diff mbox series

[bpf-next,v8] selftests/bpf: Add benchmark for bpf memory allocator

Message ID 20230703141332.3319271-1-houtao@huaweicloud.com (mailing list archive)
State Superseded
Delegated to: BPF
Headers show
Series [bpf-next,v8] selftests/bpf: Add benchmark for bpf memory allocator | expand

Checks

Context Check Description
netdev/series_format success Single patches do not need cover letters
netdev/tree_selection success Clearly marked for bpf-next, async
netdev/fixes_present success Fixes tag not required for -next series
netdev/header_inline success No static functions without inline keyword in header files
netdev/build_32bit success Errors and warnings before: 8 this patch: 8
netdev/cc_maintainers warning 3 maintainers not CCed: mykolal@fb.com shuah@kernel.org linux-kselftest@vger.kernel.org
netdev/build_clang fail Errors and warnings before: 18 this patch: 18
netdev/verify_signedoff success Signed-off-by tag matches author and committer
netdev/deprecated_api success None detected
netdev/check_selftest success net selftest script(s) already in Makefile
netdev/verify_fixes success No Fixes tag
netdev/build_allmodconfig_warn success Errors and warnings before: 8 this patch: 8
netdev/checkpatch warning WARNING: Move const after static - use 'static const struct htab_mem_use_case ' WARNING: added, moved or deleted file(s), does MAINTAINERS need updating? WARNING: externs should be avoided in .c files WARNING: line length of 100 exceeds 80 columns WARNING: line length of 81 exceeds 80 columns WARNING: line length of 83 exceeds 80 columns WARNING: line length of 84 exceeds 80 columns WARNING: line length of 85 exceeds 80 columns WARNING: line length of 87 exceeds 80 columns WARNING: line length of 89 exceeds 80 columns WARNING: line length of 90 exceeds 80 columns WARNING: line length of 93 exceeds 80 columns WARNING: line length of 94 exceeds 80 columns WARNING: line length of 95 exceeds 80 columns WARNING: quoted string split across lines
netdev/kdoc success Errors and warnings before: 0 this patch: 0
netdev/source_inline success Was 0 now: 0
bpf/vmtest-bpf-next-PR fail PR summary
bpf/vmtest-bpf-next-VM_Test-1 success Logs for ${{ matrix.test }} on ${{ matrix.arch }} with ${{ matrix.toolchain_full }}
bpf/vmtest-bpf-next-VM_Test-2 success Logs for ShellCheck
bpf/vmtest-bpf-next-VM_Test-3 success Logs for build for aarch64 with gcc
bpf/vmtest-bpf-next-VM_Test-4 success Logs for build for s390x with gcc
bpf/vmtest-bpf-next-VM_Test-5 success Logs for build for x86_64 with gcc
bpf/vmtest-bpf-next-VM_Test-6 fail Logs for build for x86_64 with llvm-16
bpf/vmtest-bpf-next-VM_Test-7 success Logs for set-matrix
bpf/vmtest-bpf-next-VM_Test-8 success Logs for veristat

Commit Message

Hou Tao July 3, 2023, 2:13 p.m. UTC
From: Hou Tao <houtao1@huawei.com>

The benchmark could be used to compare the performance of hash map
operations and the memory usage between different flavors of bpf memory
allocator (e.g., no bpf ma vs bpf ma vs reuse-after-gp bpf ma). It also
could be used to check the performance improvement or the memory saving
provided by optimization.

The benchmark creates a non-preallocated hash map which uses bpf memory
allocator and shows the operation performance and the memory usage of
the hash map under different use cases:
(1) overwrite
Each CPU overwrites nonoverlapping part of hash map. When each CPU
completes overwriting of 64 elements in hash map, it increases the
op_count.
(2) batch_add_batch_del
Each CPU adds then deletes nonoverlapping part of hash map in batch.
When each CPU adds and deletes 64 elements in hash map, it increases
the op_count twice.
(3) add_del_on_diff_cpu
Each two-CPUs pair adds and deletes nonoverlapping part of map
cooperatively. When each CPU adds or deletes 64 elements in hash map,
it will increase the op_count.

The following is the benchmark results when comparing between different
flavors of bpf memory allocator. These tests are conducted on a KVM guest
with 8 CPUs and 16 GB memory. The command line below is used to do all
the following benchmarks:

  ./bench htab-mem --use-case $name ${OPTS} -w3 -d10 -a -p8

These results show that preallocated hash map has both better performance
and smaller memory footprint.

