Skip to contentSkip to conversation

amdgpu: stop using the hash table for fd_tab

PATCH In review Christian König 5 participants 14 messages Aug 3, 2018 – Aug 10, 2018 7 messages not in the archive

lore
Share & tools
lore
#0 lore raw

PATCH 1/6 amdgpu: stop using the hash table for fd_tab

We have so few devices that just walking a linked list is probably faster.

---
 amdgpu/amdgpu_device.c   | 49 ++++++++++++++++--------------------------------
 amdgpu/amdgpu_internal.h |  1 +
 2 files changed, 17 insertions(+), 33 deletions(-)
amdgpu/amdgpu_device.c +16 -33
index d7aec6a4..38fd186d 100644
@@ -43,10 +43,9 @@
 #include "util_math.h"
 
 #define PTR_TO_UINT(x) ((unsigned)((intptr_t)(x)))
-#define UINT_TO_PTR(x) ((void *)((intptr_t)(x)))
 
 static pthread_mutex_t fd_mutex = PTHREAD_MUTEX_INITIALIZER;
-static struct util_hash_table *fd_tab;
+static amdgpu_device_handle fd_list;
 
 static unsigned handle_hash(void *key)
 {
@@ -58,28 +57,8 @@ static int handle_compare(void *key1, void *key2)
 	return PTR_TO_UINT(key1) != PTR_TO_UINT(key2);
 }
 
-static unsigned fd_hash(void *key)
+static int fd_compare(int fd1, int fd2)
 {
-	int fd = PTR_TO_UINT(key);
-	char *name = drmGetPrimaryDeviceNameFromFd(fd);
-	unsigned result = 0;
-	char *c;
-
-	if (name == NULL)
-		return 0;
-
-	for (c = name; *c; ++c)
-		result += *c;
-
-	free(name);
-
-	return result;
-}
-
-static int fd_compare(void *key1, void *key2)
-{
-	int fd1 = PTR_TO_UINT(key1);
-	int fd2 = PTR_TO_UINT(key2);
 	char *name1 = drmGetPrimaryDeviceNameFromFd(fd1);
 	char *name2 = drmGetPrimaryDeviceNameFromFd(fd2);
 	int result;
@@ -127,16 +106,17 @@ 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;
+
 	pthread_mutex_lock(&fd_mutex);
-	util_hash_table_remove(fd_tab, UINT_TO_PTR(dev->fd));
-	if (util_hash_table_count(fd_tab) == 0) {
-		util_hash_table_destroy(fd_tab);
-		fd_tab = NULL;
-	}
+	while (*node != dev && (*node)->next)
+		node = &(*node)->next;
+	*node = (*node)->next;
+	pthread_mutex_unlock(&fd_mutex);
+
 	close(dev->fd);
 	if ((dev->flink_fd >= 0) && (dev->fd != dev->flink_fd))
 		close(dev->flink_fd);
-	pthread_mutex_unlock(&fd_mutex);
 
 	amdgpu_vamgr_deinit(&dev->vamgr_32);
 	amdgpu_vamgr_deinit(&dev->vamgr);
@@ -187,8 +167,6 @@ int amdgpu_device_initialize(int fd,
 	*device_handle = NULL;
 
 	pthread_mutex_lock(&fd_mutex);
-	if (!fd_tab)
-		fd_tab = util_hash_table_create(fd_hash, fd_compare);
 	r = amdgpu_get_auth(fd, &flag_auth);
 	if (r) {
 		fprintf(stderr, "%s: amdgpu_get_auth (1) failed (%i)\n",
@@ -196,7 +174,11 @@ int amdgpu_device_initialize(int fd,
 		pthread_mutex_unlock(&fd_mutex);
 		return r;
 	}
-	dev = util_hash_table_get(fd_tab, UINT_TO_PTR(fd));
+
+	for (dev = fd_list; dev; dev = dev->next)
+		if (fd_compare(dev->fd, fd) == 0)
+			break;
+
 	if (dev) {
 		r = amdgpu_get_auth(dev->fd, &flag_authexist);
 		if (r) {
@@ -297,7 +279,8 @@ int amdgpu_device_initialize(int fd,
 	*major_version = dev->major_version;
 	*minor_version = dev->minor_version;
 	*device_handle = dev;
-	util_hash_table_set(fd_tab, UINT_TO_PTR(dev->fd), dev);
+	dev->next = fd_list;
+	fd_list = dev;
 	pthread_mutex_unlock(&fd_mutex);
 
 	return 0;
amdgpu/amdgpu_internal.h +1 -0
index 99b8ce0b..83012cab 100644
@@ -65,6 +65,7 @@ struct amdgpu_va {
 
 struct amdgpu_device {
 	atomic_t refcount;
+	struct amdgpu_device *next;
 	int fd;
 	int flink_fd;
 	unsigned major_version;
Signature · 6 lines
2.14.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 20180803113458.28560-1-christian.koenig-5C7GfCeVMHo@public.gmane.org

