloc_buffer->alloc_size, args.out.handle,
-			     buf_handle);
+	r = amdgpu_bo_create(user_dev, alloc_buffer->alloc_size,
+			     args.out.handle, buf_handle);
 	pthread_mutex_unlock(&dev->bo_table_mutex);
 	if (r) {
 		amdgpu_close_kms_handle(dev->fd, args.out.handle);
@@ -108,9 +153,10 @@ out:
 	return r;
 }
 
-drm_public int amdgpu_bo_set_metadata(amdgpu_bo_handle bo,
+drm_public int amdgpu_bo_set_metadata(amdgpu_bo_handle user_bo,
 				      struct amdgpu_bo_metadata *info)
 {
+	struct amdgpu_core_bo *bo = user_bo->core;
 	struct drm_amdgpu_gem_metadata args = {};
 
 	args.handle = bo->handle;
@@ -126,17 +172,19 @@ drm_public int amdgpu_bo_set_metadata(amdgpu_bo_handle bo,
 		memcpy(args.data.data, info->umd_metadata, info->size_metadata);
 	}
 
-	return drmCommandWriteRead(bo->dev->fd,
+	return drmCommandWriteRead(user_bo->dev->core->fd,
 				   DRM_AMDGPU_GEM_METADATA,
 				   &args, sizeof(args));
 }
 
-drm_public int amdgpu_bo_query_info(amdgpu_bo_handle bo,
+drm_public int amdgpu_bo_query_info(amdgpu_bo_handle user_bo,
 				    struct amdgpu_bo_info *info)
 {
+	struct amdgpu_core_bo *bo = user_bo->core;
 	struct drm_amdgpu_gem_metadata metadata = {};
 	struct drm_amdgpu_gem_create_in bo_info = {};
 	struct drm_amdgpu_gem_op gem_op = {};
+	int fd = user_bo->dev->core->fd;
 	int r;
 
 	/* Validate the BO passed in */
@@ -147,8 +195,8 @@ drm_public int amdgpu_bo_query_info(amdgpu_bo_handle bo,
 	metadata.handle = bo->handle;
 	metadata.op = AMDGPU_GEM_METADATA_OP_GET_METADATA;
 
-	r = drmCommandWriteRead(bo->dev->fd, DRM_AMDGPU_GEM_METADATA,
-				&metadata, sizeof(metadata));
+	r = drmCommandWriteRead(fd, DRM_AMDGPU_GEM_METADATA, &metadata,
+				sizeof(metadata));
 	if (r)
 		return r;
 
@@ -161,8 +209,7 @@ drm_public int amdgpu_bo_query_info(amdgpu_bo_handle bo,
 	gem_op.op = AMDGPU_GEM_OP_GET_GEM_CREATE_INFO;
 	gem_op.value = (uintptr_t)&bo_info;
 
-	r = drmCommandWriteRead(bo->dev->fd, DRM_AMDGPU_GEM_OP,
-				&gem_op, sizeof(gem_op));
+	r = drmCommandWriteRead(fd, DRM_AMDGPU_GEM_OP, &gem_op, sizeof(gem_op));
 	if (r)
 		return r;
 
@@ -182,29 +229,30 @@ drm_public int amdgpu_bo_query_info(amdgpu_bo_handle bo,
 	return 0;
 }
 
-static int amdgpu_bo_export_flink(amdgpu_bo_handle bo)
+static int amdgpu_bo_export_flink(amdgpu_bo_handle user_bo)
 {
+	struct amdgpu_core_device *dev = user_bo->dev->core;
+	struct amdgpu_core_bo *bo = user_bo->core;
+	int user_fd = user_bo->dev->user_fd;
 	struct drm_gem_flink flink;
 	int fd, dma_fd;
 	uint32_t handle;
 	int r;
 
-	fd = bo->dev->fd;
+	fd = dev->fd;
 	handle = bo->handle;
 	if (bo->flink_name)
 		return 0;
 
-
-	if (bo->dev->flink_fd != bo->dev->fd) {
-		r = drmPrimeHandleToFD(bo->dev->fd, bo->handle, DRM_CLOEXEC,
-				       &dma_fd);
+	if (user_fd != fd) {
+		r = drmPrimeHandleToFD(fd, bo->handle, DRM_CLOEXEC, &dma_fd);
 		if (!r) {
-			r = drmPrimeFDToHandle(bo->dev->flink_fd, dma_fd, &handle);
+			r = drmPrimeFDToHandle(user_fd, dma_fd, &handle);
 			close(dma_fd);
 		}
 		if (r)
 			return r;
-		fd = bo->dev->flink_fd;
+		fd = user_fd;
 	}
 	memset(&flink, 0, sizeof(flink));
 	flink.handle = handle;
@@ -215,25 +263,26 @@ static int amdgpu_bo_export_flink(amdgpu_bo_handle bo)
 
 	bo->flink_name = flink.name;
 
-	if (bo->dev->flink_fd != bo->dev->fd)
-		amdgpu_close_kms_handle(bo->dev->flink_fd, handle);
+	if (user_fd != dev->fd)
+		amdgpu_close_kms_handle(user_fd, handle);
 
-	pthread_mutex_lock(&bo->dev->bo_table_mutex);
-	r = handle_table_insert(&bo->dev->bo_flink_names, bo->flink_name, bo);
-	pthread_mutex_unlock(&bo->dev->bo_table_mutex);
+	pthread_mutex_lock(&dev->bo_table_mutex);
+	r = handle_table_insert(&dev->bo_flink_names, bo->flink_name, bo);
+	pthread_mutex_unlock(&dev->bo_table_mutex);
 
 	return r;
 }
 
-drm_public int amdgpu_bo_export(amdgpu_bo_handle bo,
+drm_public int amdgpu_bo_export(amdgpu_bo_handle user_bo,
 				enum amdgpu_bo_handle_type type,
 				uint32_t *shared_handle)
 {
+	struct amdgpu_core_bo *bo = user_bo->core;
 	int r;
 
 	switch (type) {
 	case amdgpu_bo_handle_type_gem_flink_name:
-		r = amdgpu_bo_export_flink(bo);
+		r = amdgpu_bo_export_flink(user_bo);
 		if (r)
 			return r;
 
@@ -246,21 +295,24 @@ drm_public int amdgpu_bo_export(amdgpu_bo_handle bo,
 		return 0;
 
 	case amdgpu_bo_handle_type_dma_buf_fd:
-		return drmPrimeHandleToFD(bo->dev->fd, bo->handle,
+		return drmPrimeHandleToFD(user_bo->dev->core->fd, bo->handle,
 					  DRM_CLOEXEC | DRM_RDWR,
 					  (int*)shared_handle);
 	}
 	return -EINVAL;
 }
 
-drm_public int amdgpu_bo_import(amdgpu_device_handle dev,
+drm_public int amdgpu_bo_import(amdgpu_device_handle user_dev,
 				enum amdgpu_bo_handle_type type,
 				uint32_t shared_handle,
-		     struct amdgpu_bo_import_result *output)
+				struct amdgpu_bo_import_result *output)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	struct drm_gem_open open_arg = {};
-	struct amdgpu_bo *bo = NULL;
+	struct amdgpu_bo *user_bo = NULL;
 	uint32_t handle = 0, flink_name = 0;
+	int user_fd = user_dev->user_fd;
+	struct amdgpu_core_bo *bo;
 	uint64_t alloc_size = 0;
 	int r = 0;
 	int dma_fd;
@@ -313,37 +365,32 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle dev,
 	}
 
 	if (bo) {
-		/* The buffer already exists, just bump the refcount. */
-		atomic_inc(&bo->refcount);
-		pthread_mutex_unlock(&dev->bo_table_mutex);
-
-		output->buf_handle = bo;
-		output->alloc_size = bo->alloc_size;
-		return 0;
+		handle = bo->handle;
+		alloc_size = bo->alloc_size;
+		goto bo_create;
 	}
 
 	/* Open the handle. */
 	switch (type) {
 	case amdgpu_bo_handle_type_gem_flink_name:
 		open_arg.name = shared_handle;
-		r = drmIoctl(dev->flink_fd, DRM_IOCTL_GEM_OPEN, &open_arg);
+		r = drmIoctl(user_fd, DRM_IOCTL_GEM_OPEN, &open_arg);
 		if (r)
 			goto unlock;
 
 		flink_name = shared_handle;
 		handle = open_arg.handle;
 		alloc_size = open_arg.size;
-		if (dev->flink_fd != dev->fd) {
-			r = drmPrimeHandleToFD(dev->flink_fd, handle,
-					       DRM_CLOEXEC, &dma_fd);
+		if (user_fd != dev->fd) {
+			r = drmPrimeHandleToFD(user_fd, handle, DRM_CLOEXEC,
+					       &dma_fd);
 			if (r)
 				goto free_bo_handle;
 			r = drmPrimeFDToHandle(dev->fd, dma_fd, &handle);
 			close(dma_fd);
 			if (r)
 				goto free_bo_handle;
-			r = amdgpu_close_kms_handle(dev->flink_fd,
-						    open_arg.handle);
+			r = amdgpu_close_kms_handle(user_fd, open_arg.handle);
 			if (r)
 				goto free_bo_handle;
 		}
@@ -360,11 +407,13 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle dev,
 		assert(0); /* unreachable */
 	}
 
+bo_create:
 	/* Initialize it. */
-	r = amdgpu_bo_create(dev, alloc_size, handle, &bo);
+	r = amdgpu_bo_create(user_dev, alloc_size, handle, &user_bo);
 	if (r)
 		goto free_bo_handle;
 
+	bo = user_bo->core;
 	if (flink_name) {
 		bo->flink_name = flink_name;
 		r = handle_table_insert(&dev->bo_flink_names, flink_name,
@@ -374,17 +423,17 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle dev,
 
 	}
 
-	output->buf_handle = bo;
+	output->buf_handle = user_bo;
 	output->alloc_size = bo->alloc_size;
 	pthread_mutex_unlock(&dev->bo_table_mutex);
 	return 0;
 
 free_bo_handle:
 	if (flink_name && open_arg.handle)
-		amdgpu_close_kms_handle(dev->flink_fd, open_arg.handle);
+		amdgpu_close_kms_handle(user_fd, open_arg.handle);
 
-	if (bo)
-		amdgpu_bo_free(bo);
+	if (user_bo)
+		amdgpu_bo_free(user_bo);
 	else
 		amdgpu_close_kms_handle(dev->fd, handle);
 unlock:
@@ -392,14 +441,10 @@ unlock:
 	return r;
 }
 
-drm_public int amdgpu_bo_free(amdgpu_bo_handle buf_handle)
+static void amdgpu_core_bo_free(struct amdgpu_bo *user_bo)
 {
-	struct amdgpu_device *dev;
-	struct amdgpu_bo *bo = buf_handle;
-
-	assert(bo != NULL);
-	dev = bo->dev;
-	pthread_mutex_lock(&dev->bo_table_mutex);
+	struct amdgpu_core_device *dev = user_bo->dev->core;
+	struct amdgpu_core_bo *bo = user_bo->core;
 
 	if (update_references(&bo->refcount, NULL)) {
 		/* Remove the buffer from the hash tables. */
@@ -412,12 +457,39 @@ drm_public int amdgpu_bo_free(amdgpu_bo_handle buf_handle)
 		/* Release CPU access. */
 		if (bo->cpu_map_count > 0) {
 			bo->cpu_map_count = 1;
-			amdgpu_bo_cpu_unmap(bo);
+			amdgpu_bo_cpu_unmap(user_bo);
 		}
 
 		amdgpu_close_kms_handle(dev->fd, bo->handle);
 		pthread_mutex_destroy(&bo->cpu_access_mutex);
 		free(bo);
+	} else if (bo->user_bos == user_bo) {
+		bo->user_bos = user_bo->next;
+	} else {
+		struct amdgpu_bo *iter;
+
+		for (iter = bo->user_bos; iter->next; iter = iter->next) {
+			if (iter->next == user_bo) {
+				iter->next = user_bo->next;
+				break;
+			}
+		}
+	}
+}
+
+drm_public int amdgpu_bo_free(amdgpu_bo_handle buf_handle)
+{
+	struct amdgpu_bo *user_bo = buf_handle;
+	struct amdgpu_core_device *dev;
+
+	assert(user_bo != NULL);
+	dev = user_bo->dev->core;
+
+	pthread_mutex_lock(&dev->bo_table_mutex);
+
+	if (update_references(&user_bo->refcount, NULL)) {
+		amdgpu_core_bo_free(user_bo);
+		free(user_bo);
 	}
 
 	pthread_mutex_unlock(&dev->bo_table_mutex);
@@ -430,8 +502,10 @@ drm_public void amdgpu_bo_inc_ref(amdgpu_bo_handle bo)
 	atomic_inc(&bo->refcount);
 }
 
