From 81ed18ab3098b6519274545e80a29caacb77d160 Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Wed, 15 Mar 2017 18:26:42 -0700 Subject: bpf: add helper inlining infra and optimize map_array lookup Optimize bpf_call -> bpf_map_lookup_elem() -> array_map_lookup_elem() into a sequence of bpf instructions. When JIT is on the sequence of bpf instructions is the sequence of native cpu instructions with significantly faster performance than indirect call and two function's prologue/epilogue. Signed-off-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: David S. Miller --- kernel/bpf/arraymap.c | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) (limited to 'kernel/bpf/arraymap.c') diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c index 6b6f41f0b211..bcf9955fac95 100644 --- a/kernel/bpf/arraymap.c +++ b/kernel/bpf/arraymap.c @@ -1,4 +1,5 @@ /* Copyright (c) 2011-2014 PLUMgrid, http://plumgrid.com + * Copyright (c) 2016,2017 Facebook * * This program is free software; you can redistribute it and/or * modify it under the terms of version 2 of the GNU General Public @@ -113,6 +114,33 @@ static void *array_map_lookup_elem(struct bpf_map *map, void *key) return array->value + array->elem_size * index; } +/* emit BPF instructions equivalent to C code of array_map_lookup_elem() */ +static u32 array_map_gen_lookup(struct bpf_map *map, struct bpf_insn *insn_buf) +{ + struct bpf_array *array = container_of(map, struct bpf_array, map); + struct bpf_insn *insn = insn_buf; + u32 elem_size = array->elem_size; + const int ret = BPF_REG_0; + const int map_ptr = BPF_REG_1; + const int index = BPF_REG_2; + + *insn++ = BPF_ALU64_IMM(BPF_ADD, map_ptr, offsetof(struct bpf_array, value)); + *insn++ = BPF_LDX_MEM(BPF_W, ret, index, 0); + *insn++ = BPF_JMP_IMM(BPF_JGE, ret, array->map.max_entries, + elem_size == 1 ? 2 : 3); + if (elem_size == 1) { + /* nop */ + } else if (is_power_of_2(elem_size)) { + *insn++ = BPF_ALU64_IMM(BPF_LSH, ret, ilog2(elem_size)); + } else { + *insn++ = BPF_ALU64_IMM(BPF_MUL, ret, elem_size); + } + *insn++ = BPF_ALU64_REG(BPF_ADD, ret, map_ptr); + *insn++ = BPF_JMP_IMM(BPF_JA, 0, 0, 1); + *insn++ = BPF_MOV64_IMM(ret, 0); + return insn - insn_buf; +} + /* Called from eBPF program */ static void *percpu_array_map_lookup_elem(struct bpf_map *map, void *key) { @@ -267,6 +295,7 @@ static const struct bpf_map_ops array_ops = { .map_lookup_elem = array_map_lookup_elem, .map_update_elem = array_map_update_elem, .map_delete_elem = array_map_delete_elem, + .map_gen_lookup = array_map_gen_lookup, }; static struct bpf_map_type_list array_type __ro_after_init = { -- cgit From fad73a1a35ea61f13607a391aca669caad8c04ca Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 22 Mar 2017 10:00:32 -0700 Subject: bpf: Fix and simplifications on inline map lookup Fix in verifier: For the same bpf_map_lookup_elem() instruction (i.e. "call 1"), a broken case is "a different type of map could be used for the same lookup instruction". For example, an array in one case and a hashmap in another. We have to resort to the old dynamic call behavior in this case. The fix is to check for collision on insn_aux->map_ptr. If there is collision, don't inline the map lookup. Please see the "do_reg_lookup()" in test_map_in_map_kern.c in the later patch for how-to trigger the above case. Simplifications on array_map_gen_lookup(): 1. Calculate elem_size from map->value_size. It removes the need for 'struct bpf_array' which makes the later map-in-map implementation easier. 2. Remove the 'elem_size == 1' test Fixes: 81ed18ab3098 ("bpf: add helper inlining infra and optimize map_array lookup") Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: David S. Miller --- kernel/bpf/arraymap.c | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-) (limited to 'kernel/bpf/arraymap.c') diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c index bcf9955fac95..4d7d5d0ed76a 100644 --- a/kernel/bpf/arraymap.c +++ b/kernel/bpf/arraymap.c @@ -117,20 +117,17 @@ static void *array_map_lookup_elem(struct bpf_map *map, void *key) /* emit BPF instructions equivalent to C code of array_map_lookup_elem() */ static u32 array_map_gen_lookup(struct bpf_map *map, struct bpf_insn *insn_buf) { - struct bpf_array *array = container_of(map, struct bpf_array, map); struct bpf_insn *insn = insn_buf; - u32 elem_size = array->elem_size; + u32 elem_size = round_up(map->value_size, 8); const int ret = BPF_REG_0; const int map_ptr = BPF_REG_1; const int index = BPF_REG_2; *insn++ = BPF_ALU64_IMM(BPF_ADD, map_ptr, offsetof(struct bpf_array, value)); *insn++ = BPF_LDX_MEM(BPF_W, ret, index, 0); - *insn++ = BPF_JMP_IMM(BPF_JGE, ret, array->map.max_entries, - elem_size == 1 ? 2 : 3); - if (elem_size == 1) { - /* nop */ - } else if (is_power_of_2(elem_size)) { + *insn++ = BPF_JMP_IMM(BPF_JGE, ret, map->max_entries, 3); + + if (is_power_of_2(elem_size)) { *insn++ = BPF_ALU64_IMM(BPF_LSH, ret, ilog2(elem_size)); } else { *insn++ = BPF_ALU64_IMM(BPF_MUL, ret, elem_size); -- cgit From 56f668dfe00dcf086734f1c42ea999398fad6572 Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 22 Mar 2017 10:00:33 -0700 Subject: bpf: Add array of maps support This patch adds a few helper funcs to enable map-in-map support (i.e. outer_map->inner_map). The first outer_map type BPF_MAP_TYPE_ARRAY_OF_MAPS is also added in this patch. The next patch will introduce a hash of maps type. Any bpf map type can be acted as an inner_map. The exception is BPF_MAP_TYPE_PROG_ARRAY because the extra level of indirection makes it harder to verify the owner_prog_type and owner_jited. Multi-level map-in-map is not supported (i.e. map->map is ok but not map->map->map). When adding an inner_map to an outer_map, it currently checks the map_type, key_size, value_size, map_flags, max_entries and ops. The verifier also uses those map's properties to do static analysis. map_flags is needed because we need to ensure BPF_PROG_TYPE_PERF_EVENT is using a preallocated hashtab for the inner_hash also. ops and max_entries are needed to generate inlined map-lookup instructions. For simplicity reason, a simple '==' test is used for both map_flags and max_entries. The equality of ops is implied by the equality of map_type. During outer_map creation time, an inner_map_fd is needed to create an outer_map. However, the inner_map_fd's life time does not depend on the outer_map. The inner_map_fd is merely used to initialize the inner_map_meta of the outer_map. Also, for the outer_map: * It allows element update and delete from syscall * It allows element lookup from bpf_prog The above is similar to the current fd_array pattern. Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: David S. Miller --- kernel/bpf/arraymap.c | 63 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 63 insertions(+) (limited to 'kernel/bpf/arraymap.c') diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c index 4d7d5d0ed76a..bc9da93db403 100644 --- a/kernel/bpf/arraymap.c +++ b/kernel/bpf/arraymap.c @@ -17,6 +17,8 @@ #include #include +#include "map_in_map.h" + static void bpf_array_free_percpu(struct bpf_array *array) { int i; @@ -602,3 +604,64 @@ static int __init register_cgroup_array_map(void) } late_initcall(register_cgroup_array_map); #endif + +static struct bpf_map *array_of_map_alloc(union bpf_attr *attr) +{ + struct bpf_map *map, *inner_map_meta; + + inner_map_meta = bpf_map_meta_alloc(attr->inner_map_fd); + if (IS_ERR(inner_map_meta)) + return inner_map_meta; + + map = fd_array_map_alloc(attr); + if (IS_ERR(map)) { + bpf_map_meta_free(inner_map_meta); + return map; + } + + map->inner_map_meta = inner_map_meta; + + return map; +} + +static void array_of_map_free(struct bpf_map *map) +{ + /* map->inner_map_meta is only accessed by syscall which + * is protected by fdget/fdput. + */ + bpf_map_meta_free(map->inner_map_meta); + bpf_fd_array_map_clear(map); + fd_array_map_free(map); +} + +static void *array_of_map_lookup_elem(struct bpf_map *map, void *key) +{ + struct bpf_map **inner_map = array_map_lookup_elem(map, key); + + if (!inner_map) + return NULL; + + return READ_ONCE(*inner_map); +} + +static const struct bpf_map_ops array_of_map_ops = { + .map_alloc = array_of_map_alloc, + .map_free = array_of_map_free, + .map_get_next_key = array_map_get_next_key, + .map_lookup_elem = array_of_map_lookup_elem, + .map_delete_elem = fd_array_map_delete_elem, + .map_fd_get_ptr = bpf_map_fd_get_ptr, + .map_fd_put_ptr = bpf_map_fd_put_ptr, +}; + +static struct bpf_map_type_list array_of_map_type __ro_after_init = { + .ops = &array_of_map_ops, + .type = BPF_MAP_TYPE_ARRAY_OF_MAPS, +}; + +static int __init register_array_of_map(void) +{ + bpf_register_map_type(&array_of_map_type); + return 0; +} +late_initcall(register_array_of_map); -- cgit From 40077e0cf62206ac3c315b6991d8dcddb3703286 Mon Sep 17 00:00:00 2001 From: Johannes Berg Date: Tue, 11 Apr 2017 15:34:58 +0200 Subject: bpf: remove struct bpf_map_type_list There's no need to have struct bpf_map_type_list since it just contains a list_head, the type, and the ops pointer. Since the types are densely packed and not actually dynamically registered, it's much easier and smaller to have an array of type->ops pointer. Also initialize this array statically to remove code needed to initialize it. In order to save duplicating the list, move it to the types header file added by the previous patch and include it in the same fashion. Signed-off-by: Johannes Berg Acked-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: David S. Miller --- kernel/bpf/arraymap.c | 78 ++++----------------------------------------------- 1 file changed, 6 insertions(+), 72 deletions(-) (limited to 'kernel/bpf/arraymap.c') diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c index bc9da93db403..ec621df5a97a 100644 --- a/kernel/bpf/arraymap.c +++ b/kernel/bpf/arraymap.c @@ -287,7 +287,7 @@ static void array_map_free(struct bpf_map *map) bpf_map_area_free(array); } -static const struct bpf_map_ops array_ops = { +const struct bpf_map_ops array_map_ops = { .map_alloc = array_map_alloc, .map_free = array_map_free, .map_get_next_key = array_map_get_next_key, @@ -297,12 +297,7 @@ static const struct bpf_map_ops array_ops = { .map_gen_lookup = array_map_gen_lookup, }; -static struct bpf_map_type_list array_type __ro_after_init = { - .ops = &array_ops, - .type = BPF_MAP_TYPE_ARRAY, -}; - -static const struct bpf_map_ops percpu_array_ops = { +const struct bpf_map_ops percpu_array_map_ops = { .map_alloc = array_map_alloc, .map_free = array_map_free, .map_get_next_key = array_map_get_next_key, @@ -311,19 +306,6 @@ static const struct bpf_map_ops percpu_array_ops = { .map_delete_elem = array_map_delete_elem, }; -static struct bpf_map_type_list percpu_array_type __ro_after_init = { - .ops = &percpu_array_ops, - .type = BPF_MAP_TYPE_PERCPU_ARRAY, -}; - -static int __init register_array_map(void) -{ - bpf_register_map_type(&array_type); - bpf_register_map_type(&percpu_array_type); - return 0; -} -late_initcall(register_array_map); - static struct bpf_map *fd_array_map_alloc(union bpf_attr *attr) { /* only file descriptors can be stored in this type of map */ @@ -427,7 +409,7 @@ void bpf_fd_array_map_clear(struct bpf_map *map) fd_array_map_delete_elem(map, &i); } -static const struct bpf_map_ops prog_array_ops = { +const struct bpf_map_ops prog_array_map_ops = { .map_alloc = fd_array_map_alloc, .map_free = fd_array_map_free, .map_get_next_key = array_map_get_next_key, @@ -437,18 +419,6 @@ static const struct bpf_map_ops prog_array_ops = { .map_fd_put_ptr = prog_fd_array_put_ptr, }; -static struct bpf_map_type_list prog_array_type __ro_after_init = { - .ops = &prog_array_ops, - .type = BPF_MAP_TYPE_PROG_ARRAY, -}; - -static int __init register_prog_array_map(void) -{ - bpf_register_map_type(&prog_array_type); - return 0; -} -late_initcall(register_prog_array_map); - static struct bpf_event_entry *bpf_event_entry_gen(struct file *perf_file, struct file *map_file) { @@ -539,7 +509,7 @@ static void perf_event_fd_array_release(struct bpf_map *map, rcu_read_unlock(); } -static const struct bpf_map_ops perf_event_array_ops = { +const struct bpf_map_ops perf_event_array_map_ops = { .map_alloc = fd_array_map_alloc, .map_free = fd_array_map_free, .map_get_next_key = array_map_get_next_key, @@ -550,18 +520,6 @@ static const struct bpf_map_ops perf_event_array_ops = { .map_release = perf_event_fd_array_release, }; -static struct bpf_map_type_list perf_event_array_type __ro_after_init = { - .ops = &perf_event_array_ops, - .type = BPF_MAP_TYPE_PERF_EVENT_ARRAY, -}; - -static int __init register_perf_event_array_map(void) -{ - bpf_register_map_type(&perf_event_array_type); - return 0; -} -late_initcall(register_perf_event_array_map); - #ifdef CONFIG_CGROUPS static void *cgroup_fd_array_get_ptr(struct bpf_map *map, struct file *map_file /* not used */, @@ -582,7 +540,7 @@ static void cgroup_fd_array_free(struct bpf_map *map) fd_array_map_free(map); } -static const struct bpf_map_ops cgroup_array_ops = { +const struct bpf_map_ops cgroup_array_map_ops = { .map_alloc = fd_array_map_alloc, .map_free = cgroup_fd_array_free, .map_get_next_key = array_map_get_next_key, @@ -591,18 +549,6 @@ static const struct bpf_map_ops cgroup_array_ops = { .map_fd_get_ptr = cgroup_fd_array_get_ptr, .map_fd_put_ptr = cgroup_fd_array_put_ptr, }; - -static struct bpf_map_type_list cgroup_array_type __ro_after_init = { - .ops = &cgroup_array_ops, - .type = BPF_MAP_TYPE_CGROUP_ARRAY, -}; - -static int __init register_cgroup_array_map(void) -{ - bpf_register_map_type(&cgroup_array_type); - return 0; -} -late_initcall(register_cgroup_array_map); #endif static struct bpf_map *array_of_map_alloc(union bpf_attr *attr) @@ -644,7 +590,7 @@ static void *array_of_map_lookup_elem(struct bpf_map *map, void *key) return READ_ONCE(*inner_map); } -static const struct bpf_map_ops array_of_map_ops = { +const struct bpf_map_ops array_of_maps_map_ops = { .map_alloc = array_of_map_alloc, .map_free = array_of_map_free, .map_get_next_key = array_map_get_next_key, @@ -653,15 +599,3 @@ static const struct bpf_map_ops array_of_map_ops = { .map_fd_get_ptr = bpf_map_fd_get_ptr, .map_fd_put_ptr = bpf_map_fd_put_ptr, }; - -static struct bpf_map_type_list array_of_map_type __ro_after_init = { - .ops = &array_of_map_ops, - .type = BPF_MAP_TYPE_ARRAY_OF_MAPS, -}; - -static int __init register_array_of_map(void) -{ - bpf_register_map_type(&array_of_map_type); - return 0; -} -late_initcall(register_array_of_map); -- cgit From 8fe45924387be6b5c1be59a7eb330790c61d5d10 Mon Sep 17 00:00:00 2001 From: Teng Qin Date: Mon, 24 Apr 2017 19:00:37 -0700 Subject: bpf: map_get_next_key to return first key on NULL When iterating through a map, we need to find a key that does not exist in the map so map_get_next_key will give us the first key of the map. This often requires a lot of guessing in production systems. This patch makes map_get_next_key return the first key when the key pointer in the parameter is NULL. Signed-off-by: Teng Qin Signed-off-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: David S. Miller --- kernel/bpf/arraymap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel/bpf/arraymap.c') diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c index ec621df5a97a..5e00b2333c26 100644 --- a/kernel/bpf/arraymap.c +++ b/kernel/bpf/arraymap.c @@ -182,7 +182,7 @@ int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value) static int array_map_get_next_key(struct bpf_map *map, void *key, void *next_key) { struct bpf_array *array = container_of(map, struct bpf_array, map); - u32 index = *(u32 *)key; + u32 index = key ? *(u32 *)key : U32_MAX; u32 *next = (u32 *)next_key; if (index >= array->map.max_entries) { -- cgit