From 20f2505fb436cfa674cf1f46aaa624f44d3d1d03 Mon Sep 17 00:00:00 2001 From: Stanislav Fomichev Date: Fri, 15 Jan 2021 08:35:00 -0800 Subject: bpf: Try to avoid kzalloc in cgroup/{s,g}etsockopt When we attach a bpf program to cgroup/getsockopt any other getsockopt() syscall starts incurring kzalloc/kfree cost. Let add a small buffer on the stack and use it for small (majority) {s,g}etsockopt values. The buffer is small enough to fit into the cache line and cover the majority of simple options (most of them are 4 byte ints). It seems natural to do the same for setsockopt, but it's a bit more involved when the BPF program modifies the data (where we have to kmalloc). The assumption is that for the majority of setsockopt calls (which are doing pure BPF options or apply policy) this will bring some benefit as well. Without this patch (we remove about 1% __kmalloc): 3.38% 0.07% tcp_mmap [kernel.kallsyms] [k] __cgroup_bpf_run_filter_getsockopt | --3.30%--__cgroup_bpf_run_filter_getsockopt | --0.81%--__kmalloc Signed-off-by: Stanislav Fomichev Signed-off-by: Alexei Starovoitov Acked-by: Martin KaFai Lau Link: https://lore.kernel.org/bpf/20210115163501.805133-3-sdf@google.com --- include/linux/filter.h | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'include/linux/filter.h') diff --git a/include/linux/filter.h b/include/linux/filter.h index 7fdce5407214..5b3137d7b690 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -1298,6 +1298,11 @@ struct bpf_sysctl_kern { u64 tmp_reg; }; +#define BPF_SOCKOPT_KERN_BUF_SIZE 32 +struct bpf_sockopt_buf { + u8 data[BPF_SOCKOPT_KERN_BUF_SIZE]; +}; + struct bpf_sockopt_kern { struct sock *sk; u8 *optval; -- cgit v1.2.3 From 700d4796ef59f5faf240d307839bd419e2b6bdff Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Tue, 9 Feb 2021 19:36:26 -0800 Subject: bpf: Optimize program stats Move bpf_prog_stats from prog->aux into prog to avoid one extra load in critical path of program execution. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20210210033634.62081-2-alexei.starovoitov@gmail.com --- include/linux/bpf.h | 8 -------- include/linux/filter.h | 14 +++++++++++--- kernel/bpf/core.c | 8 ++++---- kernel/bpf/syscall.c | 2 +- kernel/bpf/trampoline.c | 2 +- kernel/bpf/verifier.c | 2 +- 6 files changed, 18 insertions(+), 18 deletions(-) (limited to 'include/linux/filter.h') diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 079162bbd387..5a388955e6b6 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -14,7 +14,6 @@ #include #include #include -#include #include #include #include @@ -507,12 +506,6 @@ enum bpf_cgroup_storage_type { */ #define MAX_BPF_FUNC_ARGS 12 -struct bpf_prog_stats { - u64 cnt; - u64 nsecs; - struct u64_stats_sync syncp; -} __aligned(2 * sizeof(u64)); - struct btf_func_model { u8 ret_size; u8 nr_args; @@ -845,7 +838,6 @@ struct bpf_prog_aux { u32 linfo_idx; u32 num_exentries; struct exception_table_entry *extable; - struct bpf_prog_stats __percpu *stats; union { struct work_struct work; struct rcu_head rcu; diff --git a/include/linux/filter.h b/include/linux/filter.h index 5b3137d7b690..cecb03c9d251 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -22,6 +22,7 @@ #include #include #include +#include #include @@ -539,6 +540,12 @@ struct bpf_binary_header { u8 image[] __aligned(BPF_IMAGE_ALIGNMENT); }; +struct bpf_prog_stats { + u64 cnt; + u64 nsecs; + struct u64_stats_sync syncp; +} __aligned(2 * sizeof(u64)); + struct bpf_prog { u16 pages; /* Number of allocated pages */ u16 jited:1, /* Is our filter JIT'ed? */ @@ -557,10 +564,11 @@ struct bpf_prog { u32 len; /* Number of filter blocks */ u32 jited_len; /* Size of jited insns in bytes */ u8 tag[BPF_TAG_SIZE]; - struct bpf_prog_aux *aux; /* Auxiliary fields */ - struct sock_fprog_kern *orig_prog; /* Original BPF program */ + struct bpf_prog_stats __percpu *stats; unsigned int (*bpf_func)(const void *ctx, const struct bpf_insn *insn); + struct bpf_prog_aux *aux; /* Auxiliary fields */ + struct sock_fprog_kern *orig_prog; /* Original BPF program */ /* Instructions for interpreter */ struct sock_filter insns[0]; struct bpf_insn insnsi[]; @@ -581,7 +589,7 @@ DECLARE_STATIC_KEY_FALSE(bpf_stats_enabled_key); struct bpf_prog_stats *__stats; \ u64 __start = sched_clock(); \ __ret = dfunc(ctx, (prog)->insnsi, (prog)->bpf_func); \ - __stats = this_cpu_ptr(prog->aux->stats); \ + __stats = this_cpu_ptr(prog->stats); \ u64_stats_update_begin(&__stats->syncp); \ __stats->cnt++; \ __stats->nsecs += sched_clock() - __start; \ diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index 5bbd4884ff7a..2cf71fd39c22 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -114,8 +114,8 @@ struct bpf_prog *bpf_prog_alloc(unsigned int size, gfp_t gfp_extra_flags) if (!prog) return NULL; - prog->aux->stats = alloc_percpu_gfp(struct bpf_prog_stats, gfp_flags); - if (!prog->aux->stats) { + prog->stats = alloc_percpu_gfp(struct bpf_prog_stats, gfp_flags); + if (!prog->stats) { kfree(prog->aux); vfree(prog); return NULL; @@ -124,7 +124,7 @@ struct bpf_prog *bpf_prog_alloc(unsigned int size, gfp_t gfp_extra_flags) for_each_possible_cpu(cpu) { struct bpf_prog_stats *pstats; - pstats = per_cpu_ptr(prog->aux->stats, cpu); + pstats = per_cpu_ptr(prog->stats, cpu); u64_stats_init(&pstats->syncp); } return prog; @@ -249,10 +249,10 @@ void __bpf_prog_free(struct bpf_prog *fp) if (fp->aux) { mutex_destroy(&fp->aux->used_maps_mutex); mutex_destroy(&fp->aux->dst_mutex); - free_percpu(fp->aux->stats); kfree(fp->aux->poke_tab); kfree(fp->aux); } + free_percpu(fp->stats); vfree(fp); } diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index e5999d86c76e..f7df56a704de 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -1739,7 +1739,7 @@ static void bpf_prog_get_stats(const struct bpf_prog *prog, unsigned int start; u64 tnsecs, tcnt; - st = per_cpu_ptr(prog->aux->stats, cpu); + st = per_cpu_ptr(prog->stats, cpu); do { start = u64_stats_fetch_begin_irq(&st->syncp); tnsecs = st->nsecs; diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c index 35c5887d82ff..5be3beeedd74 100644 --- a/kernel/bpf/trampoline.c +++ b/kernel/bpf/trampoline.c @@ -412,7 +412,7 @@ void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start) * Hence check that 'start' is not zero. */ start) { - stats = this_cpu_ptr(prog->aux->stats); + stats = this_cpu_ptr(prog->stats); u64_stats_update_begin(&stats->syncp); stats->cnt++; stats->nsecs += sched_clock() - start; diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 400d79e99fc8..424c1ba0f52f 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -11253,7 +11253,7 @@ static int jit_subprogs(struct bpf_verifier_env *env) /* BPF_PROG_RUN doesn't call subprogs directly, * hence main prog stats include the runtime of subprogs. * subprogs don't have IDs and not reachable via prog_get_next_id - * func[i]->aux->stats will never be accessed and stays NULL + * func[i]->stats will never be accessed and stays NULL */ func[i] = bpf_prog_alloc_no_stats(bpf_prog_size(len), GFP_USER); if (!func[i]) -- cgit v1.2.3 From ca06f55b90020cd97f4cc6d52db95436162e7dcf Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Tue, 9 Feb 2021 19:36:29 -0800 Subject: bpf: Add per-program recursion prevention mechanism Since both sleepable and non-sleepable programs execute under migrate_disable add recursion prevention mechanism to both types of programs when they're executed via bpf trampoline. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20210210033634.62081-5-alexei.starovoitov@gmail.com --- arch/x86/net/bpf_jit_comp.c | 15 ++++++++++++++ include/linux/bpf.h | 6 +++--- include/linux/filter.h | 1 + kernel/bpf/core.c | 8 ++++++++ kernel/bpf/trampoline.c | 23 ++++++++++++++++++---- .../selftests/bpf/prog_tests/fexit_stress.c | 4 ++-- .../selftests/bpf/prog_tests/trampoline_count.c | 4 ++-- 7 files changed, 50 insertions(+), 11 deletions(-) (limited to 'include/linux/filter.h') diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c index d11b9bcebbea..79e7a0ec1da5 100644 --- a/arch/x86/net/bpf_jit_comp.c +++ b/arch/x86/net/bpf_jit_comp.c @@ -1740,8 +1740,11 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog, struct bpf_prog *p, int stack_size, bool mod_ret) { u8 *prog = *pprog; + u8 *jmp_insn; int cnt = 0; + /* arg1: mov rdi, progs[i] */ + emit_mov_imm64(&prog, BPF_REG_1, (long) p >> 32, (u32) (long) p); if (emit_call(&prog, p->aux->sleepable ? __bpf_prog_enter_sleepable : __bpf_prog_enter, prog)) @@ -1749,6 +1752,14 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog, /* remember prog start time returned by __bpf_prog_enter */ emit_mov_reg(&prog, true, BPF_REG_6, BPF_REG_0); + /* if (__bpf_prog_enter*(prog) == 0) + * goto skip_exec_of_prog; + */ + EMIT3(0x48, 0x85, 0xC0); /* test rax,rax */ + /* emit 2 nops that will be replaced with JE insn */ + jmp_insn = prog; + emit_nops(&prog, 2); + /* arg1: lea rdi, [rbp - stack_size] */ EMIT4(0x48, 0x8D, 0x7D, -stack_size); /* arg2: progs[i]->insnsi for interpreter */ @@ -1767,6 +1778,10 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog, if (mod_ret) emit_stx(&prog, BPF_DW, BPF_REG_FP, BPF_REG_0, -8); + /* replace 2 nops with JE insn, since jmp target is known */ + jmp_insn[0] = X86_JE; + jmp_insn[1] = prog - jmp_insn - 2; + /* arg1: mov rdi, progs[i] */ emit_mov_imm64(&prog, BPF_REG_1, (long) p >> 32, (u32) (long) p); /* arg2: mov rsi, rbx <- start time in nsec */ diff --git a/include/linux/bpf.h b/include/linux/bpf.h index e1840ceaf55f..1c8ea682c007 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -529,7 +529,7 @@ struct btf_func_model { /* Each call __bpf_prog_enter + call bpf_func + call __bpf_prog_exit is ~50 * bytes on x86. Pick a number to fit into BPF_IMAGE_SIZE / 2 */ -#define BPF_MAX_TRAMP_PROGS 40 +#define BPF_MAX_TRAMP_PROGS 38 struct bpf_tramp_progs { struct bpf_prog *progs[BPF_MAX_TRAMP_PROGS]; @@ -561,9 +561,9 @@ int arch_prepare_bpf_trampoline(void *image, void *image_end, struct bpf_tramp_progs *tprogs, void *orig_call); /* these two functions are called from generated trampoline */ -u64 notrace __bpf_prog_enter(void); +u64 notrace __bpf_prog_enter(struct bpf_prog *prog); void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start); -u64 notrace __bpf_prog_enter_sleepable(void); +u64 notrace __bpf_prog_enter_sleepable(struct bpf_prog *prog); void notrace __bpf_prog_exit_sleepable(struct bpf_prog *prog, u64 start); struct bpf_ksym { diff --git a/include/linux/filter.h b/include/linux/filter.h index cecb03c9d251..6a06f3c69f4e 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -565,6 +565,7 @@ struct bpf_prog { u32 jited_len; /* Size of jited insns in bytes */ u8 tag[BPF_TAG_SIZE]; struct bpf_prog_stats __percpu *stats; + int __percpu *active; unsigned int (*bpf_func)(const void *ctx, const struct bpf_insn *insn); struct bpf_prog_aux *aux; /* Auxiliary fields */ diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index 2cf71fd39c22..334070c4b8a1 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -91,6 +91,12 @@ struct bpf_prog *bpf_prog_alloc_no_stats(unsigned int size, gfp_t gfp_extra_flag vfree(fp); return NULL; } + fp->active = alloc_percpu_gfp(int, GFP_KERNEL_ACCOUNT | gfp_extra_flags); + if (!fp->active) { + vfree(fp); + kfree(aux); + return NULL; + } fp->pages = size / PAGE_SIZE; fp->aux = aux; @@ -116,6 +122,7 @@ struct bpf_prog *bpf_prog_alloc(unsigned int size, gfp_t gfp_extra_flags) prog->stats = alloc_percpu_gfp(struct bpf_prog_stats, gfp_flags); if (!prog->stats) { + free_percpu(prog->active); kfree(prog->aux); vfree(prog); return NULL; @@ -253,6 +260,7 @@ void __bpf_prog_free(struct bpf_prog *fp) kfree(fp->aux); } free_percpu(fp->stats); + free_percpu(fp->active); vfree(fp); } diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c index 48eb021e1421..89ef6320d19b 100644 --- a/kernel/bpf/trampoline.c +++ b/kernel/bpf/trampoline.c @@ -381,13 +381,16 @@ out: mutex_unlock(&trampoline_mutex); } -#define NO_START_TIME 0 +#define NO_START_TIME 1 static u64 notrace bpf_prog_start_time(void) { u64 start = NO_START_TIME; - if (static_branch_unlikely(&bpf_stats_enabled_key)) + if (static_branch_unlikely(&bpf_stats_enabled_key)) { start = sched_clock(); + if (unlikely(!start)) + start = NO_START_TIME; + } return start; } @@ -397,12 +400,20 @@ static u64 notrace bpf_prog_start_time(void) * call __bpf_prog_enter * call prog->bpf_func * call __bpf_prog_exit + * + * __bpf_prog_enter returns: + * 0 - skip execution of the bpf prog + * 1 - execute bpf prog + * [2..MAX_U64] - excute bpf prog and record execution time. + * This is start time. */ -u64 notrace __bpf_prog_enter(void) +u64 notrace __bpf_prog_enter(struct bpf_prog *prog) __acquires(RCU) { rcu_read_lock(); migrate_disable(); + if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) + return 0; return bpf_prog_start_time(); } @@ -430,21 +441,25 @@ void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start) __releases(RCU) { update_prog_stats(prog, start); + __this_cpu_dec(*(prog->active)); migrate_enable(); rcu_read_unlock(); } -u64 notrace __bpf_prog_enter_sleepable(void) +u64 notrace __bpf_prog_enter_sleepable(struct bpf_prog *prog) { rcu_read_lock_trace(); migrate_disable(); might_fault(); + if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) + return 0; return bpf_prog_start_time(); } void notrace __bpf_prog_exit_sleepable(struct bpf_prog *prog, u64 start) { update_prog_stats(prog, start); + __this_cpu_dec(*(prog->active)); migrate_enable(); rcu_read_unlock_trace(); } diff --git a/tools/testing/selftests/bpf/prog_tests/fexit_stress.c b/tools/testing/selftests/bpf/prog_tests/fexit_stress.c index 3b9dbf7433f0..7c9b62e971f1 100644 --- a/tools/testing/selftests/bpf/prog_tests/fexit_stress.c +++ b/tools/testing/selftests/bpf/prog_tests/fexit_stress.c @@ -2,8 +2,8 @@ /* Copyright (c) 2019 Facebook */ #include -/* x86-64 fits 55 JITed and 43 interpreted progs into half page */ -#define CNT 40 +/* that's kernel internal BPF_MAX_TRAMP_PROGS define */ +#define CNT 38 void test_fexit_stress(void) { diff --git a/tools/testing/selftests/bpf/prog_tests/trampoline_count.c b/tools/testing/selftests/bpf/prog_tests/trampoline_count.c index 781c8d11604b..f3022d934e2d 100644 --- a/tools/testing/selftests/bpf/prog_tests/trampoline_count.c +++ b/tools/testing/selftests/bpf/prog_tests/trampoline_count.c @@ -4,7 +4,7 @@ #include #include -#define MAX_TRAMP_PROGS 40 +#define MAX_TRAMP_PROGS 38 struct inst { struct bpf_object *obj; @@ -52,7 +52,7 @@ void test_trampoline_count(void) struct bpf_link *link; char comm[16] = {}; - /* attach 'allowed' 40 trampoline programs */ + /* attach 'allowed' trampoline programs */ for (i = 0; i < MAX_TRAMP_PROGS; i++) { obj = bpf_object__open_file(object, NULL); if (CHECK(IS_ERR(obj), "obj_open_file", "err %ld\n", PTR_ERR(obj))) { -- cgit v1.2.3 From 9ed9e9ba2337205311398a312796c213737bac35 Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Tue, 9 Feb 2021 19:36:31 -0800 Subject: bpf: Count the number of times recursion was prevented Add per-program counter for number of times recursion prevention mechanism was triggered and expose it via show_fdinfo and bpf_prog_info. Teach bpftool to print it. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20210210033634.62081-7-alexei.starovoitov@gmail.com --- include/linux/filter.h | 1 + include/uapi/linux/bpf.h | 1 + kernel/bpf/syscall.c | 14 ++++++++++---- kernel/bpf/trampoline.c | 18 ++++++++++++++++-- tools/bpf/bpftool/prog.c | 4 ++++ tools/include/uapi/linux/bpf.h | 1 + 6 files changed, 33 insertions(+), 6 deletions(-) (limited to 'include/linux/filter.h') diff --git a/include/linux/filter.h b/include/linux/filter.h index 6a06f3c69f4e..3b00fc906ccd 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -543,6 +543,7 @@ struct bpf_binary_header { struct bpf_prog_stats { u64 cnt; u64 nsecs; + u64 misses; struct u64_stats_sync syncp; } __aligned(2 * sizeof(u64)); diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index c001766adcbc..c547ad1ffe43 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -4501,6 +4501,7 @@ struct bpf_prog_info { __aligned_u64 prog_tags; __u64 run_time_ns; __u64 run_cnt; + __u64 recursion_misses; } __attribute__((aligned(8))); struct bpf_map_info { diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index f7df56a704de..c859bc46d06c 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -1731,25 +1731,28 @@ static int bpf_prog_release(struct inode *inode, struct file *filp) static void bpf_prog_get_stats(const struct bpf_prog *prog, struct bpf_prog_stats *stats) { - u64 nsecs = 0, cnt = 0; + u64 nsecs = 0, cnt = 0, misses = 0; int cpu; for_each_possible_cpu(cpu) { const struct bpf_prog_stats *st; unsigned int start; - u64 tnsecs, tcnt; + u64 tnsecs, tcnt, tmisses; st = per_cpu_ptr(prog->stats, cpu); do { start = u64_stats_fetch_begin_irq(&st->syncp); tnsecs = st->nsecs; tcnt = st->cnt; + tmisses = st->misses; } while (u64_stats_fetch_retry_irq(&st->syncp, start)); nsecs += tnsecs; cnt += tcnt; + misses += tmisses; } stats->nsecs = nsecs; stats->cnt = cnt; + stats->misses = misses; } #ifdef CONFIG_PROC_FS @@ -1768,14 +1771,16 @@ static void bpf_prog_show_fdinfo(struct seq_file *m, struct file *filp) "memlock:\t%llu\n" "prog_id:\t%u\n" "run_time_ns:\t%llu\n" - "run_cnt:\t%llu\n", + "run_cnt:\t%llu\n" + "recursion_misses:\t%llu\n", prog->type, prog->jited, prog_tag, prog->pages * 1ULL << PAGE_SHIFT, prog->aux->id, stats.nsecs, - stats.cnt); + stats.cnt, + stats.misses); } #endif @@ -3438,6 +3443,7 @@ static int bpf_prog_get_info_by_fd(struct file *file, bpf_prog_get_stats(prog, &stats); info.run_time_ns = stats.nsecs; info.run_cnt = stats.cnt; + info.recursion_misses = stats.misses; if (!bpf_capable()) { info.jited_prog_len = 0; diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c index 89ef6320d19b..7bc3b3209224 100644 --- a/kernel/bpf/trampoline.c +++ b/kernel/bpf/trampoline.c @@ -394,6 +394,16 @@ static u64 notrace bpf_prog_start_time(void) return start; } +static void notrace inc_misses_counter(struct bpf_prog *prog) +{ + struct bpf_prog_stats *stats; + + stats = this_cpu_ptr(prog->stats); + u64_stats_update_begin(&stats->syncp); + stats->misses++; + u64_stats_update_end(&stats->syncp); +} + /* The logic is similar to BPF_PROG_RUN, but with an explicit * rcu_read_lock() and migrate_disable() which are required * for the trampoline. The macro is split into @@ -412,8 +422,10 @@ u64 notrace __bpf_prog_enter(struct bpf_prog *prog) { rcu_read_lock(); migrate_disable(); - if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) + if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) { + inc_misses_counter(prog); return 0; + } return bpf_prog_start_time(); } @@ -451,8 +463,10 @@ u64 notrace __bpf_prog_enter_sleepable(struct bpf_prog *prog) rcu_read_lock_trace(); migrate_disable(); might_fault(); - if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) + if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) { + inc_misses_counter(prog); return 0; + } return bpf_prog_start_time(); } diff --git a/tools/bpf/bpftool/prog.c b/tools/bpf/bpftool/prog.c index 1fe3ba255bad..f2b915b20546 100644 --- a/tools/bpf/bpftool/prog.c +++ b/tools/bpf/bpftool/prog.c @@ -368,6 +368,8 @@ static void print_prog_header_json(struct bpf_prog_info *info) jsonw_uint_field(json_wtr, "run_time_ns", info->run_time_ns); jsonw_uint_field(json_wtr, "run_cnt", info->run_cnt); } + if (info->recursion_misses) + jsonw_uint_field(json_wtr, "recursion_misses", info->recursion_misses); } static void print_prog_json(struct bpf_prog_info *info, int fd) @@ -446,6 +448,8 @@ static void print_prog_header_plain(struct bpf_prog_info *info) if (info->run_time_ns) printf(" run_time_ns %lld run_cnt %lld", info->run_time_ns, info->run_cnt); + if (info->recursion_misses) + printf(" recursion_misses %lld", info->recursion_misses); printf("\n"); } diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index c001766adcbc..c547ad1ffe43 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -4501,6 +4501,7 @@ struct bpf_prog_info { __aligned_u64 prog_tags; __u64 run_time_ns; __u64 run_cnt; + __u64 recursion_misses; } __attribute__((aligned(8))); struct bpf_map_info { -- cgit v1.2.3