#2 lore raw

PATCH 2/6 amdgpu: add handle table implementation v2

The kernel handles are dense and the kernel always tries to use the lowest free id. Use this to implement a more efficient handle table by using a resizeable array instead of a hash.

v2: add handle_table_fini function, extra key checks,
fix typo in function name

---
 amdgpu/Makefile.sources |  4 ++-
 amdgpu/handle_table.c   | 72 +++++++++++++++++++++++++++++++++++++++++++++++++
 amdgpu/handle_table.h   | 41 ++++++++++++++++++++++++++++
 3 files changed, 116 insertions(+), 1 deletion(-)
 create mode 100644 amdgpu/handle_table.c
 create mode 100644 amdgpu/handle_table.h
amdgpu/Makefile.sources +3 -1
index 498b64cc..62577ba5 100644
@@ -10,7 +10,9 @@ LIBDRM_AMDGPU_FILES := \
 	util_hash.c \
 	util_hash.h \
 	util_hash_table.c \
-	util_hash_table.h
+	util_hash_table.h \
+	handle_table.c \
+	handle_table.h
 
 LIBDRM_AMDGPU_H_FILES := \
 	amdgpu.h
amdgpu/handle_table.c new file +72 -0

72 lines not on this page · raw message

amdgpu/handle_table.h new file +41 -0

41 lines not on this page · raw message

Signature · 6 lines
2.14.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 20180803113458.28560-2-christian.koenig-5C7GfCeVMHo@public.gmane.org

#4 lore raw

Re: [PATCH libdrm 2/6] amdgpu: add handle table implementation v2

Quoted 63 lines · Christian König

The kernel handles are dense and the kernel always tries to use the lowest free id. Use this to implement a more efficient handle table by using a resizeable array instead of a hash.

v2: add handle_table_fini function, extra key checks,
fix typo in function name

---
  amdgpu/Makefile.sources |  4 ++-
  amdgpu/handle_table.c   | 72 +++++++++++++++++++++++++++++++++++++++++++++++++
  amdgpu/handle_table.h   | 41 ++++++++++++++++++++++++++++
  3 files changed, 116 insertions(+), 1 deletion(-)

create mode 100644 amdgpu/handle_table.c
create mode 100644 amdgpu/handle_table.h

amdgpu/Makefile.sources +3 -1
index 498b64cc..62577ba5 100644
@@ -10,7 +10,9 @@ LIBDRM_AMDGPU_FILES := \
  	util_hash.c \
  	util_hash.h \
  	util_hash_table.c \
-	util_hash_table.h
+	util_hash_table.h \
+	handle_table.c \
+	handle_table.h
 
  LIBDRM_AMDGPU_H_FILES := \
  	amdgpu.h
amdgpu/handle_table.c new file +26 -0
new file mode 100644
index 00000000..15cd4763
@@ -0,0 +1,72 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <stdlib.h>
+#include <string.h>
+#include <errno.h>

amdgpu/handle_table.c line 36

Quoted 10 lines
amdgpu/handle_table.c new file +10 -0
+#include <unistd.h>
+#include "handle_table.h"
+#include "util_math.h"
+
+drm_private int handle_table_insert(struct handle_table *table, uint32_t key,
+				    void *value)
+{
+	if (key >= table->max_key) {
+		uint32_t alignment = sysconf(_SC_PAGESIZE) / sizeof(void*);
+		uint32_t max_key = ALIGN(key, alignment);

On 2nd thought, we may add table->count to take the role of current table->max_key, and table_max_key could be updated by key rather than aligned value.

Then we could traverse the table by max_key instead of the entire table.

Regards,
Jerry

Quoted 74 lines
amdgpu/handle_table.c new file +36 -0
+		void **values;
+
+		values = realloc(table->values, max_key * sizeof(void *));
+		if (!values)
+			return -ENOMEM;
+
+		memset(values + table->max_key, 0, (max_key - table->max_key) *
+		       sizeof(void *));
+
+		table->max_key = max_key;
+		table->values = values;
+	}
+	table->values[key] = value;
+	return 0;
+}
+
+drm_private void handle_table_remove(struct handle_table *table, uint32_t key)
+{
+	if (key < table->max_key)
+		table->values[key] = NULL;
+}
+
+drm_private void *handle_table_lookup(struct handle_table *table, uint32_t key)
+{
+	if (key < table->max_key)
+		return table->values[key];
+	else
+		return NULL;
+}
+
+drm_private void handle_table_fini(struct handle_table *table)
+{
+	free(table->values);
+	table->max_key = 0;
+	table->values = NULL;
+}
amdgpu/handle_table.h new file +31 -0
new file mode 100644
index 00000000..461193f6
@@ -0,0 +1,41 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _HANDLE_TABLE_H_
+#define _HANDLE_TABLE_H_
+
+#include <stdint.h>
+#include "libdrm_macros.h"
+
+struct handle_table {
+	uint32_t	max_key;

amdgpu/handle_table.h line 41

Quoted 10 lines
amdgpu/handle_table.h new file +10 -0
+	void		**values;
+};
+
+drm_private int handle_table_insert(struct handle_table *table, uint32_t key,
+				    void *value);
+drm_private void handle_table_remove(struct handle_table *table, uint32_t key);
+drm_private void *handle_table_lookup(struct handle_table *table, uint32_t key);
+drm_private void handle_table_fini(struct handle_table *table);
+
+#endif /* _HANDLE_TABLE_H_ */

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

#5 lore raw

PATCH 3/6 amdgpu: use handle table for KMS handles

In reply to a message not in the archive

Instead of the hash use the handle table.