-drm_public int amdgpu_bo_cpu_map(amdgpu_bo_handle bo, void **cpu)
+drm_public int amdgpu_bo_cpu_map(amdgpu_bo_handle user_bo, void **cpu)
 {
+	struct amdgpu_core_bo *bo = user_bo->core;
+	int fd = user_bo->dev->core->fd;
 	union drm_amdgpu_gem_mmap args;
 	void *ptr;
 	int r;
@@ -455,8 +529,7 @@ drm_public int amdgpu_bo_cpu_map(amdgpu_bo_handle bo, void **cpu)
 	 * The kernel driver ignores the offset and size parameters. */
 	args.in.handle = bo->handle;
 
-	r = drmCommandWriteRead(bo->dev->fd, DRM_AMDGPU_GEM_MMAP, &args,
-				sizeof(args));
+	r = drmCommandWriteRead(fd, DRM_AMDGPU_GEM_MMAP, &args, sizeof(args));
 	if (r) {
 		pthread_mutex_unlock(&bo->cpu_access_mutex);
 		return r;
@@ -464,7 +537,7 @@ drm_public int amdgpu_bo_cpu_map(amdgpu_bo_handle bo, void **cpu)
 
 	/* Map the buffer. */
 	ptr = drm_mmap(NULL, bo->alloc_size, PROT_READ | PROT_WRITE, MAP_SHARED,
-		       bo->dev->fd, args.out.addr_ptr);
+		       fd, args.out.addr_ptr);
 	if (ptr == MAP_FAILED) {
 		pthread_mutex_unlock(&bo->cpu_access_mutex);
 		return -errno;
@@ -478,8 +551,9 @@ drm_public int amdgpu_bo_cpu_map(amdgpu_bo_handle bo, void **cpu)
 	return 0;
 }
 
-drm_public int amdgpu_bo_cpu_unmap(amdgpu_bo_handle bo)
+drm_public int amdgpu_bo_cpu_unmap(amdgpu_bo_handle user_bo)
 {
+	struct amdgpu_core_bo *bo = user_bo->core;
 	int r;
 
 	pthread_mutex_lock(&bo->cpu_access_mutex);
@@ -504,18 +578,21 @@ drm_public int amdgpu_bo_cpu_unmap(amdgpu_bo_handle bo)
 	return r;
 }
 
-drm_public int amdgpu_query_buffer_size_alignment(amdgpu_device_handle dev,
+drm_public int amdgpu_query_buffer_size_alignment(amdgpu_device_handle user_dev,
 				struct amdgpu_buffer_size_alignments *info)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
+
 	info->size_local = dev->dev_info.pte_fragment_size;
 	info->size_remote = dev->dev_info.gart_page_size;
 	return 0;
 }
 
-drm_public int amdgpu_bo_wait_for_idle(amdgpu_bo_handle bo,
+drm_public int amdgpu_bo_wait_for_idle(amdgpu_bo_handle user_bo,
 				       uint64_t timeout_ns,
 			    bool *busy)
 {
+	struct amdgpu_core_bo *bo = user_bo->core;
 	union drm_amdgpu_gem_wait_idle args;
 	int r;
 
@@ -523,7 +600,7 @@ drm_public int amdgpu_bo_wait_for_idle(amdgpu_bo_handle bo,
 	args.in.handle = bo->handle;
 	args.in.timeout = amdgpu_cs_calculate_timeout(timeout_ns);
 
-	r = drmCommandWriteRead(bo->dev->fd, DRM_AMDGPU_GEM_WAIT_IDLE,
+	r = drmCommandWriteRead(user_bo->dev->core->fd, DRM_AMDGPU_GEM_WAIT_IDLE,
 				&args, sizeof(args));
 
 	if (r == 0) {
@@ -535,13 +612,14 @@ drm_public int amdgpu_bo_wait_for_idle(amdgpu_bo_handle bo,
 	}
 }
 
-drm_public int amdgpu_find_bo_by_cpu_mapping(amdgpu_device_handle dev,
+drm_public int amdgpu_find_bo_by_cpu_mapping(amdgpu_device_handle user_dev,
 					     void *cpu,
 					     uint64_t size,
 					     amdgpu_bo_handle *buf_handle,
 					     uint64_t *offset_in_bo)
 {
-	struct amdgpu_bo *bo;
+	struct amdgpu_core_device *dev = user_dev->core;
+	struct amdgpu_core_bo *bo;
 	uint32_t i;
 	int r = 0;
 
@@ -564,8 +642,8 @@ drm_public int amdgpu_find_bo_by_cpu_mapping(amdgpu_device_handle dev,
 	}
 
 	if (i < dev->bo_handles.max_key) {
-		atomic_inc(&bo->refcount);
-		*buf_handle = bo;
+		r = amdgpu_bo_create(user_dev, bo->alloc_size, bo->handle,
+				     buf_handle);
 		*offset_in_bo = (uintptr_t)cpu - (uintptr_t)bo->cpu_ptr;
 	} else {
 		*buf_handle = NULL;
@@ -577,11 +655,12 @@ drm_public int amdgpu_find_bo_by_cpu_mapping(amdgpu_device_handle dev,
 	return r;
 }
 
-drm_public int amdgpu_create_bo_from_user_mem(amdgpu_device_handle dev,
+drm_public int amdgpu_create_bo_from_user_mem(amdgpu_device_handle user_dev,
 					      void *cpu,
 					      uint64_t size,
 					      amdgpu_bo_handle *buf_handle)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	int r;
 	struct drm_amdgpu_gem_userptr args;
 
@@ -595,7 +674,7 @@ drm_public int amdgpu_create_bo_from_user_mem(amdgpu_device_handle dev,
 		goto out;
 
 	pthread_mutex_lock(&dev->bo_table_mutex);
-	r = amdgpu_bo_create(dev, size, args.handle, buf_handle);
+	r = amdgpu_bo_create(user_dev, size, args.handle, buf_handle);
 	pthread_mutex_unlock(&dev->bo_table_mutex);
 	if (r) {
 		amdgpu_close_kms_handle(dev->fd, args.handle);
@@ -605,11 +684,12 @@ out:
 	return r;
 }
 
-drm_public int amdgpu_bo_list_create_raw(amdgpu_device_handle dev,
+drm_public int amdgpu_bo_list_create_raw(amdgpu_device_handle user_dev,
 					 uint32_t number_of_buffers,
 					 struct drm_amdgpu_bo_list_entry *buffers,
 					 uint32_t *result)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	union drm_amdgpu_bo_list args;
 	int r;
 
@@ -626,9 +706,10 @@ drm_public int amdgpu_bo_list_create_raw(amdgpu_device_handle dev,
 	return r;
 }
 
-drm_public int amdgpu_bo_list_destroy_raw(amdgpu_device_handle dev,
+drm_public int amdgpu_bo_list_destroy_raw(amdgpu_device_handle user_dev,
 					  uint32_t bo_list)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	union drm_amdgpu_bo_list args;
 
 	memset(&args, 0, sizeof(args));
@@ -639,12 +720,13 @@ drm_public int amdgpu_bo_list_destroy_raw(amdgpu_device_handle dev,
 				   &args, sizeof(args));
 }
 
-drm_public int amdgpu_bo_list_create(amdgpu_device_handle dev,
+drm_public int amdgpu_bo_list_create(amdgpu_device_handle user_dev,
 				     uint32_t number_of_resources,
 				     amdgpu_bo_handle *resources,
 				     uint8_t *resource_prios,
 				     amdgpu_bo_list_handle *result)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	struct drm_amdgpu_bo_list_entry *list;
 	union drm_amdgpu_bo_list args;
 	unsigned i;
@@ -674,7 +756,7 @@ drm_public int amdgpu_bo_list_create(amdgpu_device_handle dev,
 	args.in.bo_info_ptr = (uint64_t)(uintptr_t)list;
 
 	for (i = 0; i < number_of_resources; i++) {
-		list[i].bo_handle = resources[i]->handle;
+		list[i].bo_handle = resources[i]->core->handle;
 		if (resource_prios)
 			list[i].bo_priority = resource_prios[i];
 		else
@@ -740,7 +822,7 @@ drm_public int amdgpu_bo_list_update(amdgpu_bo_list_handle handle,
 	args.in.bo_info_ptr = (uintptr_t)list;
 
 	for (i = 0; i < number_of_resources; i++) {
-		list[i].bo_handle = resources[i]->handle;
+		list[i].bo_handle = resources[i]->core->handle;
 		if (resource_prios)
 			list[i].bo_priority = resource_prios[i];
 		else
@@ -770,7 +852,7 @@ drm_public int amdgpu_bo_va_op(amdgpu_bo_handle bo,
 				   AMDGPU_VM_PAGE_EXECUTABLE, ops);
 }
 
-drm_public int amdgpu_bo_va_op_raw(amdgpu_device_handle dev,
+drm_public int amdgpu_bo_va_op_raw(amdgpu_device_handle user_dev,
 				   amdgpu_bo_handle bo,
 				   uint64_t offset,
 				   uint64_t size,
@@ -778,6 +860,7 @@ drm_public int amdgpu_bo_va_op_raw(amdgpu_device_handle dev,
 				   uint64_t flags,
 				   uint32_t ops)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	struct drm_amdgpu_gem_va va;
 	int r;
 
@@ -786,7 +869,7 @@ drm_public int amdgpu_bo_va_op_raw(amdgpu_device_handle dev,
 		return -EINVAL;
 
 	memset(&va, 0, sizeof(va));
-	va.handle = bo ? bo->handle : 0;
+	va.handle = bo ? bo->core->handle : 0;
 	va.operation = ops;
 	va.flags = flags;
 	va.va_address = addr;
diff --git a/amdgpu/amdgpu_cs.c b/amdgpu/amdgpu_cs.c
index 20d5aef2..98130105 100644
--- a/amdgpu/amdgpu_cs.c
+++ b/amdgpu/amdgpu_cs.c
@@ -48,22 +48,24 @@ static int amdgpu_cs_reset_sem(amdgpu_semaphore_handle sem);
  *
  * \return  0 on success otherwise POSIX Error code
 */
-drm_public int amdgpu_cs_ctx_create2(amdgpu_device_handle dev,
+drm_public int amdgpu_cs_ctx_create2(amdgpu_device_handle user_dev,
 				     uint32_t priority,
 				     amdgpu_context_handle *context)
 {
 	struct amdgpu_context *gpu_context;
+	struct amdgpu_core_device *dev;
 	union drm_amdgpu_ctx args;
 	int i, j, k;
 	int r;
 
-	if (!dev || !context)
+	if (!user_dev || !context)
 		return -EINVAL;
 
 	gpu_context = calloc(1, sizeof(struct amdgpu_context));
 	if (!gpu_context)
 		return -ENOMEM;
 
