Skip to contentSkip to conversation

Implement sw_sync test

PATCH Superseded [email protected] 2 participants 18 messages Sep 12, 2016 – Sep 14, 2016 Library code

v2 of 11 Superseded Latest is v11 →

lore
Share & tools
lore
#0 lore raw

v2 cover Implement sw_sync test

From: Robert Foss <[email protected]>

This is a resubmission of v2.

This series implements the sw_sync test and the lib/sw_sync helper functions for said test.

Gustavo Padovans sw_sync series was just de-staged in gregkh-staging/staging-next [1], and this test is targeted at verifying the functionality implemented in that series.

The sw_sync subtests range from very basic tests of the sw_sync functionality, to stress testing and randomized tests.

[1] http://git.kernel.org/cgit/linux/kernel/git/gregkh/staging.git/

Changes since v1:
Added "Reviewed-by: Eric Engestrom <[email protected]>" tag

lib/sw_sync:
- Fixed fd value checking
- Fixed fd validity check in sw_sync_fd_close()
- Moved sw_sync related paths to define
- Switched from malloc+memset to calloc in sync_file_info()
- Switched sizeof to dereferenced ptr

tests/sw_sync:
- Moved lib/sw_sync related code to lib/sw_sync commit
- Replaced memset with assignment in loop

Robert Foss (13):
lib/sw_sync: Add helper functions for managing synchronization
primitives
tests/sw_sync: Add sw_sync test
tests/sw_sync: Add subtest test_alloc_fence tests/sw_sync: Add subtest test_alloc_fence_invalid_timeline tests/sw_sync: Add subtest test_alloc_merge_fence tests/sw_sync: Add subtest test_sync_wait
tests/sw_sync: Add subtest test_sync_merge tests/sw_sync: Add subtest test_sync_merge_same tests/sw_sync: Add subtest test_sync_multi_consumer tests/sw_sync: Add subtest test_sync_multi_consumer_producer tests/sw_sync: Add subtest test_sync_random_merge tests/sw_sync: Add subtest test_sync_multi_timeline_wait tests/sw_sync: Add subtest test_sync_multi_producer_single_consumer

 lib/Makefile.sources   |   2 +
 lib/sw_sync.c          | 237 +++++++++++++++++
 lib/sw_sync.h          |  49 ++++
 tests/Makefile.sources |   1 +
 tests/sw_sync.c        | 693 +++++++++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 982 insertions(+)
 create mode 100644 lib/sw_sync.c
 create mode 100644 lib/sw_sync.h
 create mode 100644 tests/sw_sync.c
Signature · 6 lines
2.9.3

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

PATCH v2 1/13 lib/sw_sync: Add helper functions for managing synchronization primitives

From: Robert Foss <[email protected]>

Base functions to help testing the Sync File Framework (explicit fencing mechanism ported from Android).
These functions allow you to create, use and destroy timelines and fences.

---
 lib/Makefile.sources |   2 +
 lib/sw_sync.c        | 237 +++++++++++++++++++++++++++++++++++++++++++++++++++
 lib/sw_sync.h        |  49 +++++++++++
 3 files changed, 288 insertions(+)
 create mode 100644 lib/sw_sync.c
 create mode 100644 lib/sw_sync.h
lib/Makefile.sources +2 -0
index bac9a7f..3dc7c3c 100644
@@ -61,6 +61,8 @@ lib_source_list =	 	\
 	rendercopy_gen8.c	\
 	rendercopy_gen9.c	\
 	rendercopy.h		\
+	sw_sync.c		\
+	sw_sync.h		\
 	intel_reg_map.c		\
 	intel_iosf.c		\
 	igt_kms.c		\
lib/sw_sync.c new file +237 -0

237 lines not on this page · raw message

lib/sw_sync.h new file +49 -0

49 lines not on this page · raw message

Signature · 6 lines
2.9.3

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

Re: [PATCH i-g-t v2 01/13] lib/sw_sync: Add helper functions for managing synchronization primitives

Quoted 105 lines · On Mon, Sep 12, 2016 at 06:08:30PM -0400, [email protected] wrote:

From: Robert Foss <[email protected]>

Base functions to help testing the Sync File Framework (explicit fencing mechanism ported from Android).
These functions allow you to create, use and destroy timelines and fences.

---
 lib/Makefile.sources |   2 +
 lib/sw_sync.c        | 237 +++++++++++++++++++++++++++++++++++++++++++++++++++
 lib/sw_sync.h        |  49 +++++++++++
 3 files changed, 288 insertions(+)
 create mode 100644 lib/sw_sync.c
 create mode 100644 lib/sw_sync.h
lib/Makefile.sources +2 -0
index bac9a7f..3dc7c3c 100644
@@ -61,6 +61,8 @@ lib_source_list =	 	\
 	rendercopy_gen8.c	\
 	rendercopy_gen9.c	\
 	rendercopy.h		\
+	sw_sync.c		\
+	sw_sync.h		\
 	intel_reg_map.c		\
 	intel_iosf.c		\
 	igt_kms.c		\
lib/sw_sync.c new file +69 -0
new file mode 100644
index 0000000..179a473
@@ -0,0 +1,237 @@
+/*
+ * Copyright 2012 Google, Inc
+ * Copyright © 2016 Collabora, Ltd.
+ *
+ * Based on the implementation from the Android Open Source Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Robert Foss <[email protected]>
+ */
+
+#ifndef ANDROID
+#define _GNU_SOURCE
+#else
+#include <libgen.h>
+#endif
+#include <fcntl.h>
+#include <poll.h>
+#include <stdint.h>
+#include <linux/sync_file.h>
+#include <sys/ioctl.h>
+
+#include "sw_sync.h"
+#include "drmtest.h"
+#include "ioctl_wrappers.h"
+
+#ifndef SW_SYNC_IOC_INC
+struct sw_sync_create_fence_data {
+	__u32	value;
+	char	name[32];
+	__s32	fence;
+};
+
+#define SW_SYNC_IOC_MAGIC		'W'
+#define SW_SYNC_IOC_CREATE_FENCE	_IOWR(SW_SYNC_IOC_MAGIC, 0,\
+						struct sw_sync_create_fence_data)
+#define SW_SYNC_IOC_INC			_IOW(SW_SYNC_IOC_MAGIC, 1, __u32)
+#endif
+
+#define DEVFS_SW_SYNC   "/dev/sw_sync"
+#define DEBUGFS_SW_SYNC "/sys/kernel/debug/sync/sw_sync"
+
+
+int sw_sync_fd_is_valid(int fd)
+{
+	int status;
+
+	if (fd < 0)
+		return 0;
+
+	status = fcntl(fd, F_GETFD, 0);

lib/sw_sync.c line 79

Quoted 10 lines
lib/sw_sync.c new file +10 -0
+	return status >= 0;
+}
+
+static
+void sw_sync_fd_close(int fd)
+{
+	if (!sw_sync_fd_is_valid(fd))
+		return;
+
+	close(fd);

What bug are you trying to prevent? This is the same as just calling close(fd) and seeing EINVAL.

lib/sw_sync.c line 84

Quoted 5 lines
lib/sw_sync.c new file +5 -0
+}
+
+int sw_sync_timeline_create(void)
+{
+	int fd = open(DEVFS_SW_SYNC, O_RDWR);

if (fd < 0)

	fd = open(DEBUGFS_SW_SYNC, O_RDWR);

igt_require(fd >= 0);

lib/sw_sync.c line 96

Quoted 11 lines
lib/sw_sync.c new file +11 -0
+
+		fd = open(DEBUGFS_SW_SYNC, O_RDWR);
+
+	igt_assert(sw_sync_fd_is_valid(fd));
+
+	return fd;
+}
+
+void sw_sync_timeline_destroy(int fd)
+{
+	return sw_sync_fd_close(fd);