---
 amdgpu/amdgpu_bo.c       | 19 ++++++++++---------
 amdgpu/amdgpu_device.c   |  3 +--
 amdgpu/amdgpu_internal.h |  3 ++-
 3 files changed, 13 insertions(+), 12 deletions(-)
amdgpu/amdgpu_bo.c +10 -9
index d29be244..66edb8d4 100644
@@ -90,8 +90,12 @@ int amdgpu_bo_alloc(amdgpu_device_handle dev,
 
 	pthread_mutex_init(&bo->cpu_access_mutex, NULL);
 
-	*buf_handle = bo;
-	return 0;
+	if (r)
+		amdgpu_bo_free(bo);
+	else
+		*buf_handle = bo;
+
+	return r;
 }
 
 int amdgpu_bo_set_metadata(amdgpu_bo_handle bo,
@@ -171,8 +175,7 @@ int amdgpu_bo_query_info(amdgpu_bo_handle bo,
 static void amdgpu_add_handle_to_table(amdgpu_bo_handle bo)
 {
 	pthread_mutex_lock(&bo->dev->bo_table_mutex);
-	util_hash_table_set(bo->dev->bo_handles,
-			    (void*)(uintptr_t)bo->handle, bo);
+	handle_table_insert(&bo->dev->bo_handles, bo->handle, bo);
 	pthread_mutex_unlock(&bo->dev->bo_table_mutex);
 }
 
@@ -303,8 +306,7 @@ int amdgpu_bo_import(amdgpu_device_handle dev,
 		break;
 
 	case amdgpu_bo_handle_type_dma_buf_fd:
-		bo = util_hash_table_get(dev->bo_handles,
-					 (void*)(uintptr_t)shared_handle);
+		bo = handle_table_lookup(&dev->bo_handles, shared_handle);
 		break;
 
 	case amdgpu_bo_handle_type_kms:
@@ -387,7 +389,7 @@ int amdgpu_bo_import(amdgpu_device_handle dev,
 	bo->dev = dev;
 	pthread_mutex_init(&bo->cpu_access_mutex, NULL);
 
-	util_hash_table_set(dev->bo_handles, (void*)(uintptr_t)bo->handle, bo);
+	handle_table_insert(&dev->bo_handles, bo->handle, bo);
 	pthread_mutex_unlock(&dev->bo_table_mutex);
 
 	output->buf_handle = bo;
@@ -406,8 +408,7 @@ int amdgpu_bo_free(amdgpu_bo_handle buf_handle)
 
 	if (update_references(&bo->refcount, NULL)) {
 		/* Remove the buffer from the hash tables. */
-		util_hash_table_remove(dev->bo_handles,
-					(void*)(uintptr_t)bo->handle);
+		handle_table_remove(&dev->bo_handles, bo->handle);
 
 		if (bo->flink_name) {
 			util_hash_table_remove(dev->bo_flink_names,
amdgpu/amdgpu_device.c +1 -2
index 38fd186d..824c1dae 100644
@@ -122,8 +122,8 @@ static void amdgpu_device_free_internal(amdgpu_device_handle dev)
 	amdgpu_vamgr_deinit(&dev->vamgr);
 	amdgpu_vamgr_deinit(&dev->vamgr_high_32);
 	amdgpu_vamgr_deinit(&dev->vamgr_high);
+	handle_table_fini(&dev->bo_handles);
 	util_hash_table_destroy(dev->bo_flink_names);
-	util_hash_table_destroy(dev->bo_handles);
 	pthread_mutex_destroy(&dev->bo_table_mutex);
 	free(dev->marketing_name);
 	free(dev);
@@ -230,7 +230,6 @@ int amdgpu_device_initialize(int fd,
 
 	dev->bo_flink_names = util_hash_table_create(handle_hash,
 						     handle_compare);
-	dev->bo_handles = util_hash_table_create(handle_hash, handle_compare);
 	pthread_mutex_init(&dev->bo_table_mutex, NULL);
 
 	/* Check if acceleration is working. */
amdgpu/amdgpu_internal.h +2 -1
index 83012cab..36ebc738 100644
@@ -32,6 +32,7 @@
 #include "xf86atomic.h"
 #include "amdgpu.h"
 #include "util_double_list.h"
+#include "handle_table.h"
 
 #define AMDGPU_CS_MAX_RINGS 8
 /* do not use below macro if b is not power of 2 aligned value */
@@ -73,7 +74,7 @@ struct amdgpu_device {
 
 	char *marketing_name;
 	/** List of buffer handles. Protected by bo_table_mutex. */
-	struct util_hash_table *bo_handles;
+	struct handle_table bo_handles;
 	/** List of buffer GEM flink names. Protected by bo_table_mutex. */
 	struct util_hash_table *bo_flink_names;
 	/** This protects all hash tables. */
Signature · 6 lines
2.14.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
#6 lore raw

PATCH 4/6 amdgpu: use handle table for flink names

In reply to a message not in the archive

Instead of the hash use the handle table.