+	dev = user_dev->core;
 	gpu_context->dev = dev;
 
 	r = pthread_mutex_init(&gpu_context->sequence_mutex, NULL);
@@ -156,7 +158,7 @@ drm_public int amdgpu_cs_ctx_override_priority(amdgpu_device_handle dev,
 	memset(&args, 0, sizeof(args));
 
 	args.in.op = AMDGPU_SCHED_OP_CONTEXT_PRIORITY_OVERRIDE;
-	args.in.fd = dev->fd;
+	args.in.fd = dev->core->fd;
 	args.in.priority = priority;
 	args.in.ctx_id = context->id;
 
@@ -269,7 +271,7 @@ static int amdgpu_cs_submit_one(amdgpu_context_handle context,
 		chunks[i].chunk_data = (uint64_t)(uintptr_t)&chunk_data[i];
 
 		/* fence bo handle */
-		chunk_data[i].fence_data.handle = ibs_request->fence_info.handle->handle;
+		chunk_data[i].fence_data.handle = ibs_request->fence_info.handle->core->handle;
 		/* offset */
 		chunk_data[i].fence_data.offset = 
 			ibs_request->fence_info.offset * sizeof(uint64_t);
@@ -409,7 +411,7 @@ static int amdgpu_ioctl_wait_cs(amdgpu_context_handle context,
 				uint64_t flags,
 				bool *busy)
 {
-	amdgpu_device_handle dev = context->dev;
+	struct amdgpu_core_device *dev = context->dev;
 	union drm_amdgpu_wait_cs args;
 	int r;
 
@@ -471,8 +473,8 @@ static int amdgpu_ioctl_wait_fences(struct amdgpu_cs_fence *fences,
 				    uint32_t *status,
 				    uint32_t *first)
 {
+	struct amdgpu_core_device *dev = fences[0].context->dev;
 	struct drm_amdgpu_fence *drm_fences;
-	amdgpu_device_handle dev = fences[0].context->dev;
 	union drm_amdgpu_wait_fences args;
 	int r;
 	uint32_t i;
@@ -633,7 +635,7 @@ drm_public int amdgpu_cs_create_syncobj2(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjCreate(dev->fd, flags, handle);
+	return drmSyncobjCreate(dev->core->fd, flags, handle);
 }
 
 drm_public int amdgpu_cs_create_syncobj(amdgpu_device_handle dev,
@@ -642,7 +644,7 @@ drm_public int amdgpu_cs_create_syncobj(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjCreate(dev->fd, 0, handle);
+	return drmSyncobjCreate(dev->core->fd, 0, handle);
 }
 
 drm_public int amdgpu_cs_destroy_syncobj(amdgpu_device_handle dev,
@@ -651,7 +653,7 @@ drm_public int amdgpu_cs_destroy_syncobj(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjDestroy(dev->fd, handle);
+	return drmSyncobjDestroy(dev->core->fd, handle);
 }
 
 drm_public int amdgpu_cs_syncobj_reset(amdgpu_device_handle dev,
@@ -661,7 +663,7 @@ drm_public int amdgpu_cs_syncobj_reset(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjReset(dev->fd, syncobjs, syncobj_count);
+	return drmSyncobjReset(dev->core->fd, syncobjs, syncobj_count);
 }
 
 drm_public int amdgpu_cs_syncobj_signal(amdgpu_device_handle dev,
@@ -671,7 +673,7 @@ drm_public int amdgpu_cs_syncobj_signal(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjSignal(dev->fd, syncobjs, syncobj_count);
+	return drmSyncobjSignal(dev->core->fd, syncobjs, syncobj_count);
 }
 
 drm_public int amdgpu_cs_syncobj_timeline_signal(amdgpu_device_handle dev,
@@ -682,7 +684,7 @@ drm_public int amdgpu_cs_syncobj_timeline_signal(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjTimelineSignal(dev->fd, syncobjs,
+	return drmSyncobjTimelineSignal(dev->core->fd, syncobjs,
 					points, syncobj_count);
 }
 
@@ -694,7 +696,7 @@ drm_public int amdgpu_cs_syncobj_wait(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjWait(dev->fd, handles, num_handles, timeout_nsec,
+	return drmSyncobjWait(dev->core->fd, handles, num_handles, timeout_nsec,
 			      flags, first_signaled);
 }
 
@@ -707,7 +709,7 @@ drm_public int amdgpu_cs_syncobj_timeline_wait(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjTimelineWait(dev->fd, handles, points, num_handles,
+	return drmSyncobjTimelineWait(dev->core->fd, handles, points, num_handles,
 				      timeout_nsec, flags, first_signaled);
 }
 
@@ -718,7 +720,7 @@ drm_public int amdgpu_cs_syncobj_query(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjQuery(dev->fd, handles, points, num_handles);
+	return drmSyncobjQuery(dev->core->fd, handles, points, num_handles);
 }
 
 drm_public int amdgpu_cs_export_syncobj(amdgpu_device_handle dev,
@@ -728,7 +730,7 @@ drm_public int amdgpu_cs_export_syncobj(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjHandleToFD(dev->fd, handle, shared_fd);
+	return drmSyncobjHandleToFD(dev->core->fd, handle, shared_fd);
 }
 
 drm_public int amdgpu_cs_import_syncobj(amdgpu_device_handle dev,
@@ -738,7 +740,7 @@ drm_public int amdgpu_cs_import_syncobj(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjFDToHandle(dev->fd, shared_fd, handle);
+	return drmSyncobjFDToHandle(dev->core->fd, shared_fd, handle);
 }
 
 drm_public int amdgpu_cs_syncobj_export_sync_file(amdgpu_device_handle dev,
@@ -748,7 +750,7 @@ drm_public int amdgpu_cs_syncobj_export_sync_file(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjExportSyncFile(dev->fd, syncobj, sync_file_fd);
+	return drmSyncobjExportSyncFile(dev->core->fd, syncobj, sync_file_fd);
 }
 
 drm_public int amdgpu_cs_syncobj_import_sync_file(amdgpu_device_handle dev,
@@ -758,21 +760,24 @@ drm_public int amdgpu_cs_syncobj_import_sync_file(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjImportSyncFile(dev->fd, syncobj, sync_file_fd);
+	return drmSyncobjImportSyncFile(dev->core->fd, syncobj, sync_file_fd);
 }
 
-drm_public int amdgpu_cs_syncobj_export_sync_file2(amdgpu_device_handle dev,
+drm_public int amdgpu_cs_syncobj_export_sync_file2(amdgpu_device_handle user_dev,
 						   uint32_t syncobj,
 						   uint64_t point,
 						   uint32_t flags,
 						   int *sync_file_fd)
 {
+	struct amdgpu_core_device *dev;
 	uint32_t binary_handle;
 	int ret;
 
-	if (NULL == dev)
+	if (!user_dev)
 		return -EINVAL;
 
+	dev = user_dev->core;
+
 	if (!point)
 		return drmSyncobjExportSyncFile(dev->fd, syncobj, sync_file_fd);
 
@@ -790,17 +795,20 @@ out:
 	return ret;
 }
 
-drm_public int amdgpu_cs_syncobj_import_sync_file2(amdgpu_device_handle dev,
+drm_public int amdgpu_cs_syncobj_import_sync_file2(amdgpu_device_handle user_dev,
 						   uint32_t syncobj,
 						   uint64_t point,
 						   int sync_file_fd)
 {
+	struct amdgpu_core_device *dev;
 	uint32_t binary_handle;
 	int ret;
 
-	if (NULL == dev)
+	if (!user_dev)
 		return -EINVAL;
 
+	dev = user_dev->core;
+
 	if (!point)
 		return drmSyncobjImportSyncFile(dev->fd, syncobj, sync_file_fd);
 
@@ -827,7 +835,7 @@ drm_public int amdgpu_cs_syncobj_transfer(amdgpu_device_handle dev,
 	if (NULL == dev)
 		return -EINVAL;
 
-	return drmSyncobjTransfer(dev->fd,
+	return drmSyncobjTransfer(dev->core->fd,
 				  dst_handle, dst_point,
 				  src_handle, src_point,
 				  flags);
@@ -854,7 +862,7 @@ drm_public int amdgpu_cs_submit_raw(amdgpu_device_handle dev,
 	cs.in.ctx_id = context->id;
 	cs.in.bo_list_handle = bo_list_handle ? bo_list_handle->handle : 0;
 	cs.in.num_chunks = num_chunks;
-	r = drmCommandWriteRead(dev->fd, DRM_AMDGPU_CS,
+	r = drmCommandWriteRead(dev->core->fd, DRM_AMDGPU_CS,
 				&cs, sizeof(cs));
 	if (r)
 		return r;
@@ -883,7 +891,7 @@ drm_public int amdgpu_cs_submit_raw2(amdgpu_device_handle dev,
 	cs.in.ctx_id = context->id;
 	cs.in.bo_list_handle = bo_list_handle;
 	cs.in.num_chunks = num_chunks;
-	r = drmCommandWriteRead(dev->fd, DRM_AMDGPU_CS,
+	r = drmCommandWriteRead(dev->core->fd, DRM_AMDGPU_CS,
 				&cs, sizeof(cs));
 	if (!r && seq_no)
 		*seq_no = cs.out.handle;
@@ -893,7 +901,7 @@ drm_public int amdgpu_cs_submit_raw2(amdgpu_device_handle dev,
 drm_public void amdgpu_cs_chunk_fence_info_to_data(struct amdgpu_cs_fence_info *fence_info,
 					struct drm_amdgpu_cs_chunk_data *data)
 {
-	data->fence_data.handle = fence_info->handle->handle;
+	data->fence_data.handle = fence_info->handle->core->handle;
 	data->fence_data.offset = fence_info->offset * sizeof(uint64_t);
 }
 
@@ -923,7 +931,7 @@ drm_public int amdgpu_cs_fence_to_handle(amdgpu_device_handle dev,
 	fth.in.fence.seq_no = fence->fence;
 	fth.in.what = what;
 
-	r = drmCommandWriteRead(dev->fd, DRM_AMDGPU_FENCE_TO_HANDLE,
+	r = drmCommandWriteRead(dev->core->fd, DRM_AMDGPU_FENCE_TO_HANDLE,
 				&fth, sizeof(fth));
 	if (r == 0)
 		*out_handle = fth.out.handle;
diff --git a/amdgpu/amdgpu_device.c b/amdgpu/amdgpu_device.c
index 76b4e5eb..abf5f942 100644
--- a/amdgpu/amdgpu_device.c
+++ b/amdgpu/amdgpu_device.c
@@ -44,7 +44,7 @@
 #define PTR_TO_UINT(x) ((unsigned)((intptr_t)(x)))
 
 static pthread_mutex_t dev_mutex = PTHREAD_MUTEX_INITIALIZER;
-static amdgpu_device_handle dev_list;
+static struct amdgpu_core_device *dev_list;
 
 static int fd_compare(int fd1, int fd2)
 {
@@ -65,47 +65,15 @@ static int fd_compare(int fd1, int fd2)
 	return result;
 }
 
-/**
-* Get the authenticated form fd,
-*
-* \param   fd   - \c [in]  File descriptor for AMD GPU device
-* \param   auth - \c [out] Pointer to output the fd is authenticated or not
-*                          A render node fd, output auth = 0
-*                          A legacy fd, get the authenticated for compatibility root
-*
-* \return   0 on success\n
-*          >0 - AMD specific error code\n
-*          <0 - Negative POSIX Error code
-*/
-static int amdgpu_get_auth(int fd, int *auth)
+static void amdgpu_device_free(struct amdgpu_core_device *dev)
 {
-	int r = 0;
-	drm_client_t client = {};
-
-	if (drmGetNodeTypeFromFd(fd) == DRM_NODE_RENDER)
-		*auth = 0;
-	else {
-		client.idx = 0;
-		r = drmIoctl(fd, DRM_IOCTL_GET_CLIENT, &client);
-		if (!r)
-			*auth = client.auth;
-	}
-	return r;
-}
+	struct amdgpu_core_device **node = &dev_list;
 
