From: Hou Tao <houtao1@xxxxxxxxxx> Hi, The initial motivation for qp-trie map is to reduce memory usage for string keys special those with large differencies in length as discussed in [0]. And as a big-endian lexicographical ordered map, it can also be used for any binary data with fixed or variable length. Now the basic functionality of qp-trie is ready, so posting it to get more feedback or suggestions about qp-trie. Specially feedback about the following questions: (1) Use bpf_dynptr in map key Now qp-trie uses bpf_dynptr as map key. For simplicity a new map flags BPF_F_DYNPTR_KEY is introduced, but it is not usable for map key in which a bpf_dynptr is embedded. So maybe a better and generic way is to check the offset of bpf_dynptr in map key during map creation and use the offset when initializing or verifying map key ? (2) Use cases for qp-trie Andrii had proposed to re-implement lpm-trie by using qp-trie. The advantage would be the speed up of lookup operations due to lower tree depth of qp-trie and the performance of update may also increase. But is there any other use cases for qp-trie ? Specially those cases which need both ordering and memory efficiency or cases in which qp-trie will have high branch factor and its lookup performance will be much better than hash-table as shown below: Randomly-generated binary data (key size=255, max entries=16K, key length range:[1, 255]) htab lookup (1 thread) 4.968 ± 0.009M/s (drops 0.002 ± 0.000M/s mem 8.169 MiB) htab lookup (2 thread) 10.118 ± 0.010M/s (drops 0.007 ± 0.000M/s mem 8.169 MiB) htab lookup (4 thread) 20.084 ± 0.022M/s (drops 0.007 ± 0.000M/s mem 8.168 MiB) htab lookup (8 thread) 39.866 ± 0.047M/s (drops 0.010 ± 0.000M/s mem 8.168 MiB) htab lookup (16 thread) 79.412 ± 0.065M/s (drops 0.049 ± 0.000M/s mem 8.169 MiB) qp-trie lookup (1 thread) 10.291 ± 0.007M/s (drops 0.004 ± 0.000M/s mem 4.899 MiB) qp-trie lookup (2 thread) 20.797 ± 0.009M/s (drops 0.006 ± 0.000M/s mem 4.879 MiB) qp-trie lookup (4 thread) 41.943 ± 0.019M/s (drops 0.015 ± 0.000M/s mem 4.262 MiB) qp-trie lookup (8 thread) 81.985 ± 0.032M/s (drops 0.025 ± 0.000M/s mem 4.215 MiB) qp-trie lookup (16 thread) 164.681 ± 0.051M/s (drops 0.050 ± 0.000M/s mem 4.261 MiB) * non-zero drops is due to duplicated keys in generated keys. (3) Improve update performance for qp-trie Now qp-trie is divided into 256 sub-trees by using the first byte as the key and one sub-tree is protected one spinlock. I also had tried the hierarchical lock by using hand-over-hand lock scheme [1], but it doesn't scale well. Update performance of hash-table has been optimized a lot by "bpf: BPF specific memory allocator", but because qp-trie allocates memory for a variable-length key and bpf_mem_alloc() only support <=4KB memory, so it doesn't fill well for qp-trie. Strings in /proc/kallsyms (key size=83, max entries=170958) htab update (1 thread) 3.135 ± 0.355M/s (drops 0.000 ± 0.000M/s mem 30.842 MiB) htab update (2 thread) 6.269 ± 0.686M/s (drops 0.000 ± 0.000M/s mem 30.841 MiB) htab update (4 thread) 11.607 ± 1.632M/s (drops 0.000 ± 0.000M/s mem 30.840 MiB) htab update (8 thread) 23.041 ± 0.806M/s (drops 0.000 ± 0.000M/s mem 30.842 MiB) htab update (16 thread) 31.393 ± 0.307M/s (drops 0.000 ± 0.000M/s mem 30.835 MiB) qp-trie update (1 thread) 1.191 ± 0.093M/s (drops 0.000 ± 0.000M/s mem 17.170 MiB) qp-trie update (2 thread) 2.057 ± 0.041M/s (drops 0.000 ± 0.000M/s mem 17.058 MiB) qp-trie update (4 thread) 2.975 ± 0.035M/s (drops 0.000 ± 0.000M/s mem 17.411 MiB) qp-trie update (8 thread) 3.596 ± 0.031M/s (drops 0.000 ± 0.000M/s mem 17.110 MiB) qp-trie update (16 thread) 4.200 ± 0.048M/s (drops 0.000 ± 0.000M/s mem 17.228 MiB) (4) Improve memory efficiency further When using strings in BTF string section as a data set for qp-trie, the slab memory usage showed in cgroup memory.stats file is about 11MB for qp-trie and 22MB for hash table as shown below. However the theoretical memory usage for qp-trie is ~6.8MB (is ~4.9MB if removing "parent" & "rcu" fields from qp_trie_branch) and the extra memory usage (about 38% of total usage) mainly comes from internal fragment in slab (namely 2^n alignment for allocation) and overhead in kmem-cgroup accounting. We can reduce the internal fragment by creating separated kmem_cache for qp_trie_branch with different child nodes, but not sure whether it is worthy or not. And in order to prevent allocating a rcu_head for each leaf node, now only branch node is RCU-freed, so when replacing a leaf node, a new branch node and a new leaf node will be allocated instead of replacing the old leaf node and RCU-freed the old leaf node. Also it seems rcu_head can be replaced by a pointer as BPF specific memory allocator does, so 8-bytes can be saved for each node, and again not sure it's worth it. Sorted strings in BTF string sections (entries=115980) htab lookup (1 thread) 6.915 ± 0.029M/s (drops 0.000 ± 0.000M/s mem 22.216 MiB) qp-trie lookup (1 thread) 6.791 ± 0.005M/s (drops 0.000 ± 0.000M/s mem 11.273 MiB) All files under linux kernel source directory (entries=74359) htab lookup (1 thread) 7.978 ± 0.009M/s (drops 0.000 ± 0.000M/s mem 14.272 MiB) qp-trie lookup (1 thread) 5.521 ± 0.003M/s (drops 0.000 ± 0.000M/s mem 9.367 MiB) Domain names for Alexa top million web site (entries=1000000) htab lookup (1 thread) 3.148 ± 0.039M/s (drops 0.000 ± 0.000M/s mem 190.831 MiB) qp-trie lookup (1 thread) 2.374 ± 0.026M/s (drops 0.000 ± 0.000M/s mem 83.733 MiB) Comments and suggestions are always welcome. Change Log: v1: * Use bpf_dynptr as map key type instead of bpf_lpm_trie_key-styled key (Suggested by Andrii) * Fix build error and RCU related sparse errors reported by lkp robot * Copy the passed key firstly in qp_trie_update_elem(), because the content of passed key may change and may break the assumption in two-round lookup process during update. * Add the missing rcu_barrier in qp_trie_free() RFC: https://lore.kernel.org/bpf/20220726130005.3102470-1-houtao1@xxxxxxxxxx/ [0]: https://lore.kernel.org/bpf/CAEf4Bzb7keBS8vXgV5JZzwgNGgMV0X3_guQ_m9JW3X6fJBDpPQ@xxxxxxxxxxxxxx/ [1]: https://lore.kernel.org/bpf/db34696a-cbfe-16e8-6dd5-8174b97dcf1d@xxxxxxxxxx/ Hou Tao (10): bpf: Export bpf_dynptr_{get|set}_size bpf: Add helper btf_type_is_bpf_dynptr() bpf: Support bpf_dynptr-typed map key for bpf syscall bpf: Support bpf_dynptr-typed map key for bpf program libbpf: Add helpers for bpf_dynptr_user bpf: Add support for qp-trie map selftests/bpf: Add two new dynptr_fail cases for map key selftests/bpf: Add test case for basic qp-trie map operations selftests/bpf: Add benchmark for qp-trie map selftests/bpf: Add tests for qp-trie map by using bpf syscalls include/linux/bpf.h | 4 + include/linux/bpf_types.h | 1 + include/linux/btf.h | 1 + include/uapi/linux/bpf.h | 10 + kernel/bpf/Makefile | 1 + kernel/bpf/bpf_qp_trie.c | 1055 +++++++++++++++ kernel/bpf/btf.c | 7 + kernel/bpf/helpers.c | 9 +- kernel/bpf/syscall.c | 108 +- kernel/bpf/verifier.c | 18 +- tools/include/uapi/linux/bpf.h | 9 + tools/lib/bpf/bpf.h | 19 + tools/testing/selftests/bpf/DENYLIST.s390x | 1 + tools/testing/selftests/bpf/Makefile | 5 +- tools/testing/selftests/bpf/bench.c | 10 + .../selftests/bpf/benchs/bench_qp_trie.c | 511 ++++++++ .../selftests/bpf/benchs/run_bench_qp_trie.sh | 55 + .../selftests/bpf/map_tests/qp_trie_map.c | 1165 +++++++++++++++++ .../testing/selftests/bpf/prog_tests/dynptr.c | 2 + .../selftests/bpf/prog_tests/qp_trie_test.c | 91 ++ .../testing/selftests/bpf/progs/dynptr_fail.c | 43 + .../selftests/bpf/progs/qp_trie_bench.c | 236 ++++ .../selftests/bpf/progs/qp_trie_test.c | 110 ++ 23 files changed, 3452 insertions(+), 19 deletions(-) create mode 100644 kernel/bpf/bpf_qp_trie.c create mode 100644 tools/testing/selftests/bpf/benchs/bench_qp_trie.c create mode 100755 tools/testing/selftests/bpf/benchs/run_bench_qp_trie.sh create mode 100644 tools/testing/selftests/bpf/map_tests/qp_trie_map.c create mode 100644 tools/testing/selftests/bpf/prog_tests/qp_trie_test.c create mode 100644 tools/testing/selftests/bpf/progs/qp_trie_bench.c create mode 100644 tools/testing/selftests/bpf/progs/qp_trie_test.c -- 2.29.2