---
 amdgpu/amdgpu_bo.c       | 26 +++++++++++++-------------
 amdgpu/amdgpu_device.c   | 15 +--------------
 amdgpu/amdgpu_internal.h |  2 +-
 3 files changed, 15 insertions(+), 28 deletions(-)
amdgpu/amdgpu_bo.c +13 -13
index 66edb8d4..02592377 100644
@@ -37,7 +37,6 @@
 #include "xf86drm.h"
 #include "amdgpu_drm.h"
 #include "amdgpu_internal.h"
-#include "util_hash_table.h"
 #include "util_math.h"
 
 static void amdgpu_close_kms_handle(amdgpu_device_handle dev,
@@ -219,12 +218,10 @@ static int amdgpu_bo_export_flink(amdgpu_bo_handle bo)
 	}
 
 	pthread_mutex_lock(&bo->dev->bo_table_mutex);
-	util_hash_table_set(bo->dev->bo_flink_names,
-			    (void*)(uintptr_t)bo->flink_name,
-			    bo);
+	r = handle_table_insert(&bo->dev->bo_flink_names, bo->flink_name, bo);
 	pthread_mutex_unlock(&bo->dev->bo_table_mutex);
 
-	return 0;
+	return r;
 }
 
 int amdgpu_bo_export(amdgpu_bo_handle bo,
@@ -301,8 +298,7 @@ int amdgpu_bo_import(amdgpu_device_handle dev,
 	/* If we have already created a buffer with this handle, find it. */
 	switch (type) {
 	case amdgpu_bo_handle_type_gem_flink_name:
-		bo = util_hash_table_get(dev->bo_flink_names,
-					 (void*)(uintptr_t)shared_handle);
+		bo = handle_table_lookup(&dev->bo_flink_names, shared_handle);
 		break;
 
 	case amdgpu_bo_handle_type_dma_buf_fd:
@@ -370,8 +366,13 @@ int amdgpu_bo_import(amdgpu_device_handle dev,
 		}
 		bo->flink_name = shared_handle;
 		bo->alloc_size = open_arg.size;
-		util_hash_table_set(dev->bo_flink_names,
-				    (void*)(uintptr_t)bo->flink_name, bo);
+		r = handle_table_insert(&dev->bo_flink_names, shared_handle,
+					bo);
+		if (r) {
+			pthread_mutex_unlock(&dev->bo_table_mutex);
+			amdgpu_bo_free(bo);
+			return r;
+		}
 		break;
 
 	case amdgpu_bo_handle_type_dma_buf_fd:
@@ -410,10 +411,9 @@ int amdgpu_bo_free(amdgpu_bo_handle buf_handle)
 		/* Remove the buffer from the hash tables. */
 		handle_table_remove(&dev->bo_handles, bo->handle);
 
-		if (bo->flink_name) {
-			util_hash_table_remove(dev->bo_flink_names,
-						(void*)(uintptr_t)bo->flink_name);
-		}
+		if (bo->flink_name)
+			handle_table_remove(&dev->bo_flink_names,
+					    bo->flink_name);
 
 		/* Release CPU access. */
 		if (bo->cpu_map_count > 0) {
amdgpu/amdgpu_device.c +1 -14
index 824c1dae..73eeeed6 100644
@@ -39,7 +39,6 @@
 #include "xf86drm.h"
 #include "amdgpu_drm.h"
 #include "amdgpu_internal.h"
-#include "util_hash_table.h"
 #include "util_math.h"
 
 #define PTR_TO_UINT(x) ((unsigned)((intptr_t)(x)))
@@ -47,16 +46,6 @@
 static pthread_mutex_t fd_mutex = PTHREAD_MUTEX_INITIALIZER;
 static amdgpu_device_handle fd_list;
 
-static unsigned handle_hash(void *key)
-{
-	return PTR_TO_UINT(key);
-}
-
-static int handle_compare(void *key1, void *key2)
-{
-	return PTR_TO_UINT(key1) != PTR_TO_UINT(key2);
-}
-
 static int fd_compare(int fd1, int fd2)
 {
 	char *name1 = drmGetPrimaryDeviceNameFromFd(fd1);
@@ -123,7 +112,7 @@ static void amdgpu_device_free_internal(amdgpu_device_handle dev)
 	amdgpu_vamgr_deinit(&dev->vamgr_high_32);
 	amdgpu_vamgr_deinit(&dev->vamgr_high);
 	handle_table_fini(&dev->bo_handles);
-	util_hash_table_destroy(dev->bo_flink_names);
+	handle_table_fini(&dev->bo_flink_names);
 	pthread_mutex_destroy(&dev->bo_table_mutex);
 	free(dev->marketing_name);
 	free(dev);
@@ -228,8 +217,6 @@ int amdgpu_device_initialize(int fd,
 	dev->minor_version = version->version_minor;
 	drmFreeVersion(version);
 