-static void amdgpu_device_free_internal(amdgpu_device_handle dev)
-{
-	amdgpu_device_handle *node = &dev_list;
-
-	pthread_mutex_lock(&dev_mutex);
 	while (*node != dev && (*node)->next)
 		node = &(*node)->next;
 	*node = (*node)->next;
-	pthread_mutex_unlock(&dev_mutex);
 
 	close(dev->fd);
-	if ((dev->flink_fd >= 0) && (dev->fd != dev->flink_fd))
-		close(dev->flink_fd);
 
 	amdgpu_vamgr_deinit(&dev->vamgr_32);
 	amdgpu_vamgr_deinit(&dev->vamgr);
@@ -118,87 +86,39 @@ static void amdgpu_device_free_internal(amdgpu_device_handle dev)
 	free(dev);
 }
 
-/**
- * Assignment between two amdgpu_device pointers with reference counting.
- *
- * Usage:
- *    struct amdgpu_device *dst = ... , *src = ...;
- *
- *    dst = src;
- *    // No reference counting. Only use this when you need to move
- *    // a reference from one pointer to another.
- *
- *    amdgpu_device_reference(&dst, src);
- *    // Reference counters are updated. dst is decremented and src is
- *    // incremented. dst is freed if its reference counter is 0.
- */
-static void amdgpu_device_reference(struct amdgpu_device **dst,
-				    struct amdgpu_device *src)
+static int amdgpu_device_init(amdgpu_device_handle user_dev)
 {
-	if (update_references(&(*dst)->refcount, &src->refcount))
-		amdgpu_device_free_internal(*dst);
-	*dst = src;
-}
-
-drm_public int amdgpu_device_initialize(int fd,
-					uint32_t *major_version,
-					uint32_t *minor_version,
-					amdgpu_device_handle *device_handle)
-{
-	struct amdgpu_device *dev;
+	struct amdgpu_core_device *dev;
 	drmVersionPtr version;
-	int r;
-	int flag_auth = 0;
-	int flag_authexist=0;
-	uint32_t accel_working = 0;
 	uint64_t start, max;
-
-	*device_handle = NULL;
-
-	pthread_mutex_lock(&dev_mutex);
-	r = amdgpu_get_auth(fd, &flag_auth);
-	if (r) {
-		fprintf(stderr, "%s: amdgpu_get_auth (1) failed (%i)\n",
-			__func__, r);
-		pthread_mutex_unlock(&dev_mutex);
-		return r;
-	}
+	int r;
 
 	for (dev = dev_list; dev; dev = dev->next)
-		if (fd_compare(dev->fd, fd) == 0)
+		if (fd_compare(dev->fd, user_dev->user_fd) == 0)
 			break;
 
 	if (dev) {
-		r = amdgpu_get_auth(dev->fd, &flag_authexist);
-		if (r) {
-			fprintf(stderr, "%s: amdgpu_get_auth (2) failed (%i)\n",
-				__func__, r);
-			pthread_mutex_unlock(&dev_mutex);
-			return r;
-		}
-		if ((flag_auth) && (!flag_authexist)) {
-			dev->flink_fd = fcntl(fd, F_DUPFD_CLOEXEC, 0);
-		}
-		*major_version = dev->major_version;
-		*minor_version = dev->minor_version;
-		amdgpu_device_reference(device_handle, dev);
-		pthread_mutex_unlock(&dev_mutex);
+		atomic_inc(&dev->refcount);
+		user_dev->core = dev;
 		return 0;
 	}
 
-	dev = calloc(1, sizeof(struct amdgpu_device));
+	dev = calloc(1, sizeof(struct amdgpu_core_device));
 	if (!dev) {
 		fprintf(stderr, "%s: calloc failed\n", __func__);
-		pthread_mutex_unlock(&dev_mutex);
 		return -ENOMEM;
 	}
 
-	dev->fd = -1;
-	dev->flink_fd = -1;
-
 	atomic_set(&dev->refcount, 1);
+	pthread_mutex_init(&dev->bo_table_mutex, NULL);
+
+	dev->fd = user_dev->user_fd;
+	user_dev->core = dev;
+
+	dev->next = dev_list;
+	dev_list = dev;
 
-	version = drmGetVersion(fd);
+	version = drmGetVersion(dev->fd);
 	if (version->version_major != 3) {
 		fprintf(stderr, "%s: DRM version is %d.%d.%d but this driver is "
 			"only compatible with 3.x.x.\n",
@@ -211,28 +131,11 @@ drm_public int amdgpu_device_initialize(int fd,
 		goto cleanup;
 	}
 
-	dev->fd = fcntl(fd, F_DUPFD_CLOEXEC, 0);
-	dev->flink_fd = dev->fd;
 	dev->major_version = version->version_major;
 	dev->minor_version = version->version_minor;
 	drmFreeVersion(version);
 
-	pthread_mutex_init(&dev->bo_table_mutex, NULL);
-
-	/* Check if acceleration is working. */
-	r = amdgpu_query_info(dev, AMDGPU_INFO_ACCEL_WORKING, 4, &accel_working);
-	if (r) {
-		fprintf(stderr, "%s: amdgpu_query_info(ACCEL_WORKING) failed (%i)\n",
-			__func__, r);
-		goto cleanup;
-	}
-	if (!accel_working) {
-		fprintf(stderr, "%s: AMDGPU_INFO_ACCEL_WORKING = 0\n", __func__);
-		r = -EBADF;
-		goto cleanup;
-	}
-
-	r = amdgpu_query_gpu_info_init(dev);
+	r = amdgpu_query_gpu_info_init(user_dev);
 	if (r) {
 		fprintf(stderr, "%s: amdgpu_query_gpu_info_init failed\n", __func__);
 		goto cleanup;
@@ -261,39 +164,97 @@ drm_public int amdgpu_device_initialize(int fd,
 			  dev->dev_info.virtual_address_alignment);
 
 	amdgpu_parse_asic_ids(dev);
+	return 0;
 
-	*major_version = dev->major_version;
-	*minor_version = dev->minor_version;
-	*device_handle = dev;
-	dev->next = dev_list;
-	dev_list = dev;
+cleanup:
+	user_dev->core = NULL;
+	close(dev->fd);
+	free(dev);
+	return r;
+}
+
+drm_public int amdgpu_device_initialize(int fd,
+					uint32_t *major_version,
+					uint32_t *minor_version,
+					amdgpu_device_handle *device_handle)
+{
+	struct amdgpu_device *user_dev;
+	uint32_t accel_working = 0;
+	int r;
+
+	*device_handle = NULL;
+
+	user_dev = calloc(1, sizeof(struct amdgpu_device));
+	if (!user_dev) {
+		fprintf(stderr, "%s: calloc failed\n", __func__);
+		return -ENOMEM;
+	}
+
+	user_dev->user_fd = fcntl(fd, F_DUPFD_CLOEXEC, 0);
+
+	pthread_mutex_lock(&dev_mutex);
+
+	r = amdgpu_device_init(user_dev);
+	if (r != 0)
+		goto cleanup;
+
+	/* Check if acceleration is working. */
+	r = amdgpu_query_info(user_dev, AMDGPU_INFO_ACCEL_WORKING, 4, &accel_working);
+	if (r) {
+		fprintf(stderr, "%s: amdgpu_query_info(ACCEL_WORKING) failed (%i)\n",
+			__func__, r);
+		goto cleanup;
+	}
+	if (!accel_working) {
+		fprintf(stderr, "%s: AMDGPU_INFO_ACCEL_WORKING = 0\n", __func__);
+		r = -EBADF;
+		goto cleanup;
+	}
+
+	*major_version = user_dev->core->major_version;
+	*minor_version = user_dev->core->minor_version;
+	*device_handle = user_dev;
 	pthread_mutex_unlock(&dev_mutex);
 
 	return 0;
 
 cleanup:
-	if (dev->fd >= 0)
-		close(dev->fd);
-	free(dev);
+	if (!user_dev->core || user_dev->user_fd != user_dev->core->fd)
+		close(user_dev->user_fd);
+	if (user_dev->core && update_references(&user_dev->core->refcount, NULL))
+		amdgpu_device_free(user_dev->core);
+	free(user_dev);
 	pthread_mutex_unlock(&dev_mutex);
 	return r;
 }
 
-drm_public int amdgpu_device_deinitialize(amdgpu_device_handle dev)
+drm_public int amdgpu_device_deinitialize(amdgpu_device_handle user_dev)
 {
-	amdgpu_device_reference(&dev, NULL);
+	struct amdgpu_core_device *dev = user_dev->core;
+
+	pthread_mutex_lock(&dev_mutex);
+
+	if (user_dev->user_fd != dev->fd)
+		close(user_dev->user_fd);
+
+	if (update_references(&dev->refcount, NULL))
+		amdgpu_device_free(dev);
+
+	pthread_mutex_unlock(&dev_mutex);
+	free(user_dev);
 	return 0;
 }
 
-drm_public const char *amdgpu_get_marketing_name(amdgpu_device_handle dev)
+drm_public const char *amdgpu_get_marketing_name(amdgpu_device_handle user_dev)
 {
-	return dev->marketing_name;
+	return user_dev->core->marketing_name;
 }
 
-drm_public int amdgpu_query_sw_info(amdgpu_device_handle dev,
+drm_public int amdgpu_query_sw_info(amdgpu_device_handle user_dev,
 				    enum amdgpu_sw_info info,
 				    void *value)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	uint32_t *val32 = (uint32_t*)value;
 
 	switch (info) {
diff --git a/amdgpu/amdgpu_gpu_info.c b/amdgpu/amdgpu_gpu_info.c
index 777087f2..7253fbea 100644
--- a/amdgpu/amdgpu_gpu_info.c
+++ b/amdgpu/amdgpu_gpu_info.c
@@ -40,7 +40,7 @@ drm_public int amdgpu_query_info(amdgpu_device_handle dev, unsigned info_id,
 	request.return_size = size;
 	request.query = info_id;
 
-	return drmCommandWrite(dev->fd, DRM_AMDGPU_INFO, &request,
+	return drmCommandWrite(dev->core->fd, DRM_AMDGPU_INFO, &request,
 			       sizeof(struct drm_amdgpu_info));
 }
 
@@ -55,7 +55,7 @@ drm_public int amdgpu_query_crtc_from_id(amdgpu_device_handle dev, unsigned id,
 	request.query = AMDGPU_INFO_CRTC_FROM_ID;
 	request.mode_crtc.id = id;
 
-	return drmCommandWrite(dev->fd, DRM_AMDGPU_INFO, &request,
+	return drmCommandWrite(dev->core->fd, DRM_AMDGPU_INFO, &request,
 			       sizeof(struct drm_amdgpu_info));
 }
 
@@ -74,7 +74,7 @@ drm_public int amdgpu_read_mm_registers(amdgpu_device_handle dev,
 	request.read_mmr_reg.instance = instance;
 	request.read_mmr_reg.flags = flags;
 
-	return drmCommandWrite(dev->fd, DRM_AMDGPU_INFO, &request,
+	return drmCommandWrite(dev->core->fd, DRM_AMDGPU_INFO, &request,
 			       sizeof(struct drm_amdgpu_info));
 }
 
@@ -90,7 +90,7 @@ drm_public int amdgpu_query_hw_ip_count(amdgpu_device_handle dev,
 	request.query = AMDGPU_INFO_HW_IP_COUNT;
 	request.query_hw_ip.type = type;
 
-	return drmCommandWrite(dev->fd, DRM_AMDGPU_INFO, &request,
+	return drmCommandWrite(dev->core->fd, DRM_AMDGPU_INFO, &request,
 			       sizeof(struct drm_amdgpu_info));
 }
 
@@ -107,7 +107,7 @@ drm_public int amdgpu_query_hw_ip_info(amdgpu_device_handle dev, unsigned type,
 	request.query_hw_ip.type = type;
 	request.query_hw_ip.ip_instance = ip_instance;
 
-	return drmCommandWrite(dev->fd, DRM_AMDGPU_INFO, &request,
+	return drmCommandWrite(dev->core->fd, DRM_AMDGPU_INFO, &request,
 			       sizeof(struct drm_amdgpu_info));
 }
 
@@ -127,7 +127,7 @@ drm_public int amdgpu_query_firmware_version(amdgpu_device_handle dev,
 	request.query_fw.ip_instance = ip_instance;
 	request.query_fw.index = index;
 
-	r = drmCommandWrite(dev->fd, DRM_AMDGPU_INFO, &request,
+	r = drmCommandWrite(dev->core->fd, DRM_AMDGPU_INFO, &request,
 			    sizeof(struct drm_amdgpu_info));
 	if (r)
 		return r;
@@ -137,11 +137,12 @@ drm_public int amdgpu_query_firmware_version(amdgpu_device_handle dev,
 	return 0;
 }
 
-drm_private int amdgpu_query_gpu_info_init(amdgpu_device_handle dev)
+drm_private int amdgpu_query_gpu_info_init(amdgpu_device_handle user_dev)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	int r, i;
 
-	r = amdgpu_query_info(dev, AMDGPU_INFO_DEV_INFO, sizeof(dev->dev_info),
+	r = amdgpu_query_info(user_dev, AMDGPU_INFO_DEV_INFO, sizeof(dev->dev_info),
 			      &dev->dev_info);
 	if (r)
 		return r;
@@ -172,7 +173,7 @@ drm_private int amdgpu_query_gpu_info_init(amdgpu_device_handle dev)
 					    (AMDGPU_INFO_MMR_SH_INDEX_MASK <<
 					     AMDGPU_INFO_MMR_SH_INDEX_SHIFT);
 
-			r = amdgpu_read_mm_registers(dev, 0x263d, 1, instance, 0,
+			r = amdgpu_read_mm_registers(user_dev, 0x263d, 1, instance, 0,
 						     &dev->info.backend_disable[i]);
 			if (r)
 				return r;
@@ -180,13 +181,13 @@ drm_private int amdgpu_query_gpu_info_init(amdgpu_device_handle dev)
 			dev->info.backend_disable[i] =
 				(dev->info.backend_disable[i] >> 16) & 0xff;
 
-			r = amdgpu_read_mm_registers(dev, 0xa0d4, 1, instance, 0,
+			r = amdgpu_read_mm_registers(user_dev, 0xa0d4, 1, instance, 0,
 						     &dev->info.pa_sc_raster_cfg[i]);
 			if (r)
 				return r;
 
 			if (dev->info.family_id >= AMDGPU_FAMILY_CI) {
-				r = amdgpu_read_mm_registers(dev, 0xa0d5, 1, instance, 0,
+				r = amdgpu_read_mm_registers(user_dev, 0xa0d5, 1, instance, 0,
 						     &dev->info.pa_sc_raster_cfg1[i]);
 				if (r)
 					return r;
@@ -194,25 +195,25 @@ drm_private int amdgpu_query_gpu_info_init(amdgpu_device_handle dev)
 		}
 	}
 
-	r = amdgpu_read_mm_registers(dev, 0x263e, 1, 0xffffffff, 0,
+	r = amdgpu_read_mm_registers(user_dev, 0x263e, 1, 0xffffffff, 0,
 					     &dev->info.gb_addr_cfg);
 	if (r)
 		return r;
 
 	if (dev->info.family_id < AMDGPU_FAMILY_AI) {
-		r = amdgpu_read_mm_registers(dev, 0x2644, 32, 0xffffffff, 0,
+		r = amdgpu_read_mm_registers(user_dev, 0x2644, 32, 0xffffffff, 0,
 					     dev->info.gb_tile_mode);
 		if (r)
 			return r;
 
 		if (dev->info.family_id >= AMDGPU_FAMILY_CI) {
-			r = amdgpu_read_mm_registers(dev, 0x2664, 16, 0xffffffff, 0,
+			r = amdgpu_read_mm_registers(user_dev, 0x2664, 16, 0xffffffff, 0,
 						     dev->info.gb_macro_tile_mode);
 			if (r)
 				return r;
 		}
 
-		r = amdgpu_read_mm_registers(dev, 0x9d8, 1, 0xffffffff, 0,
+		r = amdgpu_read_mm_registers(user_dev, 0x9d8, 1, 0xffffffff, 0,
 					     &dev->info.mc_arb_ramcfg);
 		if (r)
 			return r;
@@ -235,7 +236,7 @@ drm_public int amdgpu_query_gpu_info(amdgpu_device_handle dev,
 		return -EINVAL;
 
 	/* Get ASIC info*/
-	*info = dev->info;
+	*info = dev->core->info;
 
 	return 0;
 }
@@ -328,6 +329,6 @@ drm_public int amdgpu_query_sensor_info(amdgpu_device_handle dev, unsigned senso
 	request.query = AMDGPU_INFO_SENSOR;
 	request.sensor_info.type = sensor_type;
 
-	return drmCommandWrite(dev->fd, DRM_AMDGPU_INFO, &request,
+	return drmCommandWrite(dev->core->fd, DRM_AMDGPU_INFO, &request,
 			       sizeof(struct drm_amdgpu_info));
 }
diff --git a/amdgpu/amdgpu_internal.h b/amdgpu/amdgpu_internal.h
index a340abbd..3a2ab74c 100644
--- a/amdgpu/amdgpu_internal.h
+++ b/amdgpu/amdgpu_internal.h
@@ -64,14 +64,13 @@ struct amdgpu_va {
 	struct amdgpu_bo_va_mgr *vamgr;
 };
 
-struct amdgpu_device {
+struct amdgpu_core_device {
 	atomic_t refcount;
-	struct amdgpu_device *next;
 	int fd;
-	int flink_fd;
 	unsigned major_version;
 	unsigned minor_version;
 
+	struct amdgpu_core_device *next;
 	char *marketing_name;
 	/** List of buffer handles. Protected by bo_table_mutex. */
 	struct handle_table bo_handles;
@@ -91,9 +90,14 @@ struct amdgpu_device {
 	struct amdgpu_bo_va_mgr vamgr_high_32;
 };
 
-struct amdgpu_bo {
+struct amdgpu_device {
+	int user_fd;
+	struct amdgpu_core_device *core;
+};
+
+struct amdgpu_core_bo {
 	atomic_t refcount;
-	struct amdgpu_device *dev;
+	amdgpu_bo_handle user_bos;
 
 	uint64_t alloc_size;
 
@@ -105,14 +109,21 @@ struct amdgpu_bo {
 	int cpu_map_count;
 };
 
-struct amdgpu_bo_list {
+struct amdgpu_bo {
+	atomic_t refcount;
+	struct amdgpu_bo *next;
+	struct amdgpu_core_bo *core;
 	struct amdgpu_device *dev;
+};
+
+struct amdgpu_bo_list {
+	struct amdgpu_core_device *dev;
 
 	uint32_t handle;
 };
 
 struct amdgpu_context {
-	struct amdgpu_device *dev;
+	struct amdgpu_core_device *dev;
 	/** Mutex for accessing fences and to maintain command submissions
 	    in good sequence. */
 	pthread_mutex_t sequence_mutex;
@@ -141,7 +152,7 @@ drm_private void amdgpu_vamgr_init(struct amdgpu_bo_va_mgr *mgr, uint64_t start,
 
 drm_private void amdgpu_vamgr_deinit(struct amdgpu_bo_va_mgr *mgr);
 
-drm_private void amdgpu_parse_asic_ids(struct amdgpu_device *dev);
+drm_private void amdgpu_parse_asic_ids(struct amdgpu_core_device *dev);
 
 drm_private int amdgpu_query_gpu_info_init(amdgpu_device_handle dev);
 
diff --git a/amdgpu/amdgpu_vamgr.c b/amdgpu/amdgpu_vamgr.c
index d25d4216..560b84e7 100644
--- a/amdgpu/amdgpu_vamgr.c
+++ b/amdgpu/amdgpu_vamgr.c
@@ -29,10 +29,12 @@
 #include "amdgpu_internal.h"
 #include "util_math.h"
 
-drm_public int amdgpu_va_range_query(amdgpu_device_handle dev,
+drm_public int amdgpu_va_range_query(amdgpu_device_handle user_dev,
 				     enum amdgpu_gpu_va_range type,
 				     uint64_t *start, uint64_t *end)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
+
 	if (type != amdgpu_gpu_va_range_general)
 		return -EINVAL;
 
@@ -186,7 +188,7 @@ out:
 	pthread_mutex_unlock(&mgr->bo_va_mutex);
 }
 
-drm_public int amdgpu_va_range_alloc(amdgpu_device_handle dev,
+drm_public int amdgpu_va_range_alloc(amdgpu_device_handle user_dev,
 				     enum amdgpu_gpu_va_range va_range_type,
 				     uint64_t size,
 				     uint64_t va_base_alignment,
@@ -195,6 +197,7 @@ drm_public int amdgpu_va_range_alloc(amdgpu_device_handle dev,
 				     amdgpu_va_handle *va_range_handle,
 				     uint64_t flags)
 {
+	struct amdgpu_core_device *dev = user_dev->core;
 	struct amdgpu_bo_va_mgr *vamgr;
 
 	/* Clear the flag when the high VA manager is not initialized */
@@ -237,7 +240,7 @@ drm_public int amdgpu_va_range_alloc(amdgpu_device_handle dev,
 			amdgpu_vamgr_free_va(vamgr, *va_base_allocated, size);
 			return -ENOMEM;
 		}
-		va->dev = dev;
+		va->dev = user_dev;
 		va->address = *va_base_allocated;
 		va->size = size;
 		va->range = va_range_type;
diff --git a/amdgpu/amdgpu_vm.c b/amdgpu/amdgpu_vm.c
index 7e6e28f0..de44e6c2 100644
--- a/amdgpu/amdgpu_vm.c
+++ b/amdgpu/amdgpu_vm.c
@@ -33,7 +33,7 @@ drm_public int amdgpu_vm_reserve_vmid(amdgpu_device_handle dev, uint32_t flags)
 	vm.in.op = AMDGPU_VM_OP_RESERVE_VMID;
 	vm.in.flags = flags;
 
-	return drmCommandWriteRead(dev->fd, DRM_AMDGPU_VM,
+	return drmCommandWriteRead(dev->core->fd, DRM_AMDGPU_VM,
 				   &vm, sizeof(vm));
 }
 
@@ -45,6 +45,6 @@ drm_public int amdgpu_vm_unreserve_vmid(amdgpu_device_handle dev,
 	vm.in.op = AMDGPU_VM_OP_UNRESERVE_VMID;
 	vm.in.flags = flags;
 
-	return drmCommandWriteRead(dev->fd, DRM_AMDGPU_VM,
+	return drmCommandWriteRead(dev->core->fd, DRM_AMDGPU_VM,
 				   &vm, sizeof(vm));
 }
diff --git a/tests/amdgpu/amdgpu_test.c b/tests/amdgpu/amdgpu_test.c
index 73403fb4..7095c4e4 100644
--- a/tests/amdgpu/amdgpu_test.c
+++ b/tests/amdgpu/amdgpu_test.c
@@ -428,7 +428,7 @@ static void amdgpu_disable_suites()
 				   &minor_version, &device_handle))
 		return;
 