(1) non-preallocated + no bpf memory allocator (v6.0.19)
use kmalloc() + call_rcu

overwrite            per-prod-op: 11.24 ± 0.07k/s, avg mem: 82.64 ± 26.32MiB, peak mem: 119.18MiB
batch_add_batch_del  per-prod-op: 18.45 ± 0.10k/s, avg mem: 50.47 ± 14.51MiB, peak mem: 94.96MiB
add_del_on_diff_cpu  per-prod-op: 14.50 ± 0.03k/s, avg mem: 4.64 ± 0.73MiB, peak mem: 7.20MiB

(2) preallocated
OPTS=--preallocated

overwrite            per-prod-op: 191.92 ± 0.07k/s, avg mem: 1.23 ± 0.00MiB, peak mem: 1.49MiB
batch_add_batch_del  per-prod-op: 218.10 ± 0.25k/s, avg mem: 1.23 ± 0.00MiB, peak mem: 1.49MiB
add_del_on_diff_cpu  per-prod-op: 39.59 ± 0.41k/s, avg mem: 1.48 ± 0.11MiB, peak mem: 1.74MiB

(3) normal bpf memory allocator

overwrite            per-prod-op: 134.81 ± 0.22k/s, avg mem: 1.67 ± 0.12MiB, peak mem: 2.74MiB
batch_add_batch_del  per-prod-op: 90.44 ± 0.34k/s, avg mem: 2.27 ± 0.00MiB, peak mem: 2.74MiB
add_del_on_diff_cpu  per-prod-op: 28.20 ± 0.15k/s, avg mem: 1.73 ± 0.17MiB, peak mem: 2.06MiB

Signed-off-by: Hou Tao <houtao1@huawei.com>
---
Hi,

After hacking htab to use bpf_mem_cache_free_rcu() [0], the htab-mem
benchmark result is as follows:

htab-mem:

overwrite            per-prod-op: 69.37 ± 0.83k/s, avg mem: 137.00 ± 30.10MiB, peak mem: 201.45MiB
batch_add_batch_del  per-prod-op: 76.26 ± 0.54k/s, avg mem: 45.88 ± 2.19MiB, peak mem: 56.27MiB
add_del_on_diff_cpu  per-prod-op: 30.58 ± 0.13k/s, avg mem: 20.06 ± 2.31MiB, peak mem: 28.54MiB

For reference, hash_map_perf benchmark results are shown below. The
command line for the benchmark is "./map_perf_test 4 8 16384".

htab-with-bpf_mem_cache_free()
2:hash_map_perf kmalloc 496369 events per sec
0:hash_map_perf kmalloc 250241 events per sec
1:hash_map_perf kmalloc 248366 events per sec
3:hash_map_perf kmalloc 240521 events per sec
6:hash_map_perf kmalloc 250032 events per sec
7:hash_map_perf kmalloc 250798 events per sec
4:hash_map_perf kmalloc 152847 events per sec
5:hash_map_perf kmalloc 150083 events per sec

htab-with-bpf_mem_cache_free_rc()
2:hash_map_perf kmalloc 294190 events per sec
3:hash_map_perf kmalloc 172039 events per sec
0:hash_map_perf kmalloc 170092 events per sec
5:hash_map_perf kmalloc 170396 events per sec
6:hash_map_perf kmalloc 170030 events per sec
7:hash_map_perf kmalloc 167629 events per sec
1:hash_map_perf kmalloc 162975 events per sec
4:hash_map_perf kmalloc 162673 events per sec

[0]: https://lore.kernel.org/bpf/20230628015634.33193-1-alexei.starovoitov@gmail.com/

Change Log:

v8:
 * use MAX() from sys/param.h instead of redefining it
 * remove unnecessary patch which adds min() & max() in bpf_util.h

v7: https://lore.kernel.org/rcu/20230628115910.3817966-1-houtao@huaweicloud.com/
 * Rename name of producer threads to avoid confusion
 * Make the comments in producer threads more clear
 * Remove unnecessary check of ctx->from in bpf program
 * Split add_del_on_diff bpf program to two bpf program for clarity

v6: https://lore.kernel.org/bpf/20230613080921.1623219-1-houtao@huaweicloud.com/
  * add fix patches for benchmark framework
  * updates for htab-mem benchmark (Most of updates are suggested by Alexei)
    * remove --full and --max-entries and use a fixed 8k size for htab
    * remove op_factor and increase op_cnt correctly
    * use -a instead of --prod-affinity in run_bench_htab_mem.sh
    * use $RUN_BENCH in run_bench_htab_mem.sh
    * call cleanup_cgroup_environment() at the end of htab_mem_report_final()

v5: https://lore.kernel.org/bpf/ff4b2396-48aa-28f1-c91b-7c8a4b9510bb@huaweicloud.com/
 * send the benchmark patch alone (suggested by Alexei)
 * limit the max number of touched elements per-bpf-program call to 64 (from Alexei)
 * show per-producer performance (from Alexei)
 * handle the return value of read() (from BPF CI)
 * do cleanup_cgroup_environment() in htab_mem_report_final()

v4: https://lore.kernel.org/bpf/20230606035310.4026145-1-houtao@huaweicloud.com/

 tools/testing/selftests/bpf/Makefile          |   3 +
 tools/testing/selftests/bpf/bench.c           |   4 +
 .../selftests/bpf/benchs/bench_htab_mem.c     | 346 ++++++++++++++++++
 .../bpf/benchs/run_bench_htab_mem.sh          |  40 ++
 .../selftests/bpf/progs/htab_mem_bench.c      | 105 ++++++
 5 files changed, 498 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/benchs/bench_htab_mem.c
 create mode 100755 tools/testing/selftests/bpf/benchs/run_bench_htab_mem.sh
 create mode 100644 tools/testing/selftests/bpf/progs/htab_mem_bench.c