-	dev->bo_flink_names = util_hash_table_create(handle_hash,
-						     handle_compare);
 	pthread_mutex_init(&dev->bo_table_mutex, NULL);
 
 	/* Check if acceleration is working. */
amdgpu/amdgpu_internal.h +1 -1
index 36ebc738..a340abbd 100644
@@ -76,7 +76,7 @@ struct amdgpu_device {
 	/** List of buffer handles. Protected by bo_table_mutex. */
 	struct handle_table bo_handles;
 	/** List of buffer GEM flink names. Protected by bo_table_mutex. */
-	struct util_hash_table *bo_flink_names;
+	struct handle_table bo_flink_names;
 	/** This protects all hash tables. */
 	pthread_mutex_t bo_table_mutex;
 	struct drm_amdgpu_info_device dev_info;
Signature · 6 lines
2.14.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
#7 lore raw

PATCH 5/6 amdgpu: remove the hash table implementation

In reply to a message not in the archive

Not used any more.

---
 amdgpu/Makefile.sources  |   4 -
 amdgpu/util_hash.c       | 383 -----------------------------------------------
 amdgpu/util_hash.h       | 103 -------------
 amdgpu/util_hash_table.c | 270 ---------------------------------
 amdgpu/util_hash_table.h |  71 ---------
 5 files changed, 831 deletions(-)
 delete mode 100644 amdgpu/util_hash.c
 delete mode 100644 amdgpu/util_hash.h
 delete mode 100644 amdgpu/util_hash_table.c
 delete mode 100644 amdgpu/util_hash_table.h
amdgpu/Makefile.sources +0 -4
index 62577ba5..d6df324a 100644
@@ -7,10 +7,6 @@ LIBDRM_AMDGPU_FILES := \
 	amdgpu_internal.h \
 	amdgpu_vamgr.c \
 	amdgpu_vm.c \
-	util_hash.c \
-	util_hash.h \
-	util_hash_table.c \
-	util_hash_table.h \
 	handle_table.c \
 	handle_table.h
 
amdgpu/util_hash.c deleted +0 -383

383 lines not on this page · raw message

amdgpu/util_hash.h deleted +0 -103

103 lines not on this page · raw message

amdgpu/util_hash_table.c deleted +0 -270

270 lines not on this page · raw message

amdgpu/util_hash_table.h deleted +0 -71

71 lines not on this page · raw message

Signature · 6 lines
2.14.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
#8 lore raw

PATCH 6/6 amdgpu: always add all BOs to lockup table

In reply to a message not in the archive

This way we can always find a BO structure by its handle.

---
 amdgpu/amdgpu_bo.c | 14 ++++----------
 1 file changed, 4 insertions(+), 10 deletions(-)
amdgpu/amdgpu_bo.c +4 -10
index 02592377..422c7c99 100644
@@ -87,6 +87,10 @@ int amdgpu_bo_alloc(amdgpu_device_handle dev,
 
 	bo->handle = args.out.handle;
 
+	pthread_mutex_lock(&bo->dev->bo_table_mutex);
+	r = handle_table_insert(&bo->dev->bo_handles, bo->handle, bo);
+	pthread_mutex_unlock(&bo->dev->bo_table_mutex);
+
 	pthread_mutex_init(&bo->cpu_access_mutex, NULL);
 
 	if (r)
@@ -171,13 +175,6 @@ int amdgpu_bo_query_info(amdgpu_bo_handle bo,
 	return 0;
 }
 
-static void amdgpu_add_handle_to_table(amdgpu_bo_handle bo)
-{
-	pthread_mutex_lock(&bo->dev->bo_table_mutex);
-	handle_table_insert(&bo->dev->bo_handles, bo->handle, bo);
-	pthread_mutex_unlock(&bo->dev->bo_table_mutex);
-}
-
 static int amdgpu_bo_export_flink(amdgpu_bo_handle bo)
 {
 	struct drm_gem_flink flink;
@@ -240,14 +237,11 @@ int amdgpu_bo_export(amdgpu_bo_handle bo,
 		return 0;
 
 	case amdgpu_bo_handle_type_kms:
-		amdgpu_add_handle_to_table(bo);
-		/* fall through */
 	case amdgpu_bo_handle_type_kms_noimport:
 		*shared_handle = bo->handle;
 		return 0;
 
 	case amdgpu_bo_handle_type_dma_buf_fd:
-		amdgpu_add_handle_to_table(bo);
 		return drmPrimeHandleToFD(bo->dev->fd, bo->handle,
 					  DRM_CLOEXEC | DRM_RDWR,
 					  (int*)shared_handle);
Signature · 6 lines
2.14.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 20180803113458.28560-6-christian.koenig-5C7GfCeVMHo@public.gmane.org

Reviewed-by
#10 lore raw

Re: [PATCH libdrm 6/6] amdgpu: always add all BOs to lockup table

Quoted 3 lines · Christian König

This way we can always find a BO structure by its handle.