-	family_id = device_handle->info.family_id;
+	family_id = device_handle->core->info.family_id;
 
 	if (amdgpu_device_deinitialize(device_handle))
 		return;
diff --git a/tests/amdgpu/bo_tests.c b/tests/amdgpu/bo_tests.c
index 7cff4cf7..d89c944d 100644
--- a/tests/amdgpu/bo_tests.c
+++ b/tests/amdgpu/bo_tests.c
@@ -309,7 +309,7 @@ static void amdgpu_bo_find_by_cpu_mapping(void)
 					  &offset);
 	CU_ASSERT_EQUAL(r, 0);
 	CU_ASSERT_EQUAL(offset, 0);
-	CU_ASSERT_EQUAL(bo_handle->handle, find_bo_handle->handle);
+	CU_ASSERT_EQUAL(bo_handle->core->handle, find_bo_handle->core->handle);
 
 	atomic_dec(&find_bo_handle->refcount, 1);
 	r = amdgpu_bo_unmap_and_free(bo_handle, va_handle,
diff --git a/tests/amdgpu/cs_tests.c b/tests/amdgpu/cs_tests.c
index 7ad0f0dc..8a5f6ed3 100644
--- a/tests/amdgpu/cs_tests.c
+++ b/tests/amdgpu/cs_tests.c
@@ -68,7 +68,7 @@ CU_BOOL suite_cs_tests_enable(void)
 					     &minor_version, &device_handle))
 		return CU_FALSE;
 
-	family_id = device_handle->info.family_id;
+	family_id = device_handle->core->info.family_id;
 
 	if (amdgpu_device_deinitialize(device_handle))
 		return CU_FALSE;
@@ -101,10 +101,10 @@ int suite_cs_tests_init(void)
 		return CUE_SINIT_FAILED;
 	}
 
-	family_id = device_handle->info.family_id;
+	family_id = device_handle->core->info.family_id;
 	/* VI asic POLARIS10/11 have specific external_rev_id */
-	chip_rev = device_handle->info.chip_rev;
-	chip_id = device_handle->info.chip_external_rev;
+	chip_rev = device_handle->core->info.chip_rev;
+	chip_id = device_handle->core->info.chip_external_rev;
 
 	r = amdgpu_cs_ctx_create(device_handle, &context_handle);
 	if (r)
diff --git a/tests/amdgpu/deadlock_tests.c b/tests/amdgpu/deadlock_tests.c
index 91368c15..8526bae7 100644
--- a/tests/amdgpu/deadlock_tests.c
+++ b/tests/amdgpu/deadlock_tests.c
@@ -127,14 +127,14 @@ CU_BOOL suite_deadlock_tests_enable(void)
 	 * Only enable for ASICs supporting GPU reset and for which it's enabled
 	 * by default (currently GFX8/9 dGPUS)
 	 */
-	if (device_handle->info.family_id != AMDGPU_FAMILY_VI &&
-	    device_handle->info.family_id != AMDGPU_FAMILY_AI &&
-	    device_handle->info.family_id != AMDGPU_FAMILY_CI) {
+	if (device_handle->core->info.family_id != AMDGPU_FAMILY_VI &&
+	    device_handle->core->info.family_id != AMDGPU_FAMILY_AI &&
+	    device_handle->core->info.family_id != AMDGPU_FAMILY_CI) {
 		printf("\n\nGPU reset is not enabled for the ASIC, deadlock suite disabled\n");
 		enable = CU_FALSE;
 	}
 
-	if (device_handle->info.family_id >= AMDGPU_FAMILY_AI)
+	if (device_handle->core->info.family_id >= AMDGPU_FAMILY_AI)
 		use_uc_mtype = 1;
 
 	if (amdgpu_device_deinitialize(device_handle))
diff --git a/tests/amdgpu/uvd_enc_tests.c b/tests/amdgpu/uvd_enc_tests.c
index b4251bcf..856e7ae1 100644
--- a/tests/amdgpu/uvd_enc_tests.c
+++ b/tests/amdgpu/uvd_enc_tests.c
@@ -114,7 +114,7 @@ int suite_uvd_enc_tests_init(void)
 	if (r)
 		return CUE_SINIT_FAILED;
 
-	family_id = device_handle->info.family_id;
+	family_id = device_handle->core->info.family_id;
 
 	r = amdgpu_cs_ctx_create(device_handle, &context_handle);
 	if (r)
diff --git a/tests/amdgpu/vce_tests.c b/tests/amdgpu/vce_tests.c
index 0026826e..05d9ef57 100644
--- a/tests/amdgpu/vce_tests.c
+++ b/tests/amdgpu/vce_tests.c
@@ -103,10 +103,10 @@ CU_BOOL suite_vce_tests_enable(void)
 					     &minor_version, &device_handle))
 		return CU_FALSE;
 
-	family_id = device_handle->info.family_id;
-	chip_rev = device_handle->info.chip_rev;
-	chip_id = device_handle->info.chip_external_rev;
-	ids_flags = device_handle->info.ids_flags;
+	family_id = device_handle->core->info.family_id;
+	chip_rev = device_handle->core->info.chip_rev;
+	chip_id = device_handle->core->info.chip_external_rev;
+	ids_flags = device_handle->core->info.ids_flags;
 
 	amdgpu_query_firmware_version(device_handle, AMDGPU_INFO_FW_VCE, 0,
 					  0, &version, &feature);
@@ -153,8 +153,8 @@ int suite_vce_tests_init(void)
 		return CUE_SINIT_FAILED;
 	}
 
-	family_id = device_handle->info.family_id;
-	vce_harvest_config = device_handle->info.vce_harvest_config;
+	family_id = device_handle->core->info.family_id;
+	vce_harvest_config = device_handle->core->info.vce_harvest_config;
 
 	r = amdgpu_cs_ctx_create(device_handle, &context_handle);
 	if (r)
diff --git a/tests/amdgpu/vcn_tests.c b/tests/amdgpu/vcn_tests.c
index ad438f35..17248444 100644
--- a/tests/amdgpu/vcn_tests.c
+++ b/tests/amdgpu/vcn_tests.c
@@ -94,7 +94,7 @@ CU_BOOL suite_vcn_tests_enable(void)
 				   &minor_version, &device_handle))
 		return CU_FALSE;
 
-	family_id = device_handle->info.family_id;
+	family_id = device_handle->core->info.family_id;
 
 	if (amdgpu_device_deinitialize(device_handle))
 			return CU_FALSE;
@@ -132,7 +132,7 @@ int suite_vcn_tests_init(void)
 	if (r)
 		return CUE_SINIT_FAILED;
 
-	family_id = device_handle->info.family_id;
+	family_id = device_handle->core->info.family_id;
 
 	r = amdgpu_cs_ctx_create(device_handle, &context_handle);
 	if (r)
diff --git a/tests/amdgpu/vm_tests.c b/tests/amdgpu/vm_tests.c
index 69bc4683..e52aef25 100644
--- a/tests/amdgpu/vm_tests.c
+++ b/tests/amdgpu/vm_tests.c
@@ -43,7 +43,7 @@ CU_BOOL suite_vm_tests_enable(void)
 				     &minor_version, &device_handle))
 		return CU_FALSE;
 
-	if (device_handle->info.family_id == AMDGPU_FAMILY_SI) {
+	if (device_handle->core->info.family_id == AMDGPU_FAMILY_SI) {
 		printf("\n\nCurrently hangs the CP on this ASIC, VM suite disabled\n");
 		enable = CU_FALSE;
 	}
-- 
2.20.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH libdrm 5/9] amdgpu: Add amdgpu_bo_handle_type_kms_user

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

amdgpu_bo_handle_type_kms returns the handle valid for the DRM file
descriptor (fd) used for CS submission etc. This is also valid for the
fd passed to amdgpu_device_initialize the first time for a specific
GPU, but in general not for fds passed to amdgpu_device_initialize
later for the same GPU.

Because some use-cases require a handle valid for the fd passed to
amdgpu_device_initialize, amdgpu_bo_handle_type_kms_user is added for
this purpose.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 amdgpu/amdgpu.h          | 14 ++++++++-
 amdgpu/amdgpu_bo.c       | 65 +++++++++++++++++++++++++++-------------
 amdgpu/amdgpu_internal.h |  1 +
 3 files changed, 59 insertions(+), 21 deletions(-)

diff --git a/amdgpu/amdgpu.h b/amdgpu/amdgpu.h
index 66e45f73..f95c0a36 100644
--- a/amdgpu/amdgpu.h
+++ b/amdgpu/amdgpu.h
@@ -81,7 +81,14 @@ enum amdgpu_bo_handle_type {
 	/** GEM flink name (needs DRM authentication, used by DRI2) */
 	amdgpu_bo_handle_type_gem_flink_name = 0,
 
-	/** KMS handle which is used by all driver ioctls */
+	/** KMS handle which is used by all driver ioctls
+	 *
+	 * NOTE: The returned handle is valid for the DRM file description
+	 * used for command submission, which may be different from the one
+	 * referenced by the file descriptor passed to
+	 * amdgpu_device_initialize. Use amdgpu_bo_handle_type_kms_user to
+	 * get a handle valid for the latter.
+	 */
 	amdgpu_bo_handle_type_kms = 1,
 
 	/** DMA-buf fd handle */
@@ -91,6 +98,11 @@ enum amdgpu_bo_handle_type {
 	 * amdgpu_bo_handle_type_kms, use that instead of this
 	 */
 	amdgpu_bo_handle_type_kms_noimport = 3,
+
+	/** KMS handle valid for the DRM file description referenced by the
+	 * file descriptor passed to amdgpu_device_initialize.
+	 */
+	amdgpu_bo_handle_type_kms_user = 4,
 };
 
 /** Define known types of GPU VM VA ranges */
diff --git a/amdgpu/amdgpu_bo.c b/amdgpu/amdgpu_bo.c
index 7fec1f15..8d42db90 100644
--- a/amdgpu/amdgpu_bo.c
+++ b/amdgpu/amdgpu_bo.c
@@ -229,42 +229,58 @@ drm_public int amdgpu_bo_query_info(amdgpu_bo_handle user_bo,
 	return 0;
 }
 
+static int amdgpu_bo_get_user_handle(struct amdgpu_bo *user_bo,
+				     uint32_t *user_handle)
+{
+	struct amdgpu_device *user_dev = user_bo->dev;
+	struct amdgpu_core_device *dev = user_dev->core;
+	struct amdgpu_core_bo *bo = user_bo->core;
+	int dma_fd;
+	int r;
+
+	if (user_dev->user_fd == dev->fd) {
+		*user_handle = bo->handle;
+		return 0;
+	}
+
+	if (user_bo->user_handle)
+		goto out;
+
+	r = drmPrimeHandleToFD(dev->fd, bo->handle, DRM_CLOEXEC, &dma_fd);
+	if (r)
+		return r;
+
+	r = drmPrimeFDToHandle(user_dev->user_fd, dma_fd, &user_bo->user_handle);
+	close(dma_fd);
+	if (r)
+		return r;
+
+out:
+	*user_handle = user_bo->user_handle;
+	return 0;
+}
+
 static int amdgpu_bo_export_flink(amdgpu_bo_handle user_bo)
 {
 	struct amdgpu_core_device *dev = user_bo->dev->core;
 	struct amdgpu_core_bo *bo = user_bo->core;
-	int user_fd = user_bo->dev->user_fd;
 	struct drm_gem_flink flink;
-	int fd, dma_fd;
-	uint32_t handle;
 	int r;
 