Now just close(fd); and might as well not obfuscate that these are plain fd.

lib/sw_sync.c line 114

Quoted 11 lines
lib/sw_sync.c new file +11 -0
+int sw_sync_fence_create(int fd, int32_t seqno)
+{
+	struct sw_sync_create_fence_data data = {};
+	data.value = seqno;
+
+	if (fd >= 0) {
+		do_ioctl(fd, SW_SYNC_IOC_CREATE_FENCE, &data);
+		return data.fence;
+	} else {
+		do_ioctl_err(fd, SW_SYNC_IOC_CREATE_FENCE, &data, EBADF);
+		return -1;

Try:

int __sw_sync_fence_create(int fd, int32_t seqno) /* int32_t not unsigned ? */ {

	struct sw_sync_create_fence_data data;

	memset(&data, 0, sizeof(data));
	data.value = seqno;

	if (igt_ioctl(fd, SW_SYNC_IOCT_CREATE_FENCE, &data))
		return -errno;
	
	return data.fence;
}

int sw_sync_fence_create(int fd, int32_t seqno)

{
	int fence = __sw_sync_fence_create(fd, seqno);
	igt_assert(fence >= 0);
	return fence;
}

Then only in the test code do you send garbage and check for the expected errno.

lib/sw_sync.c line 123

Quoted 9 lines
lib/sw_sync.c new file +9 -0
+	}
+}
+
+void sw_sync_timeline_inc(int fd, uint32_t count)
+{
+	uint32_t arg = count;
+
+	if (fd == 0 || fd == -1)
+		return;

What's wrong with fd == 0?

See above. By default assert that we don't fail, then have a function for negative tests.

lib/sw_sync.c line 117

Quoted 2 lines
lib/sw_sync.c new file +2 -0
+
+	do_ioctl(fd, SW_SYNC_IOC_INC, &arg);

Try not to use do_ioctl(), it generates horrible messages from the macro expansions of the ioctl numbers.

lib/sw_sync.c line 137

Quoted 12 lines
lib/sw_sync.c new file +12 -0
+}
+
+int sw_sync_merge(int fd1, int fd2)
+{
+	struct sync_merge_data data = {};
+	int err;
+
+	data.fd2 = fd2;
+
+	err = ioctl(fd1, SYNC_IOC_MERGE, &data);
+	if (err < 0)
+		return err;

Our convention is to return -errno.

Quoted 20 lines
lib/sw_sync.c new file +20 -0
+
+	sw_sync_fd_is_valid(data.fence);
+
+	return data.fence;
+}
+
+int sw_sync_wait(int fence, int timeout)
+{
+	struct pollfd fds;
+	int ret;
+
+	fds.fd = fence;
+	fds.events = POLLIN | POLLERR;
+
+	ret = poll(&fds, 1, timeout);
+
+	return ret;
+}
+
+int sw_sync_fence_size(int fd)

lib/sw_sync.c line 214

Quoted 10 lines
lib/sw_sync.c new file +10 -0
+{
+	int count;
+	struct sync_file_info *info = sync_file_info(fd);
+
+	if (!info)
+		return 0;
+
+	count = info->num_fences;
+
+	sync_file_info_free(info);

Querying all those fences seems like a waste of time now.

sw_sync_fence_count not size.

	struct sync_file_info info;

	memset(&info, 0, sizeof(info));
	if (ioctl(fd, SYNC_IOC_FILE_INFO, &info))
		return -errno;
	
	return info.num_fences;
Quoted 9 lines
lib/sw_sync.c new file +9 -0
+int sw_sync_fence_count_status(int fd, int status)
+{
+	int i, count = 0;
+	struct sync_fence_info *fence_info = NULL;
+	struct sync_file_info *info = sync_file_info(fd);
+
+	if (!info)
+		return -1;
+

lib/sw_sync.c line 237

Quoted 10 lines
lib/sw_sync.c new file +10 -0
+	fence_info = (struct sync_fence_info *)(uintptr_t)info->sync_fence_info;
+	for (i = 0 ; i < info->num_fences ; i++) {
+		if (fence_info[i].status == status)
+			count++;
+	}
+
+	sync_file_info_free(info);
+
+	return count;
+}

int __sw_sync_fence_count_status(int fd, int status)

{
	struct sync_file_info info;
	struct sync_fence_info *fence_info;
	int count;

	memset(&info, 0, sizeof(info));
	if (ioctl(fd, SYNC_IOC_FILE_INFO, &info))
		return -errno;

	fence_info = calloc(info.num_fences, sizeof(*fence_info));
	if (!fence_info)
		return -ENOMEM;
	
	info.fence_info = (uintptr_t)fence_info;
	if (ioctl(fd, SYNC_IOC_FILE_INFO, &info)) {
		count = -errno;
	} else {
		count = 0;
		for (i = 0 ; i < info.num_fences ; i++)
			count += fence_info[i].status == status;
	}

	free(fence_info);

	return count;
}

int sw_sync_fence_count_status(int fd, int status)

{
	int count = __sw_sync_fence_count_status(fd, status);
	igt_assert(count >= 0);
	return count;
}

So that regular callers don't have to worry about errors. -Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw

Re: [PATCH i-g-t v2 01/13] lib/sw_sync: Add helper functions for managing synchronization primitives

Quoted 111 lines · Chris Wilson
Quoted 110 lines · On Mon, Sep 12, 2016 at 06:08:30PM -0400, [email protected] wrote:

From: Robert Foss <[email protected]>

Base functions to help testing the Sync File Framework (explicit fencing mechanism ported from Android).
These functions allow you to create, use and destroy timelines and fences.

---
 lib/Makefile.sources |   2 +
 lib/sw_sync.c        | 237 +++++++++++++++++++++++++++++++++++++++++++++++++++
 lib/sw_sync.h        |  49 +++++++++++
 3 files changed, 288 insertions(+)
 create mode 100644 lib/sw_sync.c
 create mode 100644 lib/sw_sync.h
lib/Makefile.sources +2 -0
index bac9a7f..3dc7c3c 100644
@@ -61,6 +61,8 @@ lib_source_list =	 	\
 	rendercopy_gen8.c	\
 	rendercopy_gen9.c	\
 	rendercopy.h		\
+	sw_sync.c		\
+	sw_sync.h		\
 	intel_reg_map.c		\
 	intel_iosf.c		\
 	igt_kms.c		\
lib/sw_sync.c new file +74 -0
new file mode 100644
index 0000000..179a473
@@ -0,0 +1,237 @@
+/*
+ * Copyright 2012 Google, Inc
+ * Copyright © 2016 Collabora, Ltd.
+ *
+ * Based on the implementation from the Android Open Source Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Robert Foss <[email protected]>
+ */
+
+#ifndef ANDROID
+#define _GNU_SOURCE
+#else
+#include <libgen.h>
+#endif
+#include <fcntl.h>
+#include <poll.h>
+#include <stdint.h>
+#include <linux/sync_file.h>
+#include <sys/ioctl.h>
+
+#include "sw_sync.h"
+#include "drmtest.h"
+#include "ioctl_wrappers.h"
+
+#ifndef SW_SYNC_IOC_INC
+struct sw_sync_create_fence_data {
+	__u32	value;
+	char	name[32];
+	__s32	fence;
+};
+
+#define SW_SYNC_IOC_MAGIC		'W'
+#define SW_SYNC_IOC_CREATE_FENCE	_IOWR(SW_SYNC_IOC_MAGIC, 0,\
+						struct sw_sync_create_fence_data)
+#define SW_SYNC_IOC_INC			_IOW(SW_SYNC_IOC_MAGIC, 1, __u32)
+#endif
+
+#define DEVFS_SW_SYNC   "/dev/sw_sync"
+#define DEBUGFS_SW_SYNC "/sys/kernel/debug/sync/sw_sync"
+
+
+int sw_sync_fd_is_valid(int fd)
+{
+	int status;
+
+	if (fd < 0)
+		return 0;
+
+	status = fcntl(fd, F_GETFD, 0);
+	return status >= 0;
+}
+
+static
+void sw_sync_fd_close(int fd)
Quoted 8 lines

lib/sw_sync.c line 79

Quoted 5 lines
lib/sw_sync.c new file +5 -0
+{
+	if (!sw_sync_fd_is_valid(fd))
+		return;
+
+	close(fd);