In the shortlog, should be "handle table" instead of "lockup table"?

With that fixed, the series is

Thanks for doing this!

Signature · 6 lines
Earthling Michel Dänzer               |               http://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 98a57d99-4016-8896-1791-d5550cc2c277-otUistvHUpPR7s880joybQ@public.gmane.org

#12 lore raw

Re: [PATCH libdrm 6/6] amdgpu: always add all BOs to lockup table

Am 03.08.2018 um 15:52 schrieb Michel Dänzer:

Quoted 12 lines · Christian König
Quoted 3 lines

This way we can always find a BO structure by its handle.

In the shortlog, should be "handle table" instead of "lockup table"?

With that fixed, the series is

Thanks for doing this!

No, problem. Could you fix up the subject line and push it upstream?

I still don't have a working fdo account again :( Alternatively do you know an admin I can ping about that?

Thanks,
Christian.
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): faabbb5b-fc6a-3c7e-a051-ce4283f45241-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

#14 lore raw

Re: [PATCH libdrm 6/6] amdgpu: always add all BOs to lockup table

Ping ajax or daniels or anholt on IRC.

Alex

________________________________
From: amd-gfx <amd-gfx-bounces-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org> on behalf of Christian König <[email protected]>
Sent: Friday, August 3, 2018 9:54:42 AM
To: Michel Dänzer
Cc: [email protected]
Subject: Re: [PATCH libdrm 6/6] amdgpu: always add all BOs to lockup table

Am 03.08.2018 um 15:52 schrieb Michel Dänzer:

Quoted 12 lines · Christian König
Quoted 3 lines

This way we can always find a BO structure by its handle.

In the shortlog, should be "handle table" instead of "lockup table"?

With that fixed, the series is

Thanks for doing this!

No, problem. Could you fix up the subject line and push it upstream?

I still don't have a working fdo account again :( Alternatively do you know an admin I can ping about that?

Thanks,
Christian.
_______________________________________________
amd-gfx mailing list [email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

#15 lore raw

Re: [PATCH libdrm 6/6] amdgpu: always add all BOs to lockup table

In reply to a message not in the archive

I don't think this is a good idea. Can you please explain why this won't cause performance regressions?

Thanks
Marek

Quoted 54 lines · Christian König

This way we can always find a BO structure by its handle.

---
 amdgpu/amdgpu_bo.c | 14 ++++----------
 1 file changed, 4 insertions(+), 10 deletions(-)
amdgpu/amdgpu_bo.c +4 -11
index 02592377..422c7c99 100644
@@ -87,6 +87,10 @@ int amdgpu_bo_alloc(amdgpu_device_handle dev,
 
        bo->handle = args.out.handle;
 
+       pthread_mutex_lock(&bo->dev->bo_table_mutex);
+       r = handle_table_insert(&bo->dev->bo_handles, bo->handle, bo);
+       pthread_mutex_unlock(&bo->dev->bo_table_mutex);
+
        pthread_mutex_init(&bo->cpu_access_mutex, NULL);
 
        if (r)
@@ -171,13 +175,6 @@ int amdgpu_bo_query_info(amdgpu_bo_handle bo,
        return 0;
 }
 
-static void amdgpu_add_handle_to_table(amdgpu_bo_handle bo)
-{
-       pthread_mutex_lock(&bo->dev->bo_table_mutex);
-       handle_table_insert(&bo->dev->bo_handles, bo->handle, bo);
-       pthread_mutex_unlock(&bo->dev->bo_table_mutex);
-}
-
 static int amdgpu_bo_export_flink(amdgpu_bo_handle bo)
 {
        struct drm_gem_flink flink;
@@ -240,14 +237,11 @@ int amdgpu_bo_export(amdgpu_bo_handle bo,
                return 0;
 
        case amdgpu_bo_handle_type_kms:
-               amdgpu_add_handle_to_table(bo);
-               /* fall through */
        case amdgpu_bo_handle_type_kms_noimport:
                *shared_handle = bo->handle;
                return 0;
 
        case amdgpu_bo_handle_type_dma_buf_fd:
-               amdgpu_add_handle_to_table(bo);
                return drmPrimeHandleToFD(bo->dev->fd, bo->handle,
                                          DRM_CLOEXEC | DRM_RDWR,
                                          (int*)shared_handle);
--

2.14.1

Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): CAAxE2A51vBrphE3CVPVeKixKpxS3q0vqJ8NUkMfp+bM1M3qHbA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org

#17 lore raw

Re: [PATCH libdrm 6/6] amdgpu: always add all BOs to lockup table

Why should it? Adding the handle is now not more than setting an array entry.

I've tested with allocating 250k BOs of 4k size each and there wasn't any measurable performance differences.

Christian.

Am 09.08.2018 um 18:56 schrieb Marek Olšák:

Quoted 62 lines

I don't think this is a good idea. Can you please explain why this won't cause performance regressions?

Thanks
Marek

Quoted 54 lines · Christian König

This way we can always find a BO structure by its handle.

---
  amdgpu/amdgpu_bo.c | 14 ++++----------
  1 file changed, 4 insertions(+), 10 deletions(-)
amdgpu/amdgpu_bo.c +4 -11
index 02592377..422c7c99 100644
@@ -87,6 +87,10 @@ int amdgpu_bo_alloc(amdgpu_device_handle dev,
 
         bo->handle = args.out.handle;
 
+       pthread_mutex_lock(&bo->dev->bo_table_mutex);
+       r = handle_table_insert(&bo->dev->bo_handles, bo->handle, bo);
+       pthread_mutex_unlock(&bo->dev->bo_table_mutex);
+
         pthread_mutex_init(&bo->cpu_access_mutex, NULL);
 
         if (r)
@@ -171,13 +175,6 @@ int amdgpu_bo_query_info(amdgpu_bo_handle bo,
         return 0;
  }
 
-static void amdgpu_add_handle_to_table(amdgpu_bo_handle bo)
-{
-       pthread_mutex_lock(&bo->dev->bo_table_mutex);
-       handle_table_insert(&bo->dev->bo_handles, bo->handle, bo);
-       pthread_mutex_unlock(&bo->dev->bo_table_mutex);
-}
-
  static int amdgpu_bo_export_flink(amdgpu_bo_handle bo)
  {
         struct drm_gem_flink flink;
@@ -240,14 +237,11 @@ int amdgpu_bo_export(amdgpu_bo_handle bo,
                 return 0;
 
         case amdgpu_bo_handle_type_kms:
-               amdgpu_add_handle_to_table(bo);
-               /* fall through */
         case amdgpu_bo_handle_type_kms_noimport:
                 *shared_handle = bo->handle;
                 return 0;
 
         case amdgpu_bo_handle_type_dma_buf_fd:
-               amdgpu_add_handle_to_table(bo);
                 return drmPrimeHandleToFD(bo->dev->fd, bo->handle,
                                           DRM_CLOEXEC | DRM_RDWR,
                                           (int*)shared_handle);
--

2.14.1

Quoted 4 lines
Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 7c186502-199a-88cc-e5cf-6d3106bcc184-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

#19 lore raw

Re: [PATCH libdrm 6/6] amdgpu: always add all BOs to lockup table

OK. Thanks.

Marek

Quoted 74 lines · Christian König

Why should it? Adding the handle is now not more than setting an array entry.

I've tested with allocating 250k BOs of 4k size each and there wasn't any measurable performance differences.

Christian.

Am 09.08.2018 um 18:56 schrieb Marek Olšák:

Quoted 64 lines

I don't think this is a good idea. Can you please explain why this won't cause performance regressions?

Thanks
Marek

Quoted 55 lines · Christian König

This way we can always find a BO structure by its handle.

---
  amdgpu/amdgpu_bo.c | 14 ++++----------
  1 file changed, 4 insertions(+), 10 deletions(-)
amdgpu/amdgpu_bo.c +4 -11
index 02592377..422c7c99 100644
@@ -87,6 +87,10 @@ int amdgpu_bo_alloc(amdgpu_device_handle dev,
 
         bo->handle = args.out.handle;
 
+       pthread_mutex_lock(&bo->dev->bo_table_mutex);
+       r = handle_table_insert(&bo->dev->bo_handles, bo->handle, bo);
+       pthread_mutex_unlock(&bo->dev->bo_table_mutex);
+
         pthread_mutex_init(&bo->cpu_access_mutex, NULL);
 
         if (r)
@@ -171,13 +175,6 @@ int amdgpu_bo_query_info(amdgpu_bo_handle bo,
         return 0;
  }
 
-static void amdgpu_add_handle_to_table(amdgpu_bo_handle bo)
-{
-       pthread_mutex_lock(&bo->dev->bo_table_mutex);
-       handle_table_insert(&bo->dev->bo_handles, bo->handle, bo);
-       pthread_mutex_unlock(&bo->dev->bo_table_mutex);
-}
-
  static int amdgpu_bo_export_flink(amdgpu_bo_handle bo)
  {
         struct drm_gem_flink flink;
@@ -240,14 +237,11 @@ int amdgpu_bo_export(amdgpu_bo_handle bo,
                 return 0;
 
         case amdgpu_bo_handle_type_kms:
-               amdgpu_add_handle_to_table(bo);
-               /* fall through */
         case amdgpu_bo_handle_type_kms_noimport:
                 *shared_handle = bo->handle;
                 return 0;
 
         case amdgpu_bo_handle_type_dma_buf_fd:
-               amdgpu_add_handle_to_table(bo);
                 return drmPrimeHandleToFD(bo->dev->fd, bo->handle,
                                           DRM_CLOEXEC | DRM_RDWR,
                                           (int*)shared_handle);
--

2.14.1

Quoted 6 lines
Quoted 4 lines
Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

#20 lore raw
In reply to a message not in the archive
Quoted 4 lines · Christian König

We have so few devices that just walking a linked list is probably faster.