-	fd = dev->fd;
-	handle = bo->handle;
 	if (bo->flink_name)
 		return 0;
 
-	if (user_fd != fd) {
-		r = drmPrimeHandleToFD(fd, bo->handle, DRM_CLOEXEC, &dma_fd);
-		if (!r) {
-			r = drmPrimeFDToHandle(user_fd, dma_fd, &handle);
-			close(dma_fd);
-		}
-		if (r)
-			return r;
-		fd = user_fd;
-	}
 	memset(&flink, 0, sizeof(flink));
-	flink.handle = handle;
 
-	r = drmIoctl(fd, DRM_IOCTL_GEM_FLINK, &flink);
+	r = amdgpu_bo_get_user_handle(user_bo, &flink.handle);
 	if (r)
 		return r;
 
-	bo->flink_name = flink.name;
+	r = drmIoctl(user_bo->dev->user_fd, DRM_IOCTL_GEM_FLINK, &flink);
+	if (r)
+		return r;
 
-	if (user_fd != dev->fd)
-		amdgpu_close_kms_handle(user_fd, handle);
+	bo->flink_name = flink.name;
 
 	pthread_mutex_lock(&dev->bo_table_mutex);
 	r = handle_table_insert(&dev->bo_flink_names, bo->flink_name, bo);
@@ -294,6 +310,9 @@ drm_public int amdgpu_bo_export(amdgpu_bo_handle user_bo,
 		*shared_handle = bo->handle;
 		return 0;
 
+	case amdgpu_bo_handle_type_kms_user:
+		return amdgpu_bo_get_user_handle(user_bo, shared_handle);
+
 	case amdgpu_bo_handle_type_dma_buf_fd:
 		return drmPrimeHandleToFD(user_bo->dev->core->fd, bo->handle,
 					  DRM_CLOEXEC | DRM_RDWR,
@@ -355,6 +374,7 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle user_dev,
 
 	case amdgpu_bo_handle_type_kms:
 	case amdgpu_bo_handle_type_kms_noimport:
+	case amdgpu_bo_handle_type_kms_user:
 		/* Importing a KMS handle in not allowed. */
 		r = -EPERM;
 		goto unlock;
@@ -404,6 +424,7 @@ drm_public int amdgpu_bo_import(amdgpu_device_handle user_dev,
 
 	case amdgpu_bo_handle_type_kms:
 	case amdgpu_bo_handle_type_kms_noimport:
+	case amdgpu_bo_handle_type_kms_user:
 		assert(0); /* unreachable */
 	}
 
@@ -489,6 +510,10 @@ drm_public int amdgpu_bo_free(amdgpu_bo_handle buf_handle)
 
 	if (update_references(&user_bo->refcount, NULL)) {
 		amdgpu_core_bo_free(user_bo);
+		if (user_bo->user_handle) {
+			amdgpu_close_kms_handle(user_bo->dev->user_fd,
+						user_bo->user_handle);
+		}
 		free(user_bo);
 	}
 
diff --git a/amdgpu/amdgpu_internal.h b/amdgpu/amdgpu_internal.h
index 3a2ab74c..a08a4ae8 100644
--- a/amdgpu/amdgpu_internal.h
+++ b/amdgpu/amdgpu_internal.h
@@ -111,6 +111,7 @@ struct amdgpu_core_bo {
 
 struct amdgpu_bo {
 	atomic_t refcount;
+	uint32_t user_handle;
 	struct amdgpu_bo *next;
 	struct amdgpu_core_bo *core;
 	struct amdgpu_device *dev;
-- 
2.20.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH libdrm 6/9] amdgpu: Re-use an existing amdgpu_device when possible

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

It's possible if amdgpu_device's user_fd references the same file
description as the fd passed to amdgpu_device_initialize.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 amdgpu/amdgpu_device.c   | 76 +++++++++++++++++++++++++++++-----------
 amdgpu/amdgpu_internal.h |  3 +-
 2 files changed, 58 insertions(+), 21 deletions(-)

diff --git a/amdgpu/amdgpu_device.c b/amdgpu/amdgpu_device.c
index abf5f942..8d9a85c2 100644
--- a/amdgpu/amdgpu_device.c
+++ b/amdgpu/amdgpu_device.c
@@ -28,6 +28,11 @@
  *
  */
 
+#ifdef __linux__
+#include <linux/kcmp.h>
+#include <sys/syscall.h>
+#endif
+
 #include <sys/stat.h>
 #include <errno.h>
 #include <string.h>
@@ -44,7 +49,7 @@
 #define PTR_TO_UINT(x) ((unsigned)((intptr_t)(x)))
 
 static pthread_mutex_t dev_mutex = PTHREAD_MUTEX_INITIALIZER;
-static struct amdgpu_core_device *dev_list;
+static amdgpu_device_handle dev_list;
 
 static int fd_compare(int fd1, int fd2)
 {
@@ -67,12 +72,6 @@ static int fd_compare(int fd1, int fd2)
 
 static void amdgpu_device_free(struct amdgpu_core_device *dev)
 {
-	struct amdgpu_core_device **node = &dev_list;
-
-	while (*node != dev && (*node)->next)
-		node = &(*node)->next;
-	*node = (*node)->next;
-
 	close(dev->fd);
 
 	amdgpu_vamgr_deinit(&dev->vamgr_32);
@@ -86,20 +85,35 @@ static void amdgpu_device_free(struct amdgpu_core_device *dev)
 	free(dev);
 }
 
+static bool same_file_description(int fd1, int fd2)
+{
+#ifdef __linux__
+	pid_t pid = getpid();
+
+	return syscall(SYS_kcmp, pid, pid, KCMP_FILE, fd1, fd2) == 0;
+#endif
+
+	/* NOTE: This is never true at this point, since we always duplicate the
+	 * fd passed to amdgpu_device_initialize
+	 */
+	return fd1 == fd2;
+}
+
 static int amdgpu_device_init(amdgpu_device_handle user_dev)
 {
+	struct amdgpu_device *dev_iter;
 	struct amdgpu_core_device *dev;
 	drmVersionPtr version;
 	uint64_t start, max;
 	int r;
 
-	for (dev = dev_list; dev; dev = dev->next)
-		if (fd_compare(dev->fd, user_dev->user_fd) == 0)
+	for (dev_iter = dev_list; dev_iter; dev_iter = dev_iter->next)
+		if (fd_compare(dev_iter->core->fd, user_dev->user_fd) == 0)
 			break;
 
-	if (dev) {
-		atomic_inc(&dev->refcount);
-		user_dev->core = dev;
+	if (dev_iter) {
+		atomic_inc(&dev_iter->core->refcount);
+		user_dev->core = dev_iter->core;
 		return 0;
 	}
 
@@ -115,9 +129,6 @@ static int amdgpu_device_init(amdgpu_device_handle user_dev)
 	dev->fd = user_dev->user_fd;
 	user_dev->core = dev;
 
-	dev->next = dev_list;
-	dev_list = dev;
-
 	version = drmGetVersion(dev->fd);
 	if (version->version_major != 3) {
 		fprintf(stderr, "%s: DRM version is %d.%d.%d but this driver is "
@@ -184,6 +195,17 @@ drm_public int amdgpu_device_initialize(int fd,
 
 	*device_handle = NULL;
 
+	pthread_mutex_lock(&dev_mutex);
+
+	for (user_dev = dev_list; user_dev; user_dev = user_dev->next) {
+		if (same_file_description(user_dev->user_fd, fd)) {
+			atomic_inc(&user_dev->refcount);
+			goto out;
+		}
+	}
+
+	pthread_mutex_unlock(&dev_mutex);
+
 	user_dev = calloc(1, sizeof(struct amdgpu_device));
 	if (!user_dev) {
 		fprintf(stderr, "%s: calloc failed\n", __func__);
@@ -211,6 +233,11 @@ drm_public int amdgpu_device_initialize(int fd,
 		goto cleanup;
 	}
 
+	atomic_set(&user_dev->refcount, 1);
+	user_dev->next = dev_list;
+	dev_list = user_dev;
+
+out:
 	*major_version = user_dev->core->major_version;
 	*minor_version = user_dev->core->minor_version;
 	*device_handle = user_dev;
@@ -234,14 +261,23 @@ drm_public int amdgpu_device_deinitialize(amdgpu_device_handle user_dev)
 
 	pthread_mutex_lock(&dev_mutex);
 
-	if (user_dev->user_fd != dev->fd)
-		close(user_dev->user_fd);
+	if (update_references(&user_dev->refcount, NULL)) {
+		struct amdgpu_device **node = &dev_list;
 
-	if (update_references(&dev->refcount, NULL))
-		amdgpu_device_free(dev);
+		while (*node != user_dev && (*node)->next)
+			node = &(*node)->next;
+		*node = (*node)->next;
+
+		if (user_dev->user_fd != dev->fd)
+			close(user_dev->user_fd);
+
+		if (update_references(&dev->refcount, NULL))
+			amdgpu_device_free(dev);
+
+		free(user_dev);
+	}
 
 	pthread_mutex_unlock(&dev_mutex);
-	free(user_dev);
 	return 0;
 }
 
diff --git a/amdgpu/amdgpu_internal.h b/amdgpu/amdgpu_internal.h
index a08a4ae8..686d50ec 100644
--- a/amdgpu/amdgpu_internal.h
+++ b/amdgpu/amdgpu_internal.h
@@ -70,7 +70,6 @@ struct amdgpu_core_device {
 	unsigned major_version;
 	unsigned minor_version;
 
-	struct amdgpu_core_device *next;
 	char *marketing_name;
 	/** List of buffer handles. Protected by bo_table_mutex. */
 	struct handle_table bo_handles;
@@ -91,8 +90,10 @@ struct amdgpu_core_device {
 };
 
 struct amdgpu_device {
+	atomic_t refcount;
 	int user_fd;
 	struct amdgpu_core_device *core;
+	struct amdgpu_device *next;
 };
 
 struct amdgpu_core_bo {
-- 
2.20.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH libdrm 7/9] amdgpu: Use normal integers for device / core BO reference counting

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

The dev_mutex / dev->bo_table_mutex is always locked when accessing the
reference counts, so no need for atomic operations.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 amdgpu/amdgpu_bo.c       |  6 +++---
 amdgpu/amdgpu_device.c   | 18 ++++++++++--------
 amdgpu/amdgpu_internal.h |  6 +++---
 3 files changed, 16 insertions(+), 14 deletions(-)

diff --git a/amdgpu/amdgpu_bo.c b/amdgpu/amdgpu_bo.c
index 8d42db90..f40df484 100644
--- a/amdgpu/amdgpu_bo.c
+++ b/amdgpu/amdgpu_bo.c
@@ -65,7 +65,7 @@ static int amdgpu_core_bo_create(struct amdgpu_core_device *dev,
 		return r;
 	}
 
-	atomic_set(&bo->refcount, 1);
+	bo->refcount = 1;
 	bo->alloc_size = size;
 	bo->handle = handle;
 	pthread_mutex_init(&bo->cpu_access_mutex, NULL);
@@ -95,7 +95,7 @@ static int amdgpu_bo_create(amdgpu_device_handle user_dev,
 				goto out;
 			}
 		}
-		atomic_inc(&bo->refcount);
+		bo->refcount++;
 	} else {
 		r = amdgpu_core_bo_create(dev, size, handle, &bo);
 		if (r)
@@ -467,7 +467,7 @@ static void amdgpu_core_bo_free(struct amdgpu_bo *user_bo)
 	struct amdgpu_core_device *dev = user_bo->dev->core;
 	struct amdgpu_core_bo *bo = user_bo->core;
 