Comments

John Fastabend July 3, 2023, 9:52 p.m. UTC | #1
Hou Tao wrote:
> From: Hou Tao <houtao1@huawei.com>
> 
> The benchmark could be used to compare the performance of hash map
> operations and the memory usage between different flavors of bpf memory
> allocator (e.g., no bpf ma vs bpf ma vs reuse-after-gp bpf ma). It also
> could be used to check the performance improvement or the memory saving
> provided by optimization.
> 
> The benchmark creates a non-preallocated hash map which uses bpf memory
> allocator and shows the operation performance and the memory usage of
> the hash map under different use cases:
> (1) overwrite
> Each CPU overwrites nonoverlapping part of hash map. When each CPU
> completes overwriting of 64 elements in hash map, it increases the
> op_count.
> (2) batch_add_batch_del
> Each CPU adds then deletes nonoverlapping part of hash map in batch.
> When each CPU adds and deletes 64 elements in hash map, it increases
> the op_count twice.
> (3) add_del_on_diff_cpu
> Each two-CPUs pair adds and deletes nonoverlapping part of map
> cooperatively. When each CPU adds or deletes 64 elements in hash map,
> it will increase the op_count.
> 
> The following is the benchmark results when comparing between different
> flavors of bpf memory allocator. These tests are conducted on a KVM guest
> with 8 CPUs and 16 GB memory. The command line below is used to do all
> the following benchmarks:
> 
>   ./bench htab-mem --use-case $name ${OPTS} -w3 -d10 -a -p8
> 
> These results show that preallocated hash map has both better performance
> and smaller memory footprint.
> 
> (1) non-preallocated + no bpf memory allocator (v6.0.19)
> use kmalloc() + call_rcu
> 
> overwrite            per-prod-op: 11.24 ± 0.07k/s, avg mem: 82.64 ± 26.32MiB, peak mem: 119.18MiB
> batch_add_batch_del  per-prod-op: 18.45 ± 0.10k/s, avg mem: 50.47 ± 14.51MiB, peak mem: 94.96MiB
> add_del_on_diff_cpu  per-prod-op: 14.50 ± 0.03k/s, avg mem: 4.64 ± 0.73MiB, peak mem: 7.20MiB
> 
> (2) preallocated
> OPTS=--preallocated
> 
> overwrite            per-prod-op: 191.92 ± 0.07k/s, avg mem: 1.23 ± 0.00MiB, peak mem: 1.49MiB
> batch_add_batch_del  per-prod-op: 218.10 ± 0.25k/s, avg mem: 1.23 ± 0.00MiB, peak mem: 1.49MiB
> add_del_on_diff_cpu  per-prod-op: 39.59 ± 0.41k/s, avg mem: 1.48 ± 0.11MiB, peak mem: 1.74MiB
> 
> (3) normal bpf memory allocator
> 
> overwrite            per-prod-op: 134.81 ± 0.22k/s, avg mem: 1.67 ± 0.12MiB, peak mem: 2.74MiB
> batch_add_batch_del  per-prod-op: 90.44 ± 0.34k/s, avg mem: 2.27 ± 0.00MiB, peak mem: 2.74MiB
> add_del_on_diff_cpu  per-prod-op: 28.20 ± 0.15k/s, avg mem: 1.73 ± 0.17MiB, peak mem: 2.06MiB

Acked-by: John Fastabend <john.fastabend@gmail.com>

> +
> +static error_t htab_mem_parse_arg(int key, char *arg, struct argp_state *state)
> +{
> +	switch (key) {
> +	case ARG_VALUE_SIZE:
> +		args.value_size = strtoul(arg, NULL, 10);
> +		if (args.value_size > 4096) {
> +			fprintf(stderr, "too big value size %u\n", args.value_size);
> +			argp_usage(state);
> +		}
> +		break;
> +	case ARG_USE_CASE:
> +		args.use_case = strdup(arg);

might be worth checking for null and returning an error? Only matters if we
run from CI or something and then this looks like a flake.

> +		break;
> +	case ARG_PREALLOCATED:
> +		args.preallocated = true;
> +		break;
> +	default:
> +		return ARGP_ERR_UNKNOWN;
> +	}
> +
> +	return 0;
> +}
Hou Tao July 4, 2023, 1:13 a.m. UTC | #2
Hi,

On 7/4/2023 5:52 AM, John Fastabend wrote:
> Hou Tao wrote:
>> From: Hou Tao <houtao1@huawei.com>
>>
>> The benchmark could be used to compare the performance of hash map
>> operations and the memory usage between different flavors of bpf memory
>> allocator (e.g., no bpf ma vs bpf ma vs reuse-after-gp bpf ma). It also
>> could be used to check the performance improvement or the memory saving
>> provided by optimization.
>>

SNIP
>> overwrite            per-prod-op: 191.92 ± 0.07k/s, avg mem: 1.23 ± 0.00MiB, peak mem: 1.49MiB
>> batch_add_batch_del  per-prod-op: 218.10 ± 0.25k/s, avg mem: 1.23 ± 0.00MiB, peak mem: 1.49MiB
>> add_del_on_diff_cpu  per-prod-op: 39.59 ± 0.41k/s, avg mem: 1.48 ± 0.11MiB, peak mem: 1.74MiB
>>
>> (3) normal bpf memory allocator
>>
>> overwrite            per-prod-op: 134.81 ± 0.22k/s, avg mem: 1.67 ± 0.12MiB, peak mem: 2.74MiB
>> batch_add_batch_del  per-prod-op: 90.44 ± 0.34k/s, avg mem: 2.27 ± 0.00MiB, peak mem: 2.74MiB
>> add_del_on_diff_cpu  per-prod-op: 28.20 ± 0.15k/s, avg mem: 1.73 ± 0.17MiB, peak mem: 2.06MiB
> Acked-by: John Fastabend <john.fastabend@gmail.com>
Thanks for the Ack.
>
>> +
>> +static error_t htab_mem_parse_arg(int key, char *arg, struct argp_state *state)
>> +{
>> +	switch (key) {
>> +	case ARG_VALUE_SIZE:
>> +		args.value_size = strtoul(arg, NULL, 10);
>> +		if (args.value_size > 4096) {
>> +			fprintf(stderr, "too big value size %u\n", args.value_size);
>> +			argp_usage(state);
>> +		}
>> +		break;
>> +	case ARG_USE_CASE:
>> +		args.use_case = strdup(arg);
> might be worth checking for null and returning an error? Only matters if we
> run from CI or something and then this looks like a flake.
Will fix. I also found an error reported from BPF CI llvm-16 build
procedure: "benchs/bench_htab_mem.c:45:27: error: initializer element is
not a compile-time constant". Will fix in v9.
>
>> +		break;
>> +	case ARG_PREALLOCATED:
>> +		args.preallocated = true;
>> +		break;
>> +	default:
>> +		return ARGP_ERR_UNKNOWN;
>> +	}
>> +
>> +	return 0;
>> +}
diff mbox series

Patch

diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile
index 28d2c77262be..e49234bba02e 100644
--- a/tools/testing/selftests/bpf/Makefile
+++ b/tools/testing/selftests/bpf/Makefile
@@ -645,11 +645,13 @@  $(OUTPUT)/bench_local_storage.o: $(OUTPUT)/local_storage_bench.skel.h
 $(OUTPUT)/bench_local_storage_rcu_tasks_trace.o: $(OUTPUT)/local_storage_rcu_tasks_trace_bench.skel.h
 $(OUTPUT)/bench_local_storage_create.o: $(OUTPUT)/bench_local_storage_create.skel.h
 $(OUTPUT)/bench_bpf_hashmap_lookup.o: $(OUTPUT)/bpf_hashmap_lookup.skel.h
+$(OUTPUT)/bench_htab_mem.o: $(OUTPUT)/htab_mem_bench.skel.h
 $(OUTPUT)/bench.o: bench.h testing_helpers.h $(BPFOBJ)
 $(OUTPUT)/bench: LDLIBS += -lm
 $(OUTPUT)/bench: $(OUTPUT)/bench.o \
 		 $(TESTING_HELPERS) \
 		 $(TRACE_HELPERS) \
+		 $(CGROUP_HELPERS) \
 		 $(OUTPUT)/bench_count.o \
 		 $(OUTPUT)/bench_rename.o \
 		 $(OUTPUT)/bench_trigger.o \
@@ -662,6 +664,7 @@  $(OUTPUT)/bench: $(OUTPUT)/bench.o \
 		 $(OUTPUT)/bench_local_storage_rcu_tasks_trace.o \
 		 $(OUTPUT)/bench_bpf_hashmap_lookup.o \
 		 $(OUTPUT)/bench_local_storage_create.o \
+		 $(OUTPUT)/bench_htab_mem.o \
 		 #
 	$(call msg,BINARY,,$@)
 	$(Q)$(CC) $(CFLAGS) $(LDFLAGS) $(filter %.a %.o,$^) $(LDLIBS) -o $@
diff --git a/tools/testing/selftests/bpf/bench.c b/tools/testing/selftests/bpf/bench.c
index 41fe5a82b88b..73ce11b0547d 100644
--- a/tools/testing/selftests/bpf/bench.c
+++ b/tools/testing/selftests/bpf/bench.c
@@ -279,6 +279,7 @@  extern struct argp bench_local_storage_rcu_tasks_trace_argp;
 extern struct argp bench_strncmp_argp;
 extern struct argp bench_hashmap_lookup_argp;
 extern struct argp bench_local_storage_create_argp;
+extern struct argp bench_htab_mem_argp;
 
 static const struct argp_child bench_parsers[] = {
 	{ &bench_ringbufs_argp, 0, "Ring buffers benchmark", 0 },
@@ -290,6 +291,7 @@  static const struct argp_child bench_parsers[] = {
 		"local_storage RCU Tasks Trace slowdown benchmark", 0 },
 	{ &bench_hashmap_lookup_argp, 0, "Hashmap lookup benchmark", 0 },
 	{ &bench_local_storage_create_argp, 0, "local-storage-create benchmark", 0 },
+	{ &bench_htab_mem_argp, 0, "hash map memory benchmark", 0 },
 	{},
 };
 
@@ -520,6 +522,7 @@  extern const struct bench bench_local_storage_cache_hashmap_control;
 extern const struct bench bench_local_storage_tasks_trace;
 extern const struct bench bench_bpf_hashmap_lookup;
 extern const struct bench bench_local_storage_create;
+extern const struct bench bench_htab_mem;
 
 static const struct bench *benchs[] = {
 	&bench_count_global,
@@ -561,6 +564,7 @@  static const struct bench *benchs[] = {
 	&bench_local_storage_tasks_trace,
 	&bench_bpf_hashmap_lookup,
 	&bench_local_storage_create,
+	&bench_htab_mem,
 };
 
 static void find_benchmark(void)
diff --git a/tools/testing/selftests/bpf/benchs/bench_htab_mem.c b/tools/testing/selftests/bpf/benchs/bench_htab_mem.c
new file mode 100644
index 000000000000..96172bffd91f
--- /dev/null
+++ b/tools/testing/selftests/bpf/benchs/bench_htab_mem.c
@@ -0,0 +1,346 @@ 
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (C) 2023. Huawei Technologies Co., Ltd */
+#include <argp.h>
+#include <stdbool.h>
+#include <pthread.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/param.h>
+#include <fcntl.h>
+
+#include "bench.h"
+#include "bpf_util.h"
+#include "cgroup_helpers.h"
+#include "htab_mem_bench.skel.h"
+
+struct htab_mem_use_case {
+	const char *name;
+	const char **progs;
+	/* Do synchronization between addition thread and deletion thread */
+	bool need_sync;
+};
+
+static struct htab_mem_ctx {
+	const struct htab_mem_use_case *uc;
+	struct htab_mem_bench *skel;
+	pthread_barrier_t *notify;
+	int fd;
+} ctx;
+
+const char *ow_progs[] = {"overwrite", NULL};
+const char *batch_progs[] = {"batch_add_batch_del", NULL};
+const char *add_del_progs[] = {"add_only", "del_only", NULL};
+const static struct htab_mem_use_case use_cases[] = {
+	{ .name = "overwrite", .progs = ow_progs },
+	{ .name = "batch_add_batch_del", .progs = batch_progs },
+	{ .name = "add_del_on_diff_cpu", .progs = add_del_progs, .need_sync = true },
+};
+
+static struct htab_mem_args {
+	u32 value_size;
+	const char *use_case;
+	bool preallocated;
+} args = {
+	.value_size = 8,
+	.use_case = use_cases[0].name,
+	.preallocated = false,
+};
+
+enum {
+	ARG_VALUE_SIZE = 10000,
+	ARG_USE_CASE = 10001,
+	ARG_PREALLOCATED = 10002,
+};
+
+static const struct argp_option opts[] = {
+	{ "value-size", ARG_VALUE_SIZE, "VALUE_SIZE", 0,
+	  "Set the value size of hash map (default 8)" },
+	{ "use-case", ARG_USE_CASE, "USE_CASE", 0,
+	  "Set the use case of hash map: overwrite|batch_add_batch_del|add_del_on_diff_cpu" },
+	{ "preallocated", ARG_PREALLOCATED, NULL, 0, "use preallocated hash map" },
+	{},
+};
+
+static error_t htab_mem_parse_arg(int key, char *arg, struct argp_state *state)
+{
+	switch (key) {
+	case ARG_VALUE_SIZE:
+		args.value_size = strtoul(arg, NULL, 10);
+		if (args.value_size > 4096) {
+			fprintf(stderr, "too big value size %u\n", args.value_size);
+			argp_usage(state);
+		}
+		break;
+	case ARG_USE_CASE:
+		args.use_case = strdup(arg);
+		break;
+	case ARG_PREALLOCATED:
+		args.preallocated = true;
+		break;
+	default:
+		return ARGP_ERR_UNKNOWN;
+	}
+
+	return 0;
+}
+
+const struct argp bench_htab_mem_argp = {
+	.options = opts,
+	.parser = htab_mem_parse_arg,
+};
+
+static void htab_mem_validate(void)
+{
+	if (!strcmp(use_cases[2].name, args.use_case) && env.producer_cnt % 2) {
+		fprintf(stderr, "%s needs an even number of producers\n", args.use_case);
+		exit(1);
+	}
+}
+
+static int htab_mem_bench_init_barriers(void)
+{
+	pthread_barrier_t *barriers;
+	unsigned int i, nr;
+
+	if (!ctx.uc->need_sync)
+		return 0;
+
+	nr = (env.producer_cnt + 1) / 2;
+	barriers = calloc(nr, sizeof(*barriers));
+	if (!barriers)
+		return -1;
+
+	/* Used for synchronization between two threads */
+	for (i = 0; i < nr; i++)
+		pthread_barrier_init(&barriers[i], NULL, 2);
+
+	ctx.notify = barriers;
+	return 0;
+}
+
+static void htab_mem_bench_exit_barriers(void)
+{
+	unsigned int i, nr;
+
+	if (!ctx.notify)
+		return;
+
+	nr = (env.producer_cnt + 1) / 2;
+	for (i = 0; i < nr; i++)
+		pthread_barrier_destroy(&ctx.notify[i]);
+	free(ctx.notify);
+}
+
+static const struct htab_mem_use_case *htab_mem_find_use_case_or_exit(const char *name)
+{
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(use_cases); i++) {
+		if (!strcmp(name, use_cases[i].name))
+			return &use_cases[i];
+	}
+
+	fprintf(stderr, "no such use-case: %s\n", name);
+	fprintf(stderr, "available use case:");
+	for (i = 0; i < ARRAY_SIZE(use_cases); i++)
+		fprintf(stderr, " %s", use_cases[i].name);
+	fprintf(stderr, "\n");
+	exit(1);
+}
+
+static void htab_mem_setup(void)
+{
+	struct bpf_map *map;
+	const char **names;
+	int err;
+
+	setup_libbpf();
+
+	ctx.uc = htab_mem_find_use_case_or_exit(args.use_case);
+	err = htab_mem_bench_init_barriers();
+	if (err) {
+		fprintf(stderr, "failed to init barrier\n");
+		exit(1);
+	}
+
+	ctx.fd = cgroup_setup_and_join("/htab_mem");
+	if (ctx.fd < 0)
+		goto cleanup;
+
+	ctx.skel = htab_mem_bench__open();
+	if (!ctx.skel) {
+		fprintf(stderr, "failed to open skeleton\n");
+		goto cleanup;
+	}
+
+	map = ctx.skel->maps.htab;
+	bpf_map__set_value_size(map, args.value_size);
+	/* Ensure that different CPUs can operate on different subset */
+	bpf_map__set_max_entries(map, MAX(8192, 64 * env.nr_cpus));
+	if (args.preallocated)
+		bpf_map__set_map_flags(map, bpf_map__map_flags(map) & ~BPF_F_NO_PREALLOC);
+
+	names = ctx.uc->progs;
+	while (*names) {
+		struct bpf_program *prog;
+
+		prog = bpf_object__find_program_by_name(ctx.skel->obj, *names);
+		if (!prog) {
+			fprintf(stderr, "no such program %s\n", *names);
+			goto cleanup;
+		}
+		bpf_program__set_autoload(prog, true);
+		names++;
+	}
+	ctx.skel->bss->nr_thread = env.producer_cnt;
+
+	err = htab_mem_bench__load(ctx.skel);
+	if (err) {
+		fprintf(stderr, "failed to load skeleton\n");
+		goto cleanup;
+	}
+	err = htab_mem_bench__attach(ctx.skel);
+	if (err) {
+		fprintf(stderr, "failed to attach skeleton\n");
+		goto cleanup;
+	}
+	return;
+
+cleanup:
+	htab_mem_bench__destroy(ctx.skel);
+	htab_mem_bench_exit_barriers();
+	if (ctx.fd >= 0) {
+		close(ctx.fd);
+		cleanup_cgroup_environment();
+	}
+	exit(1);
+}
+
+static void htab_mem_add_fn(pthread_barrier_t *notify)
+{
+	while (true) {
+		/* Do addition */
+		(void)syscall(__NR_getpgid, 0);
+		/* Notify deletion thread to do deletion */
+		pthread_barrier_wait(notify);
+		/* Wait for deletion to complete */
+		pthread_barrier_wait(notify);
+	}
+}
+
+static void htab_mem_delete_fn(pthread_barrier_t *notify)
+{
+	while (true) {
+		/* Wait for addition to complete */
+		pthread_barrier_wait(notify);
+		/* Do deletion */
+		(void)syscall(__NR_getppid);
+		/* Notify addition thread to do addition */
+		pthread_barrier_wait(notify);
+	}
+}
+
+static void *htab_mem_producer(void *arg)
+{
+	pthread_barrier_t *notify;
+	int seq;
+
+	if (!ctx.uc->need_sync) {
+		while (true)
+			(void)syscall(__NR_getpgid, 0);
+		return NULL;
+	}
+
+	seq = (long)arg;
+	notify = &ctx.notify[seq / 2];
+	if (seq & 1)
+		htab_mem_delete_fn(notify);
+	else
+		htab_mem_add_fn(notify);
+	return NULL;
+}
+
+static void htab_mem_read_mem_cgrp_file(const char *name, unsigned long *value)
+{
+	char buf[32];
+	ssize_t got;
+	int fd;
+
+	fd = openat(ctx.fd, name, O_RDONLY);
+	if (fd < 0) {
+		/* cgroup v1 ? */
+		fprintf(stderr, "no %s\n", name);
+		*value = 0;
+		return;
+	}
+
+	got = read(fd, buf, sizeof(buf) - 1);
+	if (got <= 0) {
+		*value = 0;
+		return;
+	}
+	buf[got] = 0;
+
+	*value = strtoull(buf, NULL, 0);
+
+	close(fd);
+}
+
+static void htab_mem_measure(struct bench_res *res)
+{
+	res->hits = atomic_swap(&ctx.skel->bss->op_cnt, 0) / env.producer_cnt;
+	htab_mem_read_mem_cgrp_file("memory.current", &res->gp_ct);
+}
+
+static void htab_mem_report_progress(int iter, struct bench_res *res, long delta_ns)
+{
+	double loop, mem;
+
+	loop = res->hits / 1000.0 / (delta_ns / 1000000000.0);
+	mem = res->gp_ct / 1048576.0;
+	printf("Iter %3d (%7.3lfus): ", iter, (delta_ns - 1000000000) / 1000.0);
+	printf("per-prod-op %7.2lfk/s, memory usage %7.2lfMiB\n", loop, mem);
+}
+
+static void htab_mem_report_final(struct bench_res res[], int res_cnt)
+{
+	double mem_mean = 0.0, mem_stddev = 0.0;
+	double loop_mean = 0.0, loop_stddev = 0.0;
+	unsigned long peak_mem;
+	int i;
+
+	for (i = 0; i < res_cnt; i++) {
+		loop_mean += res[i].hits / 1000.0 / (0.0 + res_cnt);
+		mem_mean += res[i].gp_ct / 1048576.0 / (0.0 + res_cnt);
+	}
+	if (res_cnt > 1)  {
+		for (i = 0; i < res_cnt; i++) {
+			loop_stddev += (loop_mean - res[i].hits / 1000.0) *
+				       (loop_mean - res[i].hits / 1000.0) /
+				       (res_cnt - 1.0);
+			mem_stddev += (mem_mean - res[i].gp_ct / 1048576.0) *
+				      (mem_mean - res[i].gp_ct / 1048576.0) /
+				      (res_cnt - 1.0);
+		}
+		loop_stddev = sqrt(loop_stddev);
+		mem_stddev = sqrt(mem_stddev);
+	}
+
+	htab_mem_read_mem_cgrp_file("memory.peak", &peak_mem);
+	printf("Summary: per-prod-op %7.2lf \u00B1 %7.2lfk/s, memory usage %7.2lf \u00B1 %7.2lfMiB,"
+	       " peak memory usage %7.2lfMiB\n",
+	       loop_mean, loop_stddev, mem_mean, mem_stddev, peak_mem / 1048576.0);
+
+	cleanup_cgroup_environment();
+}
+
+const struct bench bench_htab_mem = {
+	.name = "htab-mem",
+	.argp = &bench_htab_mem_argp,
+	.validate = htab_mem_validate,
+	.setup = htab_mem_setup,
+	.producer_thread = htab_mem_producer,
+	.measure = htab_mem_measure,
+	.report_progress = htab_mem_report_progress,
+	.report_final = htab_mem_report_final,
+};
diff --git a/tools/testing/selftests/bpf/benchs/run_bench_htab_mem.sh b/tools/testing/selftests/bpf/benchs/run_bench_htab_mem.sh
new file mode 100755
index 000000000000..9ff5832463a2
--- /dev/null
+++ b/tools/testing/selftests/bpf/benchs/run_bench_htab_mem.sh
@@ -0,0 +1,40 @@ 
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0
+
+source ./benchs/run_common.sh
+
+set -eufo pipefail
+
+htab_mem()
+{
+	echo -n "per-prod-op: "
+	echo -n "$*" | sed -E "s/.* per-prod-op\s+([0-9]+\.[0-9]+ ± [0-9]+\.[0-9]+k\/s).*/\1/"
+	echo -n -e ", avg mem: "
+	echo -n "$*" | sed -E "s/.* memory usage\s+([0-9]+\.[0-9]+ ± [0-9]+\.[0-9]+MiB).*/\1/"
+	echo -n ", peak mem: "
+	echo "$*" | sed -E "s/.* peak memory usage\s+([0-9]+\.[0-9]+MiB).*/\1/"
+}
+
+summarize_htab_mem()
+{
+	local bench="$1"
+	local summary=$(echo $2 | tail -n1)
+
+	printf "%-20s %s\n" "$bench" "$(htab_mem $summary)"
+}
+
+htab_mem_bench()
+{
+	local name
+
+	for name in overwrite batch_add_batch_del add_del_on_diff_cpu
+	do
+		summarize_htab_mem "$name" "$($RUN_BENCH htab-mem --use-case $name -p8 "$@")"
+	done
+}
+
+header "preallocated"
+htab_mem_bench "--preallocated"
+
+header "normal bpf ma"
+htab_mem_bench
diff --git a/tools/testing/selftests/bpf/progs/htab_mem_bench.c b/tools/testing/selftests/bpf/progs/htab_mem_bench.c
new file mode 100644
index 000000000000..b1b721b14d67
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/htab_mem_bench.c
@@ -0,0 +1,105 @@ 
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (C) 2023. Huawei Technologies Co., Ltd */
+#include <stdbool.h>
+#include <errno.h>
+#include <linux/types.h>
+#include <linux/bpf.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+#define OP_BATCH 64
+
+struct update_ctx {
+	unsigned int from;
+	unsigned int step;
+};
+
+struct {
+	__uint(type, BPF_MAP_TYPE_HASH);
+	__uint(key_size, 4);
+	__uint(map_flags, BPF_F_NO_PREALLOC);
+} htab SEC(".maps");
+
+char _license[] SEC("license") = "GPL";
+
+unsigned char zeroed_value[4096];
+unsigned int nr_thread = 0;
+long op_cnt = 0;
+
+static int write_htab(unsigned int i, struct update_ctx *ctx, unsigned int flags)
+{
+	bpf_map_update_elem(&htab, &ctx->from, zeroed_value, flags);
+	ctx->from += ctx->step;
+
+	return 0;
+}
+
+static int overwrite_htab(unsigned int i, struct update_ctx *ctx)
+{
+	return write_htab(i, ctx, 0);
+}
+
+static int newwrite_htab(unsigned int i, struct update_ctx *ctx)
+{
+	return write_htab(i, ctx, BPF_NOEXIST);
+}
+
+static int del_htab(unsigned int i, struct update_ctx *ctx)
+{
+	bpf_map_delete_elem(&htab, &ctx->from);
+	ctx->from += ctx->step;
+
+	return 0;
+}
+
+SEC("?tp/syscalls/sys_enter_getpgid")
+int overwrite(void *ctx)
+{
+	struct update_ctx update;
+
+	update.from = bpf_get_smp_processor_id();
+	update.step = nr_thread;
+	bpf_loop(OP_BATCH, overwrite_htab, &update, 0);
+	__sync_fetch_and_add(&op_cnt, 1);
+	return 0;
+}
+
+SEC("?tp/syscalls/sys_enter_getpgid")
+int batch_add_batch_del(void *ctx)
+{
+	struct update_ctx update;
+
+	update.from = bpf_get_smp_processor_id();
+	update.step = nr_thread;
+	bpf_loop(OP_BATCH, overwrite_htab, &update, 0);
+
+	update.from = bpf_get_smp_processor_id();
+	bpf_loop(OP_BATCH, del_htab, &update, 0);
+
+	__sync_fetch_and_add(&op_cnt, 2);
+	return 0;
+}
+
+SEC("?tp/syscalls/sys_enter_getpgid")
+int add_only(void *ctx)
+{
+	struct update_ctx update;
+
+	update.from = bpf_get_smp_processor_id() / 2;
+	update.step = nr_thread / 2;
+	bpf_loop(OP_BATCH, newwrite_htab, &update, 0);
+	__sync_fetch_and_add(&op_cnt, 1);
+	return 0;
+}
+
+SEC("?tp/syscalls/sys_enter_getppid")
+int del_only(void *ctx)
+{
+	struct update_ctx update;
+
+	update.from = bpf_get_smp_processor_id() / 2;
+	update.step = nr_thread / 2;
+	bpf_loop(OP_BATCH, del_htab, &update, 0);
+	__sync_fetch_and_add(&op_cnt, 1);
+	return 0;
+}