What bug are you trying to prevent? This is the same as just calling close(fd) and seeing EINVAL.

It's not really trying to prevent any bugs, but rather the idea is have low cognitive load on these functions. If an fd is valid it can always be closed.

Or are you suggesting that the above code is equivalent to close()? If so it should be replaced by close().

Quoted 11 lines

lib/sw_sync.c line 84

Quoted 6 lines
lib/sw_sync.c new file +5 -0
+}
+
+int sw_sync_timeline_create(void)
+{
+	int fd = open(DEVFS_SW_SYNC, O_RDWR);

if (fd < 0)

	fd = open(DEBUGFS_SW_SYNC, O_RDWR);

igt_require(fd >= 0);

Replacing if (!sw_sync_fd_is_valid(fd)) with fd<0 is seems to me like a good change, in this scenario they should be equivalent.

Adding igt_require() is a bit inflexbile, especially when it comes to actually testing timeline creation.

Quoted 15 lines

lib/sw_sync.c line 96

Quoted 12 lines
lib/sw_sync.c new file +11 -0
+
+		fd = open(DEBUGFS_SW_SYNC, O_RDWR);
+
+	igt_assert(sw_sync_fd_is_valid(fd));
+
+	return fd;
+}
+
+void sw_sync_timeline_destroy(int fd)
+{
+	return sw_sync_fd_close(fd);

Now just close(fd); and might as well not obfuscate that these are plain fd.

You're definitely, right. I'll remove this function from lib/sw_sync and tests/sw_sync in v3.

Quoted 30 lines

lib/sw_sync.c line 114

Quoted 12 lines
lib/sw_sync.c new file +11 -0
+int sw_sync_fence_create(int fd, int32_t seqno)
+{
+	struct sw_sync_create_fence_data data = {};
+	data.value = seqno;
+
+	if (fd >= 0) {
+		do_ioctl(fd, SW_SYNC_IOC_CREATE_FENCE, &data);
+		return data.fence;
+	} else {
+		do_ioctl_err(fd, SW_SYNC_IOC_CREATE_FENCE, &data, EBADF);
+		return -1;

Try:

int __sw_sync_fence_create(int fd, int32_t seqno) /* int32_t not unsigned ? */ {

	struct sw_sync_create_fence_data data;

	memset(&data, 0, sizeof(data));
	data.value = seqno;

	if (igt_ioctl(fd, SW_SYNC_IOCT_CREATE_FENCE, &data))
		return -errno;
	
	return data.fence;
}

int sw_sync_fence_create(int fd, int32_t seqno)

Quoted 9 lines
{
	int fence = __sw_sync_fence_create(fd, seqno);
	igt_assert(fence >= 0);
	return fence;
}

Then only in the test code do you send garbage and check for the expected errno.

What would the corresponding negative test code look like?
A call to __sw_sync_fence_create? Then __sw_sync_fence_create would have to be made accessible outside of lib/sw_sync.

Or maybe creating a second user of __sw_sync_fence_create along the lines of sw_sync_fence_create_fail with an inverted igt_assert check is what you're suggesting.

Quoted 14 lines

lib/sw_sync.c line 123

Quoted 11 lines
lib/sw_sync.c new file +9 -0
+	}
+}
+
+void sw_sync_timeline_inc(int fd, uint32_t count)
+{
+	uint32_t arg = count;
+
+	if (fd == 0 || fd == -1)
+		return;

What's wrong with fd == 0?

Quoted 2 lines

See above. By default assert that we don't fail, then have a function for negative tests.

Nothing. I'll fix this in v3.

Quoted 7 lines

lib/sw_sync.c line 117

Quoted 3 lines
lib/sw_sync.c new file +2 -0
+
+	do_ioctl(fd, SW_SYNC_IOC_INC, &arg);

Try not to use do_ioctl(), it generates horrible messages from the macro expansions of the ioctl numbers.

Quoted 14 lines

lib/sw_sync.c line 137

Quoted 12 lines
lib/sw_sync.c new file +12 -0
+}
+
+int sw_sync_merge(int fd1, int fd2)
+{
+	struct sync_merge_data data = {};
+	int err;
+
+	data.fd2 = fd2;
+
+	err = ioctl(fd1, SYNC_IOC_MERGE, &data);
+	if (err < 0)
+		return err;

Our convention is to return -errno.

Will be fixed in v3.

Quoted 25 lines
Quoted 25 lines
lib/sw_sync.c new file +24 -0
+
+	sw_sync_fd_is_valid(data.fence);
+
+	return data.fence;
+}
+
+int sw_sync_wait(int fence, int timeout)
+{
+	struct pollfd fds;
+	int ret;
+
+	fds.fd = fence;
+	fds.events = POLLIN | POLLERR;
+
+	ret = poll(&fds, 1, timeout);
+
+	return ret;
+}
+
+int sw_sync_fence_size(int fd)
+{
+	int count;
+	struct sync_file_info *info = sync_file_info(fd);
+
Quoted 8 lines

lib/sw_sync.c line 214

Quoted 6 lines
lib/sw_sync.c new file +6 -0
+	if (!info)
+		return 0;
+
+	count = info->num_fences;
+
+	sync_file_info_free(info);

Querying all those fences seems like a waste of time now.

I'm slightly confused by this comment.
sw_sync_fence_size seems to be making just one query, just the same number as the revised code attached below.

Quoted 2 lines

sw_sync_fence_count not size.

Agreed, count is the right name, will fix in v3.

Quoted 57 lines
	struct sync_file_info info;

	memset(&info, 0, sizeof(info));
	if (ioctl(fd, SYNC_IOC_FILE_INFO, &info))
		return -errno;
	
	return info.num_fences;
Quoted 9 lines
lib/sw_sync.c new file +9 -0
+int sw_sync_fence_count_status(int fd, int status)
+{
+	int i, count = 0;
+	struct sync_fence_info *fence_info = NULL;
+	struct sync_file_info *info = sync_file_info(fd);
+
+	if (!info)
+		return -1;
+

lib/sw_sync.c line 237

Quoted 10 lines
lib/sw_sync.c new file +10 -0
+	fence_info = (struct sync_fence_info *)(uintptr_t)info->sync_fence_info;
+	for (i = 0 ; i < info->num_fences ; i++) {
+		if (fence_info[i].status == status)
+			count++;
+	}
+
+	sync_file_info_free(info);
+
+	return count;
+}

int __sw_sync_fence_count_status(int fd, int status)

{
	struct sync_file_info info;
	struct sync_fence_info *fence_info;
	int count;

	memset(&info, 0, sizeof(info));
	if (ioctl(fd, SYNC_IOC_FILE_INFO, &info))
		return -errno;

	fence_info = calloc(info.num_fences, sizeof(*fence_info));
	if (!fence_info)
		return -ENOMEM;
	
	info.fence_info = (uintptr_t)fence_info;
	if (ioctl(fd, SYNC_IOC_FILE_INFO, &info)) {
		count = -errno;
	} else {
		count = 0;
		for (i = 0 ; i < info.num_fences ; i++)
			count += fence_info[i].status == status;
	}

	free(fence_info);

	return count;
}
Quoted 8 lines

int sw_sync_fence_count_status(int fd, int status)

{
	int count = __sw_sync_fence_count_status(fd, status);
	igt_assert(count >= 0);
	return count;
}

So that regular callers don't have to worry about errors.

That's a good improvement, I'll include it in v3.

Thanks for all of the feedback Chris. I'll spin and test v3 now.

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

#4 lore raw

Re: [PATCH i-g-t v2 01/13] lib/sw_sync: Add helper functions for managing synchronization primitives

Quoted 30 lines · Robert Foss
Quoted 17 lines · Chris Wilson

Try:

int __sw_sync_fence_create(int fd, int32_t seqno) /* int32_t not unsigned ? */ {

	struct sw_sync_create_fence_data data;

	memset(&data, 0, sizeof(data));
	data.value = seqno;

	if (igt_ioctl(fd, SW_SYNC_IOCT_CREATE_FENCE, &data))
		return -errno;
	
	return data.fence;
}

int sw_sync_fence_create(int fd, int32_t seqno)

Quoted 9 lines
{
	int fence = __sw_sync_fence_create(fd, seqno);
	igt_assert(fence >= 0);
	return fence;
}

Then only in the test code do you send garbage and check for the expected errno.

Quoted 7 lines

What would the corresponding negative test code look like? A call to __sw_sync_fence_create? Then __sw_sync_fence_create would have to be made accessible outside of lib/sw_sync.

Or maybe creating a second user of __sw_sync_fence_create along the lines of sw_sync_fence_create_fail with an inverted igt_assert check is what you're suggesting.

Exactly. Make the raw unchecked version available to tests. We have been using __gem_foo() and gem_foo() to identify the difference. __gem_foo() reports the error to the caller (so that they can feed in different values of garbage and check for different errno, or maybe used as a probe to see if the kernel supports such a function) and gem_foo() for everyone else where we want to just focus on writing a test and hide the error handling clutter.

Try not to put the error handling tests in the library itself. -Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#5 lore raw

Re: [PATCH i-g-t v2 01/13] lib/sw_sync: Add helper functions for managing synchronization primitives

Quoted 47 lines · Chris Wilson
Quoted 30 lines · Robert Foss
Quoted 17 lines · Chris Wilson

Try:

int __sw_sync_fence_create(int fd, int32_t seqno) /* int32_t not unsigned ? */ {

	struct sw_sync_create_fence_data data;

	memset(&data, 0, sizeof(data));
	data.value = seqno;

	if (igt_ioctl(fd, SW_SYNC_IOCT_CREATE_FENCE, &data))
		return -errno;
	
	return data.fence;
}

int sw_sync_fence_create(int fd, int32_t seqno)

Quoted 9 lines
{
	int fence = __sw_sync_fence_create(fd, seqno);
	igt_assert(fence >= 0);
	return fence;
}

Then only in the test code do you send garbage and check for the expected errno.

Quoted 7 lines

What would the corresponding negative test code look like? A call to __sw_sync_fence_create? Then __sw_sync_fence_create would have to be made accessible outside of lib/sw_sync.

Or maybe creating a second user of __sw_sync_fence_create along the lines of sw_sync_fence_create_fail with an inverted igt_assert check is what you're suggesting.

Exactly. Make the raw unchecked version available to tests. We have been using __gem_foo() and gem_foo() to identify the difference. __gem_foo() reports the error to the caller (so that they can feed in different values of garbage and check for different errno, or maybe used as a probe to see if the kernel supports such a function) and gem_foo() for everyone else where we want to just focus on writing a test and hide the error handling clutter.

Quoted 1 line

Try not to put the error handling tests in the library itself.

Thanks!
Coming up in v4.

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

#6 lore raw

PATCH v2 2/13 tests/sw_sync: Add sw_sync test

In reply to [email protected]

From: Robert Foss <[email protected]>

Add initial tests for sw_sync.

---
 tests/Makefile.sources |  1 +
 tests/sw_sync.c        | 51 ++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 52 insertions(+)
 create mode 100644 tests/sw_sync.c
tests/Makefile.sources +1 -0
index 72a58ad..0ba769f 100644
@@ -125,6 +125,7 @@ TESTS_progs_M = \
 	prime_mmap_kms \
 	prime_self_import \
 	prime_vgem \
+	sw_sync \
 	template \
 	vgem_basic \
 	vgem_slow \
tests/sw_sync.c new file +51 -0

51 lines not on this page · raw message

Signature · 6 lines
2.9.3

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

PATCH v2 3/13 tests/sw_sync: Add subtest test_alloc_fence

In reply to [email protected]

From: Robert Foss <[email protected]>

Add subtest alloc_fence that verifies that it's possible to allocate a fence on a timeline.

---
 tests/sw_sync.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
tests/sw_sync.c +16 -0
index d2d4c42..a8c8ca4 100644
@@ -43,9 +43,25 @@ static void test_alloc_timeline(void)
 	sw_sync_timeline_destroy(timeline);
 }
 
+static void test_alloc_fence(void)
+{
+	int in_fence;
+	int timeline;
+
+	timeline = sw_sync_timeline_create();
+	in_fence = sw_sync_fence_create(timeline, 0);
+
+	sw_sync_fence_destroy(in_fence);
+	sw_sync_timeline_destroy(timeline);
+}
+
+
 igt_main
 {
 	igt_subtest("alloc_timeline")
 		test_alloc_timeline();
+
+	igt_subtest("alloc_fence")
+		test_alloc_fence();
 }
 
Signature · 6 lines
2.9.3

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

PATCH v2 4/13 tests/sw_sync: Add subtest test_alloc_fence_invalid_timeline

In reply to [email protected]

From: Robert Foss <[email protected]>

This subtests tests that creating fences on negative timelines fail.

---
 tests/sw_sync.c | 7 +++++++
 1 file changed, 7 insertions(+)
tests/sw_sync.c +7 -0
index a8c8ca4..102647d 100644
@@ -55,6 +55,10 @@ static void test_alloc_fence(void)
 	sw_sync_timeline_destroy(timeline);
 }
 
+static void test_alloc_fence_invalid_timeline(void)
+{
+	sw_sync_fence_create(-1, 0);
+}
 
 igt_main
 {
@@ -63,5 +67,8 @@ igt_main
 
 	igt_subtest("alloc_fence")
 		test_alloc_fence();
+
+	igt_subtest("alloc_fence_invalid_timeline")
+		test_alloc_fence_invalid_timeline();
 }
 
Signature · 6 lines
2.9.3

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

PATCH v2 5/13 tests/sw_sync: Add subtest test_alloc_merge_fence

In reply to [email protected]

From: Robert Foss <[email protected]>

This subtest verifies that merging two fences works in the simples possible case.

---
 tests/sw_sync.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)
tests/sw_sync.c +23 -0
index 102647d..851430e 100644
@@ -60,6 +60,26 @@ static void test_alloc_fence_invalid_timeline(void)
 	sw_sync_fence_create(-1, 0);
 }
 
+static void test_alloc_merge_fence(void)
+{
+	int in_fence[2];
+	int fence_merge;
+	int timeline[2];
+
+	timeline[0] = sw_sync_timeline_create();
+	timeline[1] = sw_sync_timeline_create();
+
+	in_fence[0] = sw_sync_fence_create(timeline[0], 1);
+	in_fence[1] = sw_sync_fence_create(timeline[1], 1);
+	fence_merge = sw_sync_merge(in_fence[1], in_fence[0]);
+
+	sw_sync_fence_destroy(in_fence[0]);
+	sw_sync_fence_destroy(in_fence[1]);
+	sw_sync_fence_destroy(fence_merge);
+	sw_sync_timeline_destroy(timeline[0]);
+	sw_sync_timeline_destroy(timeline[1]);
+}
+
 igt_main
 {
 	igt_subtest("alloc_timeline")
@@ -70,5 +90,8 @@ igt_main
 
 	igt_subtest("alloc_fence_invalid_timeline")
 		test_alloc_fence_invalid_timeline();
+
+	igt_subtest("alloc_merge_fence")
+		test_alloc_merge_fence();
 }
 
Signature · 6 lines
2.9.3

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

PATCH v2 6/13 tests/sw_sync: Add subtest test_sync_wait

In reply to [email protected]

From: Robert Foss <[email protected]>

This subtest verifies that waiting on fences works properly.

---
 tests/sw_sync.c | 38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
tests/sw_sync.c +38 -0
index 851430e..07e9638 100644
@@ -80,6 +80,41 @@ static void test_alloc_merge_fence(void)
 	sw_sync_timeline_destroy(timeline[1]);
 }
 
+static void test_sync_wait(void)
+{
+	int fence, ret;
+	int timeline;
+
+	timeline = sw_sync_timeline_create();
+	fence = sw_sync_fence_create(timeline, 5);
+
+	/* Wait on fence until timeout */
+	ret = sw_sync_wait(fence, 0);
+	igt_assert_f(ret == 0, "Failure waiting on fence until timeout\n");
+
+	/* Advance timeline from 0 -> 1 */
+	sw_sync_timeline_inc(timeline, 1);
+
+	/* Wait on fence until timeout */
+	ret = sw_sync_wait(fence, 0);
+	igt_assert_f(ret == 0, "Failure waiting on fence until timeout\n");
+
+	/* Signal the fence */
+	sw_sync_timeline_inc(timeline, 4);
+
+	/* Wait successfully */
+	ret = sw_sync_wait(fence, 0);
+	igt_assert_f(ret > 0, "Failure waiting on fence\n");
+
+	/* Go even further, and confirm wait still succeeds */
+	sw_sync_timeline_inc(timeline, 10);
+	ret = sw_sync_wait(fence, 0);
+	igt_assert_f(ret > 0, "Failure waiting ahead\n");
+
+	sw_sync_fence_destroy(fence);
+	sw_sync_timeline_destroy(timeline);
+}
+
 igt_main
 {
 	igt_subtest("alloc_timeline")
@@ -93,5 +128,8 @@ igt_main
 
 	igt_subtest("alloc_merge_fence")
 		test_alloc_merge_fence();
+
+	igt_subtest("sync_wait")
+		test_sync_wait();
 }
 
Signature · 6 lines
2.9.3

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

PATCH v2 7/13 tests/sw_sync: Add subtest test_sync_merge

In reply to [email protected]

From: Robert Foss <[email protected]>

Add subtest test_sync_merge that tests merging fences and the validity of the resulting merged fence.

---
 tests/sw_sync.c | 67 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 67 insertions(+)
tests/sw_sync.c +67 -0
index 07e9638..00ac44b 100644
@@ -115,6 +115,70 @@ static void test_sync_wait(void)
 	sw_sync_timeline_destroy(timeline);
 }
 
+static void test_sync_merge(void)
+{
+	int in_fence[3];
+	int fence_merge;
+	int timeline;
+	int active, signaled;
+
+	timeline = sw_sync_timeline_create();
+	in_fence[0] = sw_sync_fence_create(timeline, 1);
+	in_fence[1] = sw_sync_fence_create(timeline, 2);
+	in_fence[2] = sw_sync_fence_create(timeline, 3);
+
+	fence_merge = sw_sync_merge(in_fence[0], in_fence[1]);
+	fence_merge = sw_sync_merge(in_fence[2], fence_merge);
+
+	/* confirm all fences have one active point (even d) */
+	active = sw_sync_fence_count_status(in_fence[0],
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	igt_assert_f(active == 1, "in_fence[0] has too many active fences\n");
+	active = sw_sync_fence_count_status(in_fence[1],
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	igt_assert_f(active == 1, "in_fence[1] has too many active fences\n");
+	active = sw_sync_fence_count_status(in_fence[2],
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	igt_assert_f(active == 1, "in_fence[2] has too many active fences\n");
+	active = sw_sync_fence_count_status(fence_merge,
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	igt_assert_f(active == 1, "fence_merge has too many active fences\n");
+
+	/* confirm that fence_merge is not signaled until the max of fence 0,1,2 */
+	sw_sync_timeline_inc(timeline, 1);
+	signaled = sw_sync_fence_count_status(in_fence[0],
+					      SW_SYNC_FENCE_STATUS_SIGNALED);
+	active = sw_sync_fence_count_status(fence_merge,
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	igt_assert_f(signaled == 1, "in_fence[0] did not signal\n");
+	igt_assert_f(active == 1, "fence_merge signaled too early\n");
+
+	sw_sync_timeline_inc(timeline, 1);
+	signaled = sw_sync_fence_count_status(in_fence[1],
+					      SW_SYNC_FENCE_STATUS_SIGNALED);
+	active = sw_sync_fence_count_status(fence_merge,
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	igt_assert_f(signaled == 1, "in_fence[1] did not signal\n");
+	igt_assert_f(active == 1, "fence_merge signaled too early\n");
+
+	sw_sync_timeline_inc(timeline, 1);
+	signaled = sw_sync_fence_count_status(in_fence[2],
+					      SW_SYNC_FENCE_STATUS_SIGNALED);
+	igt_assert_f(signaled == 1, "in_fence[2] did not signal\n");
+	signaled = sw_sync_fence_count_status(fence_merge,
+					       SW_SYNC_FENCE_STATUS_SIGNALED);
+	active = sw_sync_fence_count_status(fence_merge,
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	igt_assert_f(active == 0 && signaled == 1,
+		     "fence_merge did not signal\n");
+
+	sw_sync_fence_destroy(in_fence[0]);
+	sw_sync_fence_destroy(in_fence[1]);
+	sw_sync_fence_destroy(in_fence[2]);
+	sw_sync_fence_destroy(fence_merge);
+	sw_sync_timeline_destroy(timeline);
+}
+
 igt_main
 {
 	igt_subtest("alloc_timeline")
@@ -131,5 +195,8 @@ igt_main
 
 	igt_subtest("sync_wait")
 		test_sync_wait();
+
+	igt_subtest("sync_merge")
+		test_sync_merge();
 }
 
Signature · 6 lines
2.9.3

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

PATCH v2 8/13 tests/sw_sync: Add subtest test_sync_merge_same

In reply to [email protected]

From: Robert Foss <[email protected]>

This subtest verifies merging a fence with itself does not fail.

---
 tests/sw_sync.c | 27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)
tests/sw_sync.c +27 -0
index 00ac44b..db03f48 100644
@@ -179,6 +179,30 @@ static void test_sync_merge(void)
 	sw_sync_timeline_destroy(timeline);
 }
 
+static void test_sync_merge_same(void)
+{
+	int in_fence[2];
+	int timeline;
+	int signaled;
+
+	timeline = sw_sync_timeline_create();
+	in_fence[0] = sw_sync_fence_create(timeline, 1);
+	in_fence[1] = sw_sync_merge(in_fence[0], in_fence[0]);
+
+	signaled = sw_sync_fence_count_status(in_fence[0],
+					      SW_SYNC_FENCE_STATUS_SIGNALED);
+	igt_assert_f(signaled == 0, "fence signaled too early\n");
+
+	sw_sync_timeline_inc(timeline, 1);
+	signaled = sw_sync_fence_count_status(in_fence[0],
+					      SW_SYNC_FENCE_STATUS_SIGNALED);
+	igt_assert_f(signaled == 1, "fence did not signal\n");
+
+	sw_sync_fence_destroy(in_fence[0]);
+	sw_sync_fence_destroy(in_fence[1]);
+	sw_sync_timeline_destroy(timeline);
+}
+
 igt_main
 {
 	igt_subtest("alloc_timeline")
@@ -198,5 +222,8 @@ igt_main
 
 	igt_subtest("sync_merge")
 		test_sync_merge();
+
+	igt_subtest("sync_merge_same")
+		test_sync_merge_same();
 }
 
Signature · 6 lines
2.9.3

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

PATCH v2 9/13 tests/sw_sync: Add subtest test_sync_multi_consumer

In reply to [email protected]

From: Robert Foss <[email protected]>

This subtest verifies the access ordering of multiple consumer threads.

---
 tests/sw_sync.c | 103 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 103 insertions(+)
tests/sw_sync.c +103 -0
index db03f48..0c9c923 100644
@@ -27,6 +27,8 @@
  *    Robert Foss <[email protected]>
  */
 
+#include <pthread.h>
+#include <semaphore.h>
 #include <stdint.h>
 
 #include "sw_sync.h"
@@ -35,6 +37,15 @@
 
 IGT_TEST_DESCRIPTION("Test SW Sync Framework");
 
+typedef struct {
+	int timeline;
+	uint32_t thread_id;
+	uint32_t nbr_threads;
+	uint32_t nbr_iterations;
+	volatile uint32_t * volatile counter;
+	sem_t *sem;
+} data_t;
+
 static void test_alloc_timeline(void)
 {
 	int timeline;
@@ -203,6 +214,95 @@ static void test_sync_merge_same(void)
 	sw_sync_timeline_destroy(timeline);
 }
 
+static void * test_sync_multi_consumer_thread(void *arg)
+{
+	data_t *data = arg;
+	int thread_id = data->thread_id;
+	int nbr_threads = data->nbr_threads;
+	int timeline = data->timeline;
+	int iterations = data->nbr_iterations;
+	int ret, i;
+
+	for (i = 0; i < iterations; i++) {
+		int next_point = i * nbr_threads + thread_id;
+		int fence = sw_sync_fence_create(timeline, next_point);
+
+		ret = sw_sync_wait(fence, 1000);
+		if (ret <= 0)
+		{
+			return (void *) 1;
+		}
+
+		if (*(data->counter) != next_point)
+		{
+			return (void *) 1;
+		}
+
+		sem_post(data->sem);
+		sw_sync_fence_destroy(fence);
+	}
+	return NULL;
+}
+
+static void test_sync_multi_consumer(void)
+{
+	const uint32_t nbr_threads = 8;
+	const uint32_t nbr_iterations = 1 << 14;
+	data_t data_arr[nbr_threads];
+	pthread_t thread_arr[nbr_threads];
+	sem_t sem;
+	int timeline;
+	volatile uint32_t counter = 0;
+	uintptr_t thread_ret = 0;
+	data_t data;
+	int i, ret;
+
+	sem_init(&sem, 0, 0);
+	timeline = sw_sync_timeline_create();
+
+	data.nbr_iterations = nbr_iterations;
+	data.nbr_threads = nbr_threads;
+	data.counter = &counter;
+	data.timeline = timeline;
+	data.sem = &sem;
+
+	/* Start sync threads. */
+	for (i = 0; i < nbr_threads; i++)
+	{
+		data_arr[i] = data;
+		data_arr[i].thread_id = i;
+		ret = pthread_create(&thread_arr[i], NULL,
+				     test_sync_multi_consumer_thread,
+				     (void *) &(data_arr[i]));
+		igt_assert_eq(ret, 0);
+	}
+
+	/* Produce 'content'. */
+	for (i = 0; i < nbr_threads * nbr_iterations; i++)
+	{
+		sem_wait(&sem);
+
+		counter++;
+		sw_sync_timeline_inc(timeline, 1);
+	}
+
+	/* Wait for threads to complete. */
+	for (i = 0; i < nbr_threads; i++)
+	{
+		uintptr_t local_thread_ret;
+		pthread_join(thread_arr[i], (void **)&local_thread_ret);
+		thread_ret |= local_thread_ret;
+	}
+
+	sw_sync_timeline_destroy(timeline);
+	sem_destroy(&sem);
+
+	igt_assert_f(counter == nbr_threads * nbr_iterations,
+		     "Counter has unexpected value.\n");
+
+	igt_assert_f(thread_ret == 0, "A sync thread reported failure.\n");
+}
+
 igt_main
 {
 	igt_subtest("alloc_timeline")
@@ -225,5 +325,8 @@ igt_main
 
 	igt_subtest("sync_merge_same")
 		test_sync_merge_same();
+
+	igt_subtest("sync_multi_consumer")
+		test_sync_multi_consumer();
 }
 
Signature · 6 lines
2.9.3

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

PATCH v2 10/13 tests/sw_sync: Add subtest test_sync_multi_consumer_producer

In reply to [email protected]

From: Robert Foss <[email protected]>

This test verifies that stressing the kernel by creating multiple consumer/producer threads that wait on a single timeline to be incremented by another conumer/producer thread does not fail.
And that the order amongst the threads is maintained.

---
 tests/sw_sync.c | 83 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 83 insertions(+)
tests/sw_sync.c +83 -0
index 0c9c923..0e67ad5 100644
@@ -303,6 +303,86 @@ static void test_sync_multi_consumer(void)
 	igt_assert_f(thread_ret == 0, "A sync thread reported failure.\n");
 }
 
+static void * test_sync_multi_consumer_producer_thread(void *arg)
+{
+	data_t *data = arg;
+	int thread_id = data->thread_id;
+	int nbr_threads = data->nbr_threads;
+	int timeline = data->timeline;
+	int iterations = data->nbr_iterations;
+	int ret, i;
+
+	for (i = 0; i < iterations; i++) {
+		int next_point = i * nbr_threads + thread_id;
+		int fence = sw_sync_fence_create(timeline, next_point);
+
+		ret = sw_sync_wait(fence, 1000);
+		if (ret <= 0)
+		{
+			return (void *) 1;
+		}
+
+		if (*(data->counter) != next_point)
+		{
+			return (void *) 1;
+		}
+
+		(*data->counter)++;
+
+		/* Kick off the next thread. */
+		sw_sync_timeline_inc(timeline, 1);
+
+		sw_sync_fence_destroy(fence);
+	}
+	return NULL;
+}
+
+static void test_sync_multi_consumer_producer(void)
+{
+	const uint32_t nbr_threads = 8;
+	const uint32_t nbr_iterations = 1 << 14;
+	data_t data_arr[nbr_threads];
+	pthread_t thread_arr[nbr_threads];
+	int timeline;
+	volatile uint32_t counter = 0;
+	uintptr_t thread_ret = 0;
+	data_t data;
+	int i, ret;
+
+	timeline = sw_sync_timeline_create();
+
+	data.nbr_iterations = nbr_iterations;
+	data.nbr_threads = nbr_threads;
+	data.counter = &counter;
+	data.timeline = timeline;
+
+	/* Start consumer threads. */
+	for (i = 0; i < nbr_threads; i++)
+	{
+		data_arr[i] = data;
+		data_arr[i].thread_id = i;
+		ret = pthread_create(&thread_arr[i], NULL,
+				     test_sync_multi_consumer_producer_thread,
+				     (void *) &(data_arr[i]));
+		igt_assert_eq(ret, 0);
+	}
+
+	/* Wait for threads to complete. */
+	for (i = 0; i < nbr_threads; i++)
+	{
+		uintptr_t local_thread_ret;
+		pthread_join(thread_arr[i], (void **)&local_thread_ret);
+		thread_ret |= local_thread_ret;
+	}
+
+	sw_sync_timeline_destroy(timeline);
+
+	igt_assert_f(counter == nbr_threads * nbr_iterations,
+		     "Counter has unexpected value.\n");
+
+	igt_assert_f(thread_ret == 0, "A sync thread reported failure.\n");
+}
+
 igt_main
 {
 	igt_subtest("alloc_timeline")
@@ -328,5 +408,8 @@ igt_main
 
 	igt_subtest("sync_multi_consumer")
 		test_sync_multi_consumer();
+
+	igt_subtest("sync_multi_consumer_producer")
+		test_sync_multi_consumer_producer();
 }
 
Signature · 6 lines
2.9.3

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

PATCH v2 11/13 tests/sw_sync: Add subtest test_sync_random_merge

In reply to [email protected]

From: Robert Foss <[email protected]>

This subtest verifies that creating many timelines and merging random fences from each timeline with eachother results in merged fences that are fully functional.

---
 tests/sw_sync.c | 73 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 73 insertions(+)
tests/sw_sync.c +73 -0
index 0e67ad5..93f4151 100644
@@ -383,6 +383,76 @@ static void test_sync_multi_consumer_producer(void)
 	igt_assert_f(thread_ret == 0, "A sync thread reported failure.\n");
 }
 
+static void test_sync_random_merge(void)
+{
+	int i, size, ret;
+	const int nbr_timeline = 32;
+	const int nbr_merge = 1024;
+	int fence_map[nbr_timeline];
+	int timeline_arr[nbr_timeline];
+	int fence, tmpfence, merged;
+	int timeline, timeline_offset, sync_pt;
+
+	srand(time(NULL));
+
+	for (i = 0; i < nbr_timeline; i++) {
+		timeline_arr[i] = sw_sync_timeline_create();
+		fence_map[i] = -1;
+	}
+
+	sync_pt = rand();
+	fence = sw_sync_fence_create(timeline_arr[0], sync_pt);
+
+	fence_map[0] = sync_pt;
+
+	/* Randomly create syncpoints out of a fixed set of timelines,
+	 * and merge them together.
+	 */
+	for (i = 0; i < nbr_merge; i++) {
+		/* Generate syncpoint. */
+		timeline_offset = rand() % nbr_timeline;
+		timeline = timeline_arr[timeline_offset];
+		sync_pt = rand();
+
+		/* Keep track of the latest sync_pt in each timeline. */
+		if (fence_map[timeline_offset] == -1)
+			fence_map[timeline_offset] = sync_pt;
+		else if (fence_map[timeline_offset] < sync_pt)
+			fence_map[timeline_offset] = sync_pt;
+
+		/* Merge. */
+		tmpfence = sw_sync_fence_create(timeline, sync_pt);
+		merged = sw_sync_merge(tmpfence, fence);
+		sw_sync_fence_destroy(tmpfence);
+		sw_sync_fence_destroy(fence);
+		fence = merged;
+	}
+
+	size = 0;
+	for (i = 0; i < nbr_timeline; i++)
+		if (fence_map[i] != -1)
+			size++;
+
+	/* Trigger the merged fence. */
+	for (i = 0; i < nbr_timeline; i++) {
+		if (fence_map[i] != -1) {
+			ret = sw_sync_wait(fence, 0);
+			igt_assert_f(ret == 0,
+				    "Failure waiting on fence until timeout\n");
+			/* Increment the timeline to the last sync_pt */
+			sw_sync_timeline_inc(timeline_arr[i], fence_map[i]);
+		}
+	}
+
+	/* Check that the fence is triggered. */
+	ret = sw_sync_wait(fence, 0);
+	igt_assert_f(ret > 0, "Failure triggering fence\n");
+
+	sw_sync_fence_destroy(fence);
+	for (i = 0; i < nbr_timeline; i++)
+		sw_sync_timeline_destroy(timeline_arr[i]);
+}
+
 igt_main
 {
 	igt_subtest("alloc_timeline")
@@ -411,5 +481,8 @@ igt_main
 
 	igt_subtest("sync_multi_consumer_producer")
 		test_sync_multi_consumer_producer();
+
+	igt_subtest("sync_random_merge")
+		test_sync_random_merge();
 }
 
Signature · 6 lines
2.9.3

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

PATCH v2 12/13 tests/sw_sync: Add subtest test_sync_multi_timeline_wait

In reply to [email protected]

From: Robert Foss <[email protected]>

This subtest verifies that waiting, timing out on a wait and that counting fences in various states works.

---
 tests/sw_sync.c | 66 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)
tests/sw_sync.c +66 -0
index 93f4151..322c7b1 100644
@@ -214,6 +214,69 @@ static void test_sync_merge_same(void)
 	sw_sync_timeline_destroy(timeline);
 }
 
+static void test_sync_multi_timeline_wait(void)
+{
+	int timeline[3];
+	int in_fence[3];
+	int fence_merge;
+	int active, signaled, ret;
+
+	timeline[0] = sw_sync_timeline_create();
+	timeline[1] = sw_sync_timeline_create();
+	timeline[2] = sw_sync_timeline_create();
+
+	in_fence[0] = sw_sync_fence_create(timeline[0], 5);
+	in_fence[1] = sw_sync_fence_create(timeline[1], 5);
+	in_fence[2] = sw_sync_fence_create(timeline[2], 5);
+
+	fence_merge = sw_sync_merge(in_fence[0], in_fence[1]);
+	fence_merge = sw_sync_merge(in_fence[2], fence_merge);
+
+	/* Confirm fence isn't signaled */
+	active = sw_sync_fence_count_status(fence_merge,
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	igt_assert_f(active == 3, "Fence signaled too early\n");
+
+	ret = sw_sync_wait(fence_merge, 0);
+	igt_assert_f(ret == 0, "Failure waiting on fence until timeout\n");
+
+	sw_sync_timeline_inc(timeline[0], 5);
+	active = sw_sync_fence_count_status(fence_merge,
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	signaled = sw_sync_fence_count_status(fence_merge,
+					      SW_SYNC_FENCE_STATUS_SIGNALED);
+	igt_assert_f(active == 2 && signaled == 1,
+		    "Fence did not signal properly\n");
+
+	sw_sync_timeline_inc(timeline[1], 5);
+	active = sw_sync_fence_count_status(fence_merge,
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	signaled = sw_sync_fence_count_status(fence_merge,
+					      SW_SYNC_FENCE_STATUS_SIGNALED);
+	igt_assert_f(active == 1 && signaled == 2,
+		    "Fence did not signal properly\n");
+
+	sw_sync_timeline_inc(timeline[2], 5);
+	active = sw_sync_fence_count_status(fence_merge,
+					    SW_SYNC_FENCE_STATUS_ACTIVE);
+	signaled = sw_sync_fence_count_status(fence_merge,
+					      SW_SYNC_FENCE_STATUS_SIGNALED);
+	igt_assert_f(active == 0 && signaled == 3,
+		     "Fence did not signal properly\n");
+
+	/* confirm you can successfully wait */
+	ret = sw_sync_wait(fence_merge, 100);
+	igt_assert_f(ret > 0, "Failure waiting on signaled fence\n");
+
+	sw_sync_fence_destroy(in_fence[0]);
+	sw_sync_fence_destroy(in_fence[1]);
+	sw_sync_fence_destroy(in_fence[2]);
+	sw_sync_fence_destroy(fence_merge);
+	sw_sync_timeline_destroy(timeline[0]);
+	sw_sync_timeline_destroy(timeline[1]);
+	sw_sync_timeline_destroy(timeline[2]);
+}
+
 static void * test_sync_multi_consumer_thread(void *arg)
 {
 	data_t *data = arg;
@@ -476,6 +539,9 @@ igt_main
 	igt_subtest("sync_merge_same")
 		test_sync_merge_same();
 
+	igt_subtest("sync_multi_timeline_wait")
+		test_sync_multi_timeline_wait();
+
 	igt_subtest("sync_multi_consumer")
 		test_sync_multi_consumer();
 
Signature · 6 lines
2.9.3

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

PATCH v2 13/13 tests/sw_sync: Add subtest test_sync_multi_producer_single_consumer

In reply to [email protected]

From: Robert Foss <[email protected]>

This subtest runs a single consumer thread and multiple producer thread that are synchronized using multiple timelines.

---
 tests/sw_sync.c | 139 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 139 insertions(+)
tests/sw_sync.c +139 -0
index 322c7b1..4336659 100644
@@ -446,6 +446,142 @@ static void test_sync_multi_consumer_producer(void)
 	igt_assert_f(thread_ret == 0, "A sync thread reported failure.\n");
 }
 
+static int test_mspc_wait_on_fence(int fence)
+{
+	int error, active;
+
+	do {
+		error = sw_sync_fence_count_status(fence,
+						   SW_SYNC_FENCE_STATUS_ERROR);
+		igt_assert_f(error == 0, "Error occurred on fence\n");
+		active = sw_sync_fence_count_status(fence,
+						    SW_SYNC_FENCE_STATUS_ACTIVE);
+	} while (active);
+
+	return 0;
+}
+
+static struct {
+	int iterations;
+	int threads;
+	int counter;
+	int cons_timeline;
+	int *prod_timeline;
+	pthread_mutex_t lock;
+} test_mpsc_data;
+
+static int mpsc_producer_thread(void *d)
+{
+	int id = (long)d;
+	int fence, i;
+	int *prod_timeline = test_mpsc_data.prod_timeline;
+	int cons_timeline = test_mpsc_data.cons_timeline;
+	int iterations = test_mpsc_data.iterations;
+
+	for (i = 0; i < iterations; i++) {
+		fence = sw_sync_fence_create(cons_timeline, i);
+
+		/* Wait for the consumer to finish. Use alternate
+		 * means of waiting on the fence
+		 */
+		if ((iterations + id) % 8 != 0) {
+			igt_assert_f(sw_sync_wait(fence, -1) > 0,
+				     "Failure waiting on fence\n");
+		} else {
+			igt_assert_f(test_mspc_wait_on_fence(fence) == 0,
+				     "Failure waiting on fence\n");
+		}
+
+		/* Every producer increments the counter, the consumer
+		 * checks and erases it
+		 */
+		pthread_mutex_lock(&test_mpsc_data.lock);
+		test_mpsc_data.counter++;
+		pthread_mutex_unlock(&test_mpsc_data.lock);
+
+		sw_sync_timeline_inc(prod_timeline[id], 1);
+		sw_sync_fence_destroy(fence);
+	}
+
+	return 0;
+}
+
+static int mpsc_consumer_thread(void)
+{
+	int fence, merged, tmp, it, i;
+	int *prod_timeline = test_mpsc_data.prod_timeline;
+	int cons_timeline = test_mpsc_data.cons_timeline;
+	int iterations = test_mpsc_data.iterations;
+	int n = test_mpsc_data.threads;
+
+	for (it = 1; it <= iterations; it++) {
+		fence = sw_sync_fence_create(prod_timeline[0], it);
+		for (i = 1; i < n; i++) {
+			tmp = sw_sync_fence_create(prod_timeline[i], it);
+			merged = sw_sync_merge(tmp, fence);
+			sw_sync_fence_destroy(tmp);
+			sw_sync_fence_destroy(fence);
+			fence = merged;
+		}
+
+		/* Make sure we see an increment from every producer thread.
+		 * Vary the means by which we wait.
+		 */
+		if (iterations % 8 != 0) {
+			igt_assert_f(sw_sync_wait(fence, -1) == 0,
+				    "Producers did not increment as expected\n");
+		} else {
+			igt_assert_f(test_mspc_wait_on_fence(fence) == 0,
+				     "Failure waiting on fence\n");
+		}
+
+		igt_assert_f(test_mpsc_data.counter == n * it,
+			     "Counter value mismatch\n");
+
+		/* Release the producer threads */
+		sw_sync_timeline_inc(cons_timeline, 1);
+		sw_sync_fence_destroy(fence);
+	}
+
+	return 0;
+}
+
+/* IMPORTANT NOTE: if you see this test failing on your system, it may be
+ * due to a shortage of file descriptors. Please ensure your system has
+ * a sensible limit for this test to finish correctly.
+ */
+static void test_sync_multi_producer_single_consumer(void)
+{
+	int iterations = 1 << 12;
+	int n = 5;
+	int prod_timeline[n];
+	int cons_timeline;
+	pthread_t threads[n];
+	long i;
+
+	cons_timeline = sw_sync_timeline_create();
+	for (i = 0; i < n; i++)
+		prod_timeline[i] = sw_sync_timeline_create();
+
+	test_mpsc_data.prod_timeline = prod_timeline;
+	test_mpsc_data.cons_timeline = cons_timeline;
+	test_mpsc_data.iterations = iterations;
+	test_mpsc_data.threads = n;
+	test_mpsc_data.counter = 0;
+	pthread_mutex_init(&test_mpsc_data.lock, NULL);
+
+	for (i = 0; i < n; i++) {
+		pthread_create(&threads[i], NULL, (void * (*)(void *))
+			       mpsc_producer_thread,
+			       (void *)i);
+	}
+
+	mpsc_consumer_thread();
+
+	for (i = 0; i < n; i++)
+		pthread_join(threads[i], NULL);
+}
+
 static void test_sync_random_merge(void)
 {
 	int i, size, ret;
@@ -548,6 +684,9 @@ igt_main
 	igt_subtest("sync_multi_consumer_producer")
 		test_sync_multi_consumer_producer();
 
+	igt_subtest("sync_multi_producer_single_consumer")
+		test_sync_multi_producer_single_consumer();
+
 	igt_subtest("sync_random_merge")
 		test_sync_random_merge();
 }
Signature · 6 lines
2.9.3

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