-	if (update_references(&bo->refcount, NULL)) {
+	if (--bo->refcount == 0) {
 		/* Remove the buffer from the hash tables. */
 		handle_table_remove(&dev->bo_handles, bo->handle);
 
diff --git a/amdgpu/amdgpu_device.c b/amdgpu/amdgpu_device.c
index 8d9a85c2..8a99b8bb 100644
--- a/amdgpu/amdgpu_device.c
+++ b/amdgpu/amdgpu_device.c
@@ -72,6 +72,9 @@ static int fd_compare(int fd1, int fd2)
 
 static void amdgpu_device_free(struct amdgpu_core_device *dev)
 {
+	if (--dev->refcount > 0)
+		return;
+
 	close(dev->fd);
 
 	amdgpu_vamgr_deinit(&dev->vamgr_32);
@@ -112,7 +115,7 @@ static int amdgpu_device_init(amdgpu_device_handle user_dev)
 			break;
 
 	if (dev_iter) {
-		atomic_inc(&dev_iter->core->refcount);
+		dev_iter->core->refcount++;
 		user_dev->core = dev_iter->core;
 		return 0;
 	}
@@ -123,7 +126,7 @@ static int amdgpu_device_init(amdgpu_device_handle user_dev)
 		return -ENOMEM;
 	}
 
-	atomic_set(&dev->refcount, 1);
+	dev->refcount = 1;
 	pthread_mutex_init(&dev->bo_table_mutex, NULL);
 
 	dev->fd = user_dev->user_fd;
@@ -199,7 +202,7 @@ drm_public int amdgpu_device_initialize(int fd,
 
 	for (user_dev = dev_list; user_dev; user_dev = user_dev->next) {
 		if (same_file_description(user_dev->user_fd, fd)) {
-			atomic_inc(&user_dev->refcount);
+			user_dev->refcount++;
 			goto out;
 		}
 	}
@@ -233,7 +236,7 @@ drm_public int amdgpu_device_initialize(int fd,
 		goto cleanup;
 	}
 
-	atomic_set(&user_dev->refcount, 1);
+	user_dev->refcount = 1;
 	user_dev->next = dev_list;
 	dev_list = user_dev;
 
@@ -248,7 +251,7 @@ out:
 cleanup:
 	if (!user_dev->core || user_dev->user_fd != user_dev->core->fd)
 		close(user_dev->user_fd);
-	if (user_dev->core && update_references(&user_dev->core->refcount, NULL))
+	if (user_dev->core)
 		amdgpu_device_free(user_dev->core);
 	free(user_dev);
 	pthread_mutex_unlock(&dev_mutex);
@@ -261,7 +264,7 @@ drm_public int amdgpu_device_deinitialize(amdgpu_device_handle user_dev)
 
 	pthread_mutex_lock(&dev_mutex);
 
-	if (update_references(&user_dev->refcount, NULL)) {
+	if (--user_dev->refcount == 0) {
 		struct amdgpu_device **node = &dev_list;
 
 		while (*node != user_dev && (*node)->next)
@@ -271,8 +274,7 @@ drm_public int amdgpu_device_deinitialize(amdgpu_device_handle user_dev)
 		if (user_dev->user_fd != dev->fd)
 			close(user_dev->user_fd);
 
-		if (update_references(&dev->refcount, NULL))
-			amdgpu_device_free(dev);
+		amdgpu_device_free(dev);
 
 		free(user_dev);
 	}
diff --git a/amdgpu/amdgpu_internal.h b/amdgpu/amdgpu_internal.h
index 686d50ec..70566f98 100644
--- a/amdgpu/amdgpu_internal.h
+++ b/amdgpu/amdgpu_internal.h
@@ -65,7 +65,7 @@ struct amdgpu_va {
 };
 
 struct amdgpu_core_device {
-	atomic_t refcount;
+	int refcount;
 	int fd;
 	unsigned major_version;
 	unsigned minor_version;
@@ -90,14 +90,14 @@ struct amdgpu_core_device {
 };
 
 struct amdgpu_device {
-	atomic_t refcount;
+	int refcount;
 	int user_fd;
 	struct amdgpu_core_device *core;
 	struct amdgpu_device *next;
 };
 
 struct amdgpu_core_bo {
-	atomic_t refcount;
+	int refcount;
 	amdgpu_bo_handle user_bos;
 
 	uint64_t alloc_size;
-- 
2.20.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH libdrm 0/9] amdgpu:

[Christian König]

Patches #1 - #3 look good to me, but I'm not sure if the rest is such a 
good idea.

Basically you not only want to use the same FD for CS, but also for 
basically all buffer functions and as far as I can see we break that here.

I would rather add a new function to export the KMS handle for a certain 
BO/FD pair. Exporting the handle based on a type was also like trying to 
squeeze a round pig through a square hole in the first place.

KMS, flink and DMA-buf fd are fundamentally different and shouldn't be 
handled by the same function in the first place.

The only tricky part in this scenario is to know when to close the KMS 
handle again.

Christian.

Am 24.06.19 um 18:53 schrieb Michel Dänzer:
> From: Michel Dänzer <michel.daenzer@amd.com>
>
> The motivation for these patches is https://bugs.freedesktop.org/110903 .
>
> Patches 1-3 are preparatory.
>
> Patches 4 & 5 are the core patches allowing the issues discussed in the
> bug report to be fixed.
>
> Patches 6-8 are further optimizations / cleanups.
>
> Patch 9 is the Mesa patch making use of the new
> amdgpu_bo_handle_type_kms_user API to fix the user visible problem.
>
> Note that the libdrm patches need to land first, a libdrm release needs
> to be made, and the Mesa patch needs to bump the libdrm_amdgpu version
> requirement to that release's version before it can land.
>
> Michel Dänzer (9):
>    amdgpu: Pass file descriptor directly to amdgpu_close_kms_handle
>    amdgpu: Add BO handle to table in amdgpu_bo_create
>    amdgpu: Rename fd_mutex/list to dev_mutex/list
>    amdgpu: Add struct amdgpu_core_device and amdgpu_core_bo
>    amdgpu: Add amdgpu_bo_handle_type_kms_user
>    amdgpu: Re-use an existing amdgpu_device when possible
>    amdgpu: Use normal integers for device / core BO reference counting
>    amdgpu: Drop refcount member from struct amdgpu_core_bo/device
>    winsys/amdgpu: Use amdgpu_bo_handle_type_kms_user for API KMS handles
>
>   amdgpu/amdgpu.h               |  14 +-
>   amdgpu/amdgpu_asic_id.c       |   4 +-
>   amdgpu/amdgpu_bo.c            | 367 ++++++++++++++++++++++------------
>   amdgpu/amdgpu_cs.c            |  64 +++---
>   amdgpu/amdgpu_device.c        | 281 +++++++++++++-------------
>   amdgpu/amdgpu_gpu_info.c      |  35 ++--
>   amdgpu/amdgpu_internal.h      |  31 ++-
>   amdgpu/amdgpu_vamgr.c         |   9 +-
>   amdgpu/amdgpu_vm.c            |   4 +-
>   tests/amdgpu/amdgpu_test.c    |   2 +-
>   tests/amdgpu/bo_tests.c       |   2 +-
>   tests/amdgpu/cs_tests.c       |   8 +-
>   tests/amdgpu/deadlock_tests.c |   8 +-
>   tests/amdgpu/uvd_enc_tests.c  |   2 +-
>   tests/amdgpu/vce_tests.c      |  12 +-
>   tests/amdgpu/vcn_tests.c      |   4 +-
>   tests/amdgpu/vm_tests.c       |   2 +-
>   17 files changed, 500 insertions(+), 349 deletions(-)
>

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH libdrm 0/9] amdgpu:

[Michel Dänzer]

On 2019-06-24 7:31 p.m., Christian König wrote:
> Patches #1 - #3 look good to me, but I'm not sure if the rest is such a
> good idea.
> 
> Basically you not only want to use the same FD for CS, but also for
> basically all buffer functions and as far as I can see we break that here.

How so? The core FD is used for everything except flink and
amdgpu_bo_handle_type_kms_user.


> I would rather add a new function to export the KMS handle for a certain
> BO/FD pair. Exporting the handle based on a type was also like trying to
> squeeze a round pig through a square hole in the first place.
> 
> KMS, flink and DMA-buf fd are fundamentally different and shouldn't be
> handled by the same function in the first place.

I don't really see the problem. radeonsi & RADV are using
amdgpu_bo_handle_type_kms to get handles for command stream submission,
so those handles have to be valid for the core FD. Now there's
amdgpu_bo_handle_type_kms_user to get a handle valid for the FD passed
to amdgpu_device_initialize. Can you describe a scenario where a handle
valid for yet another file description is needed?


-- 
Earthling Michel Dänzer               |              https://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH libdrm 0/9] amdgpu:

[Koenig, Christian]

Am 25.06.19 um 10:02 schrieb Michel Dänzer:
> On 2019-06-24 7:31 p.m., Christian König wrote:
>> Patches #1 - #3 look good to me, but I'm not sure if the rest is such a
>> good idea.
>>
>> Basically you not only want to use the same FD for CS, but also for
>> basically all buffer functions and as far as I can see we break that here.
> How so? The core FD is used for everything except flink and
> amdgpu_bo_handle_type_kms_user.

IIRC in the Mesa winsys we compare the amdgpu_device and amdgpu_bo 
pointers to figure out if an opened device or imported BO is the same as 
one we already know.

With this patch that won't work any more and for example OpenGL and 
VA-API could potentially use separate amdgpu_bo pointers for the same 
underlying buffer. That in turn would break synchronization.

Additional to that we potentially could keep a bunch of file descriptors 
around which we don't necessary want. In other words we previously 
closed the extra file descriptors, but now we keep them open as well. 
But that should probably be only a minor problem.

>> I would rather add a new function to export the KMS handle for a certain
>> BO/FD pair. Exporting the handle based on a type was also like trying to
>> squeeze a round pig through a square hole in the first place.
>>
>> KMS, flink and DMA-buf fd are fundamentally different and shouldn't be
>> handled by the same function in the first place.
> I don't really see the problem. radeonsi & RADV are using
> amdgpu_bo_handle_type_kms to get handles for command stream submission,
> so those handles have to be valid for the core FD. Now there's
> amdgpu_bo_handle_type_kms_user to get a handle valid for the FD passed
> to amdgpu_device_initialize. Can you describe a scenario where a handle
> valid for yet another file description is needed?

Not yet another file descriptor, but the underlying problem here seems 
to be that we don't tell libdrm in which domain/file descriptor the KMS 
should actually be valid in.

Regards,
Christian.
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH libdrm 0/9] amdgpu:

[Michel Dänzer]

On 2019-06-25 11:44 a.m., Koenig, Christian wrote:
> Am 25.06.19 um 10:02 schrieb Michel Dänzer:
>> On 2019-06-24 7:31 p.m., Christian König wrote:
>>> Patches #1 - #3 look good to me, but I'm not sure if the rest is such a
>>> good idea.
>>>
>>> Basically you not only want to use the same FD for CS, but also for
>>> basically all buffer functions and as far as I can see we break that here.
>> How so? The core FD is used for everything except flink and
>> amdgpu_bo_handle_type_kms_user.
> 
> IIRC in the Mesa winsys we compare the amdgpu_device and amdgpu_bo 
> pointers to figure out if an opened device or imported BO is the same as 
> one we already know.
> 
> With this patch that won't work any more and for example OpenGL and 
> VA-API could potentially use separate amdgpu_bo pointers for the same 
> underlying buffer. That in turn would break synchronization.

Hmm, I was hoping patch 6 would cover this, but it looks like OpenGL and
VA-API pass file descriptors referencing different file descriptions to
amdgpu_device_initialize, so you're right. :(


>>> I would rather add a new function to export the KMS handle for a certain
>>> BO/FD pair.

I'll try this approach then.


-- 
Earthling Michel Dänzer               |              https://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx