Previously, in order to set the numa_node attribute at the time of map
creation using "libbpf", it was necessary to call bpf_create_map_node()
directly (bpf_load approach), instead of calling bpf_object_load()
that handles everything on its own, including map creation. And because
of this problem, this sample had problems with refactoring from bpf_load
to libbbpf.
However, by commit 1bdb6c9a1c ("libbpf: Add a bunch of attribute
getters/setters for map definitions") added the numa_node attribute and
allowed it to be set in the map.
By using libbpf instead of bpf_load, the inner map definition has
been explicitly declared with BTF-defined format. Also, the element of
ARRAY_OF_MAPS was also statically specified using the BTF format. And
for this reason some logic in fixup_map() was not needed and changed
or removed.
Signed-off-by: Daniel T. Lee <danieltimlee@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/20200707184855.30968-4-danieltimlee@gmail.com
Currently, BPF programs with kprobe/sys_connect does not work properly.
Commit 34745aed51 ("samples/bpf: fix kprobe attachment issue on x64")
This commit modifies the bpf_load behavior of kprobe events in the x64
architecture. If the current kprobe event target starts with "sys_*",
add the prefix "__x64_" to the front of the event.
Appending "__x64_" prefix with kprobe/sys_* event was appropriate as a
solution to most of the problems caused by the commit below.
commit d5a00528b5 ("syscalls/core, syscalls/x86: Rename struct
pt_regs-based sys_*() to __x64_sys_*()")
However, there is a problem with the sys_connect kprobe event that does
not work properly. For __sys_connect event, parameters can be fetched
normally, but for __x64_sys_connect, parameters cannot be fetched.
ffffffff818d3520 <__x64_sys_connect>:
ffffffff818d3520: e8 fb df 32 00 callq 0xffffffff81c01520
<__fentry__>
ffffffff818d3525: 48 8b 57 60 movq 96(%rdi), %rdx
ffffffff818d3529: 48 8b 77 68 movq 104(%rdi), %rsi
ffffffff818d352d: 48 8b 7f 70 movq 112(%rdi), %rdi
ffffffff818d3531: e8 1a ff ff ff callq 0xffffffff818d3450
<__sys_connect>
ffffffff818d3536: 48 98 cltq
ffffffff818d3538: c3 retq
ffffffff818d3539: 0f 1f 80 00 00 00 00 nopl (%rax)
As the assembly code for __x64_sys_connect shows, parameters should be
fetched and set into rdi, rsi, rdx registers prior to calling
__sys_connect.
Because of this problem, this commit fixes the sys_connect event by
first getting the value of the rdi register and then the value of the
rdi, rsi, and rdx register through an offset based on that value.
Fixes: 34745aed51 ("samples/bpf: fix kprobe attachment issue on x64")
Signed-off-by: Daniel T. Lee <danieltimlee@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/20200707184855.30968-2-danieltimlee@gmail.com
Fix all files in samples/bpf to include libbpf header files with the bpf/
prefix, to be consistent with external users of the library. Also ensure
that all includes of exported libbpf header files (those that are exported
on 'make install' of the library) use bracketed includes instead of quoted.
To make sure no new files are introduced that doesn't include the bpf/
prefix in its include, remove tools/lib/bpf from the include path entirely,
and use tools/lib instead.
Fixes: 6910d7d386 ("selftests/bpf: Ensure bpf_helper_defs.h are taken from selftests dir")
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/157952560911.1683545.8795966751309534150.stgit@toke.dk
Split-off PT_REGS-related helpers into bpf_tracing.h header. Adjust
selftests and samples to include it where necessary.
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20191008175942.1769476-5-andriin@fb.com
Split off few legacy things from bpf_helpers.h into separate
bpf_legacy.h file:
- load_{byte|half|word};
- remove extra inner_idx and numa_node fields from bpf_map_def and
introduce bpf_map_def_legacy for use in samples;
- move BPF_ANNOTATE_KV_PAIR into bpf_legacy.h.
Adjust samples and selftests accordingly by either including
bpf_legacy.h and using bpf_map_def_legacy, or switching to BTF-defined
maps altogether.
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20191008175942.1769476-3-andriin@fb.com
When cross-compiling the bpf sample map_perf_test for aarch64, I find that
__NR_getpgrp is undefined. This causes build errors. This syscall is deprecated
and requires defining __ARCH_WANT_SYSCALL_DEPRECATED. To avoid having to define
that, just use a different syscall (getppid) for the array map stress test.
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Joel Fernandes <joelaf@google.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Create a new case to test the LRU lookup performance.
At the beginning, the LRU map is fully loaded (i.e. the number of keys
is equal to map->max_entries). The lookup is done through key 0
to num_map_entries and then repeats from 0 again.
This patch also creates an anonymous struct to properly
name the test params in stress_lru_hmap_alloc() in map_perf_test_kern.c.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch makes the needed changes to allow each process of
the INNER_LRU_HASH_PREALLOC test to provide its numa node id
when creating the lru map.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds a map-in-map LRU example.
If we know only a subset of cores will use the
LRU, we can allocate a common LRU list per targeting core
and store it into an array-of-hashs.
It allows using the common LRU map with map-update performance
comparable to the BPF_F_NO_COMMON_LRU map but without wasting memory
on the unused cores that we know they will never access the LRU map.
BPF_F_NO_COMMON_LRU:
> map_perf_test 32 8 10000000 10000000 | awk '{sum += $3}END{print sum}'
9234314 (9.23M/s)
map-in-map LRU:
> map_perf_test 512 8 1260000 80000000 | awk '{sum += $3}END{print sum}'
9962743 (9.96M/s)
Notes that the max_entries for the map-in-map LRU test is 1260000 which
is the max_entries for each inner LRU map. 8 processes have been
started, so 8 * 1260000 = 10080000 (~10M) which is close to what is
used in the BPF_F_NO_COMMON_LRU test.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
One more LRU test will be added later in this patch series.
In this patch, we first move all existing LRU map tests into
a single syscall (connect) first so that the future new
LRU test can be added without hunting another syscall.
One of the map name is also changed from percpu_lru_hash_map
to nocommon_lru_hash_map to avoid the confusion with percpu_hash_map.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
$ map_perf_test 128
speed of HASH bpf_map_lookup_elem() in lookups per second
w/o JIT w/JIT
before 46M 58M
after 42M 74M
perf report
before:
54.23% map_perf_test [kernel.kallsyms] [k] __htab_map_lookup_elem
14.24% map_perf_test [kernel.kallsyms] [k] lookup_elem_raw
8.84% map_perf_test [kernel.kallsyms] [k] htab_map_lookup_elem
5.93% map_perf_test [kernel.kallsyms] [k] bpf_map_lookup_elem
2.30% map_perf_test [kernel.kallsyms] [k] bpf_prog_da4fc6a3f41761a2
1.49% map_perf_test [kernel.kallsyms] [k] kprobe_ftrace_handler
after:
60.03% map_perf_test [kernel.kallsyms] [k] __htab_map_lookup_elem
18.07% map_perf_test [kernel.kallsyms] [k] lookup_elem_raw
2.91% map_perf_test [kernel.kallsyms] [k] bpf_prog_da4fc6a3f41761a2
1.94% map_perf_test [kernel.kallsyms] [k] _einittext
1.90% map_perf_test [kernel.kallsyms] [k] __audit_syscall_exit
1.72% map_perf_test [kernel.kallsyms] [k] kprobe_ftrace_handler
Notice that bpf_map_lookup_elem() and htab_map_lookup_elem() are trivial
functions, yet they take sizeable amount of cpu time.
htab_map_gen_lookup() removes bpf_map_lookup_elem() and converts
htab_map_lookup_elem() into three BPF insns which causing cpu time
for bpf_prog_da4fc6a3f41761a2() slightly increase.
$ map_perf_test 256
speed of ARRAY bpf_map_lookup_elem() in lookups per second
w/o JIT w/JIT
before 97M 174M
after 64M 280M
before:
37.33% map_perf_test [kernel.kallsyms] [k] array_map_lookup_elem
13.95% map_perf_test [kernel.kallsyms] [k] bpf_map_lookup_elem
6.54% map_perf_test [kernel.kallsyms] [k] bpf_prog_da4fc6a3f41761a2
4.57% map_perf_test [kernel.kallsyms] [k] kprobe_ftrace_handler
after:
32.86% map_perf_test [kernel.kallsyms] [k] bpf_prog_da4fc6a3f41761a2
6.54% map_perf_test [kernel.kallsyms] [k] kprobe_ftrace_handler
array_map_gen_lookup() removes calls to array_map_lookup_elem()
and bpf_map_lookup_elem() and replaces them with 7 bpf insns.
The performance without JIT is slower, since executing extra insns
in the interpreter is slower than running native C code,
but with JIT the performance gains are obvious,
since native C->x86 code is replaced with fewer bpf->x86 instructions.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Extend the map_perf_test_{user,kern}.c infrastructure to stress test
lpm-trie lookups. We hook into the kprobe on sys_gettid() and measure
the latency depending on trie size and lookup count.
On my Intel Haswell i7-6400U, a single gettid() syscall with an empty
bpf program takes roughly 6.5us on my system. Lookups in empty tries
take ~1.8us on first try, ~0.9us on retries. Lookups in tries with 8192
entries take ~7.1us (on the first _and_ any subsequent try).
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Daniel Mack <daniel@zonque.org>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch has some unit tests and a test_lru_dist.
The test_lru_dist reads in the numeric keys from a file.
The files used here are generated by a modified fio-genzipf tool
originated from the fio test suit. The sample data file can be
found here: https://github.com/iamkafai/bpf-lru
The zipf.* data files have 100k numeric keys and the key is also
ranged from 1 to 100k.
The test_lru_dist outputs the number of unique keys (nr_unique).
F.e. The following means, 61239 of them is unique out of 100k keys.
nr_misses means it cannot be found in the LRU map, so nr_misses
must be >= nr_unique. test_lru_dist also simulates a perfect LRU
map as a comparison:
[root@arch-fb-vm1 ~]# ~/devshare/fb-kernel/linux/samples/bpf/test_lru_dist \
/root/zipf.100k.a1_01.out 4000 1
...
test_parallel_lru_dist (map_type:9 map_flags:0x0):
task:0 BPF LRU: nr_unique:23093(/100000) nr_misses:31603(/100000)
task:0 Perfect LRU: nr_unique:23093(/100000 nr_misses:34328(/100000)
....
test_parallel_lru_dist (map_type:9 map_flags:0x2):
task:0 BPF LRU: nr_unique:23093(/100000) nr_misses:31710(/100000)
task:0 Perfect LRU: nr_unique:23093(/100000 nr_misses:34328(/100000)
[root@arch-fb-vm1 ~]# ~/devshare/fb-kernel/linux/samples/bpf/test_lru_dist \
/root/zipf.100k.a0_01.out 40000 1
...
test_parallel_lru_dist (map_type:9 map_flags:0x0):
task:0 BPF LRU: nr_unique:61239(/100000) nr_misses:67054(/100000)
task:0 Perfect LRU: nr_unique:61239(/100000 nr_misses:66993(/100000)
...
test_parallel_lru_dist (map_type:9 map_flags:0x2):
task:0 BPF LRU: nr_unique:61239(/100000) nr_misses:67068(/100000)
task:0 Perfect LRU: nr_unique:61239(/100000 nr_misses:66993(/100000)
LRU map has also been added to map_perf_test:
/* Global LRU */
[root@kerneltest003.31.prn1 ~]# for i in 1 4 8; do echo -n "$i cpus: "; \
./map_perf_test 16 $i | awk '{r += $3}END{print r " updates"}'; done
1 cpus: 2934082 updates
4 cpus: 7391434 updates
8 cpus: 6500576 updates
/* Percpu LRU */
[root@kerneltest003.31.prn1 ~]# for i in 1 4 8; do echo -n "$i cpus: "; \
./map_perf_test 32 $i | awk '{r += $3}END{print r " updates"}'; done
1 cpus: 2896553 updates
4 cpus: 9766395 updates
8 cpus: 17460553 updates
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
performance tests for hash map and per-cpu hash map
with and without pre-allocation
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