Series is

Reviewed-and-Tested-by Junwei Zhang <[email protected]>
Quoted 113 lines
---
  amdgpu/amdgpu_device.c   | 49 ++++++++++++++++--------------------------------
  amdgpu/amdgpu_internal.h |  1 +
  2 files changed, 17 insertions(+), 33 deletions(-)
amdgpu/amdgpu_device.c +16 -33
index d7aec6a4..38fd186d 100644
@@ -43,10 +43,9 @@
  #include "util_math.h"
 
  #define PTR_TO_UINT(x) ((unsigned)((intptr_t)(x)))
-#define UINT_TO_PTR(x) ((void *)((intptr_t)(x)))
 
  static pthread_mutex_t fd_mutex = PTHREAD_MUTEX_INITIALIZER;
-static struct util_hash_table *fd_tab;
+static amdgpu_device_handle fd_list;
 
  static unsigned handle_hash(void *key)
  {
@@ -58,28 +57,8 @@ static int handle_compare(void *key1, void *key2)
  	return PTR_TO_UINT(key1) != PTR_TO_UINT(key2);
  }
 
-static unsigned fd_hash(void *key)
+static int fd_compare(int fd1, int fd2)
  {
-	int fd = PTR_TO_UINT(key);
-	char *name = drmGetPrimaryDeviceNameFromFd(fd);
-	unsigned result = 0;
-	char *c;
-
-	if (name == NULL)
-		return 0;
-
-	for (c = name; *c; ++c)
-		result += *c;
-
-	free(name);
-
-	return result;
-}
-
-static int fd_compare(void *key1, void *key2)
-{
-	int fd1 = PTR_TO_UINT(key1);
-	int fd2 = PTR_TO_UINT(key2);
  	char *name1 = drmGetPrimaryDeviceNameFromFd(fd1);
  	char *name2 = drmGetPrimaryDeviceNameFromFd(fd2);
  	int result;
@@ -127,16 +106,17 @@ 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;
+
  	pthread_mutex_lock(&fd_mutex);
-	util_hash_table_remove(fd_tab, UINT_TO_PTR(dev->fd));
-	if (util_hash_table_count(fd_tab) == 0) {
-		util_hash_table_destroy(fd_tab);
-		fd_tab = NULL;
-	}
+	while (*node != dev && (*node)->next)
+		node = &(*node)->next;
+	*node = (*node)->next;
+	pthread_mutex_unlock(&fd_mutex);
+
  	close(dev->fd);
  	if ((dev->flink_fd >= 0) && (dev->fd != dev->flink_fd))
  		close(dev->flink_fd);
-	pthread_mutex_unlock(&fd_mutex);
 
  	amdgpu_vamgr_deinit(&dev->vamgr_32);
  	amdgpu_vamgr_deinit(&dev->vamgr);
@@ -187,8 +167,6 @@ int amdgpu_device_initialize(int fd,
  	*device_handle = NULL;
 
  	pthread_mutex_lock(&fd_mutex);
-	if (!fd_tab)
-		fd_tab = util_hash_table_create(fd_hash, fd_compare);
  	r = amdgpu_get_auth(fd, &flag_auth);
  	if (r) {
  		fprintf(stderr, "%s: amdgpu_get_auth (1) failed (%i)\n",
@@ -196,7 +174,11 @@ int amdgpu_device_initialize(int fd,
  		pthread_mutex_unlock(&fd_mutex);
  		return r;
  	}
-	dev = util_hash_table_get(fd_tab, UINT_TO_PTR(fd));
+
+	for (dev = fd_list; dev; dev = dev->next)
+		if (fd_compare(dev->fd, fd) == 0)
+			break;
+
  	if (dev) {
  		r = amdgpu_get_auth(dev->fd, &flag_authexist);
  		if (r) {
@@ -297,7 +279,8 @@ int amdgpu_device_initialize(int fd,
  	*major_version = dev->major_version;
  	*minor_version = dev->minor_version;
  	*device_handle = dev;
-	util_hash_table_set(fd_tab, UINT_TO_PTR(dev->fd), dev);
+	dev->next = fd_list;
+	fd_list = dev;
  	pthread_mutex_unlock(&fd_mutex);
 
  	return 0;
amdgpu/amdgpu_internal.h +0 -0
index 99b8ce0b..83012cab 100644

amdgpu/amdgpu_internal.h struct amdgpu_va { · line 71

Quoted 7 lines
amdgpu/amdgpu_internal.h +1 -0
@@ -65,6 +65,7 @@ struct amdgpu_va {
 
  struct amdgpu_device {
  	atomic_t refcount;
+	struct amdgpu_device *next;
  	int fd;
  	int flink_fd;
  	unsigned major_version;

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx