[PATCH bpf-next 0/5] bpf: Misc improvements

[PATCH bpf-next 0/5] bpf: Misc improvements

[Alexei Starovoitov]

From: Alexei Starovoitov <ast@kernel.org>

Several improvements
- optimize prog stats
- compute stats for sleepable progs
- prevent recursion fentry/fexit progs

Alexei Starovoitov (5):
  bpf: Optimize program stats
  bpf: Compute program stats for sleepable programs
  bpf: Add per-program recursion prevention mechanism
  selftest/bpf: Add recursion test
  bpf: Count the number of times recursion was prevented

 arch/x86/net/bpf_jit_comp.c                   | 46 +++++++++------
 include/linux/bpf.h                           | 16 ++---
 include/linux/filter.h                        | 12 +++-
 include/uapi/linux/bpf.h                      |  1 +
 kernel/bpf/core.c                             | 16 +++--
 kernel/bpf/syscall.c                          | 16 +++--
 kernel/bpf/trampoline.c                       | 59 ++++++++++++++++---
 kernel/bpf/verifier.c                         |  2 +-
 tools/bpf/bpftool/prog.c                      |  5 ++
 tools/include/uapi/linux/bpf.h                |  1 +
 .../selftests/bpf/prog_tests/fexit_stress.c   |  2 +-
 .../selftests/bpf/prog_tests/recursion.c      | 33 +++++++++++
 .../bpf/prog_tests/trampoline_count.c         |  4 +-
 tools/testing/selftests/bpf/progs/recursion.c | 46 +++++++++++++++
 14 files changed, 205 insertions(+), 54 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/recursion.c
 create mode 100644 tools/testing/selftests/bpf/progs/recursion.c

-- 
2.24.1

[PATCH bpf-next 1/5] bpf: Optimize program stats

[Alexei Starovoitov]

From: Alexei Starovoitov <ast@kernel.org>

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 <ast@kernel.org>
---
 include/linux/bpf.h     |  8 --------
 include/linux/filter.h  | 10 +++++++++-
 kernel/bpf/core.c       |  8 ++++----
 kernel/bpf/syscall.c    |  2 +-
 kernel/bpf/trampoline.c |  2 +-
 kernel/bpf/verifier.c   |  2 +-
 6 files changed, 16 insertions(+), 16 deletions(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 321966fc35db..026fa8873c5d 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -14,7 +14,6 @@
 #include <linux/numa.h>
 #include <linux/mm_types.h>
 #include <linux/wait.h>
-#include <linux/u64_stats_sync.h>
 #include <linux/refcount.h>
 #include <linux/mutex.h>
 #include <linux/module.h>
@@ -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..c6592590a0b7 100644
--- a/include/linux/filter.h
+++ b/include/linux/filter.h
@@ -22,6 +22,7 @@
 #include <linux/vmalloc.h>
 #include <linux/sockptr.h>
 #include <crypto/sha1.h>
+#include <linux/u64_stats_sync.h>
 
 #include <net/sch_generic.h>
 
@@ -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? */
@@ -559,6 +566,7 @@ struct bpf_prog {
 	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);
 	/* Instructions for interpreter */
@@ -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..fa3da4cda476 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;
@@ -249,7 +249,7 @@ 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);
+		free_percpu(fp->stats);
 		kfree(fp->aux->poke_tab);
 		kfree(fp->aux);
 	}
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 15694246f854..4189edb41b73 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -10889,7 +10889,7 @@ 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])
-- 
2.24.1

[PATCH bpf-next 2/5] bpf: Compute program stats for sleepable programs

[Alexei Starovoitov]

From: Alexei Starovoitov <ast@kernel.org>

In older non-RT kernels migrate_disable() was the same as preempt_disable().
Since commit 74d862b682f5 ("sched: Make migrate_disable/enable() independent of RT")
migrate_disable() is real and doesn't prevent sleeping.
Use it to efficiently compute execution stats for sleepable bpf programs.
migrate_disable() will also be used to enable per-cpu maps in sleepable programs
in the future patches.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
---
 arch/x86/net/bpf_jit_comp.c | 31 ++++++++++++-------------------
 include/linux/bpf.h         |  4 ++--
 kernel/bpf/trampoline.c     | 27 +++++++++++++++++++++------
 3 files changed, 35 insertions(+), 27 deletions(-)

diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c
index a3dc3bd154ac..d11b9bcebbea 100644
--- a/arch/x86/net/bpf_jit_comp.c
+++ b/arch/x86/net/bpf_jit_comp.c
@@ -1742,15 +1742,12 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog,
 	u8 *prog = *pprog;
 	int cnt = 0;
 
-	if (p->aux->sleepable) {
-		if (emit_call(&prog, __bpf_prog_enter_sleepable, prog))
+	if (emit_call(&prog,
+		      p->aux->sleepable ? __bpf_prog_enter_sleepable :
+		      __bpf_prog_enter, prog))
 			return -EINVAL;
-	} else {
-		if (emit_call(&prog, __bpf_prog_enter, prog))
-			return -EINVAL;
-		/* remember prog start time returned by __bpf_prog_enter */
-		emit_mov_reg(&prog, true, BPF_REG_6, BPF_REG_0);
-	}
+	/* remember prog start time returned by __bpf_prog_enter */
+	emit_mov_reg(&prog, true, BPF_REG_6, BPF_REG_0);
 
 	/* arg1: lea rdi, [rbp - stack_size] */
 	EMIT4(0x48, 0x8D, 0x7D, -stack_size);
@@ -1770,18 +1767,14 @@ 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);
 
-	if (p->aux->sleepable) {
-		if (emit_call(&prog, __bpf_prog_exit_sleepable, prog))
+	/* 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 */
+	emit_mov_reg(&prog, true, BPF_REG_2, BPF_REG_6);
+	if (emit_call(&prog,
+		      p->aux->sleepable ? __bpf_prog_exit_sleepable :
+		      __bpf_prog_exit, prog))
 			return -EINVAL;
-	} else {
-		/* 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 */
-		emit_mov_reg(&prog, true, BPF_REG_2, BPF_REG_6);
-		if (emit_call(&prog, __bpf_prog_exit, prog))
-			return -EINVAL;
-	}
 
 	*pprog = prog;
 	return 0;
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 026fa8873c5d..2fa48439ef31 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -563,8 +563,8 @@ int arch_prepare_bpf_trampoline(void *image, void *image_end,
 /* these two functions are called from generated trampoline */
 u64 notrace __bpf_prog_enter(void);
 void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start);
-void notrace __bpf_prog_enter_sleepable(void);
-void notrace __bpf_prog_exit_sleepable(void);
+u64 notrace __bpf_prog_enter_sleepable(void);
+void notrace __bpf_prog_exit_sleepable(struct bpf_prog *prog, u64 start);
 
 struct bpf_ksym {
 	unsigned long		 start;
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index 5be3beeedd74..b1f567514b7e 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -388,10 +388,11 @@ void bpf_trampoline_put(struct bpf_trampoline *tr)
  * call prog->bpf_func
  * call __bpf_prog_exit
  */
+#define NO_START_TIME 0
 u64 notrace __bpf_prog_enter(void)
 	__acquires(RCU)
 {
-	u64 start = 0;
+	u64 start = NO_START_TIME;
 
 	rcu_read_lock();
 	migrate_disable();
@@ -400,8 +401,8 @@ u64 notrace __bpf_prog_enter(void)
 	return start;
 }
 
-void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start)
-	__releases(RCU)
+static void notrace update_prog_stats(struct bpf_prog *prog,
+				      u64 start)
 {
 	struct bpf_prog_stats *stats;
 
@@ -411,25 +412,39 @@ void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start)
 	     * And vice versa.
 	     * Hence check that 'start' is not zero.
 	     */
-	    start) {
+	    start > NO_START_TIME) {
 		stats = this_cpu_ptr(prog->stats);
 		u64_stats_update_begin(&stats->syncp);
 		stats->cnt++;
 		stats->nsecs += sched_clock() - start;
 		u64_stats_update_end(&stats->syncp);
 	}
+}
+
+void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start)
+	__releases(RCU)
+{
+	update_prog_stats(prog, start);
 	migrate_enable();
 	rcu_read_unlock();
 }
 
-void notrace __bpf_prog_enter_sleepable(void)
+u64 notrace __bpf_prog_enter_sleepable(void)
 {
+	u64 start = NO_START_TIME;
+
 	rcu_read_lock_trace();
+	migrate_disable();
 	might_fault();
+	if (static_branch_unlikely(&bpf_stats_enabled_key))
+		start = sched_clock();
+	return start;
 }
 
-void notrace __bpf_prog_exit_sleepable(void)
+void notrace __bpf_prog_exit_sleepable(struct bpf_prog *prog, u64 start)
 {
+	update_prog_stats(prog, start);
+	migrate_enable();
 	rcu_read_unlock_trace();
 }
 
-- 
2.24.1

[PATCH bpf-next 3/5] bpf: Add per-program recursion prevention mechanism

[Alexei Starovoitov]

From: Alexei Starovoitov <ast@kernel.org>

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 <ast@kernel.org>
---
 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                       | 22 ++++++++++++++-----
 .../selftests/bpf/prog_tests/fexit_stress.c   |  2 +-
 .../bpf/prog_tests/trampoline_count.c         |  4 ++--
 7 files changed, 47 insertions(+), 11 deletions(-)

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 2fa48439ef31..6f019b06a2fd 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 c6592590a0b7..9927e14ce021 100644
--- a/include/linux/filter.h
+++ b/include/linux/filter.h
@@ -567,6 +567,7 @@ struct bpf_prog {
 	struct bpf_prog_aux	*aux;		/* Auxiliary fields */
 	struct sock_fprog_kern	*orig_prog;	/* Original BPF program */
 	struct bpf_prog_stats __percpu *stats;
+	int __percpu 		*active;
 	unsigned int		(*bpf_func)(const void *ctx,
 					    const struct bpf_insn *insn);
 	/* Instructions for interpreter */
diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
index fa3da4cda476..f4560dbe7f31 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;
@@ -250,6 +257,7 @@ void __bpf_prog_free(struct bpf_prog *fp)
 		mutex_destroy(&fp->aux->used_maps_mutex);
 		mutex_destroy(&fp->aux->dst_mutex);
 		free_percpu(fp->stats);
+		free_percpu(fp->active);
 		kfree(fp->aux->poke_tab);
 		kfree(fp->aux);
 	}
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index b1f567514b7e..226f613ab289 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -388,16 +388,21 @@ void bpf_trampoline_put(struct bpf_trampoline *tr)
  * call prog->bpf_func
  * call __bpf_prog_exit
  */
-#define NO_START_TIME 0
-u64 notrace __bpf_prog_enter(void)
+#define NO_START_TIME 1
+u64 notrace __bpf_prog_enter(struct bpf_prog *prog)
 	__acquires(RCU)
 {
 	u64 start = NO_START_TIME;
 
 	rcu_read_lock();
 	migrate_disable();
-	if (static_branch_unlikely(&bpf_stats_enabled_key))
+	if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1))
+		return 0;
+	if (static_branch_unlikely(&bpf_stats_enabled_key)) {
 		start = sched_clock();
+		if (unlikely(!start))
+			start = NO_START_TIME;
+	}
 	return start;
 }
 
@@ -425,25 +430,32 @@ 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)
 {
 	u64 start = NO_START_TIME;
 
 	rcu_read_lock_trace();
 	migrate_disable();
 	might_fault();
-	if (static_branch_unlikely(&bpf_stats_enabled_key))
+	if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1))
+		return 0;
+	if (static_branch_unlikely(&bpf_stats_enabled_key)) {
 		start = sched_clock();
+		if (unlikely(!start))
+			start = NO_START_TIME;
+	}
 	return start;
 }
 
 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..4698b0d2de36 100644
--- a/tools/testing/selftests/bpf/prog_tests/fexit_stress.c
+++ b/tools/testing/selftests/bpf/prog_tests/fexit_stress.c
@@ -3,7 +3,7 @@
 #include <test_progs.h>
 
 /* x86-64 fits 55 JITed and 43 interpreted progs into half page */
-#define CNT 40
+#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 <sys/prctl.h>
 #include <test_progs.h>
 
-#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))) {
-- 
2.24.1

[PATCH bpf-next 4/5] selftest/bpf: Add recursion test

[Alexei Starovoitov]

From: Alexei Starovoitov <ast@kernel.org>

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
---
 .../selftests/bpf/prog_tests/recursion.c      | 33 +++++++++++++
 tools/testing/selftests/bpf/progs/recursion.c | 46 +++++++++++++++++++
 2 files changed, 79 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/recursion.c
 create mode 100644 tools/testing/selftests/bpf/progs/recursion.c

diff --git a/tools/testing/selftests/bpf/prog_tests/recursion.c b/tools/testing/selftests/bpf/prog_tests/recursion.c
new file mode 100644
index 000000000000..16e8eab5a29d
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/recursion.c
@@ -0,0 +1,33 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2021 Facebook */
+#include <test_progs.h>
+#include "recursion.skel.h"
+
+void test_recursion(void)
+{
+	struct recursion *skel;
+	int key = 0;
+	int err;
+
+	skel = recursion__open_and_load();
+	if (!ASSERT_OK_PTR(skel, "skel_open_and_load"))
+		return;
+
+	err = recursion__attach(skel);
+	if (!ASSERT_OK(err, "skel_attach"))
+		goto out;
+
+	ASSERT_EQ(skel->bss->pass1, 0, "pass1 == 0");
+	bpf_map_lookup_elem(bpf_map__fd(skel->maps.hash1), &key, 0);
+	ASSERT_EQ(skel->bss->pass1, 1, "pass1 == 1");
+	bpf_map_lookup_elem(bpf_map__fd(skel->maps.hash1), &key, 0);
+	ASSERT_EQ(skel->bss->pass1, 2, "pass1 == 2");
+	
+	ASSERT_EQ(skel->bss->pass2, 0, "pass2 == 0");
+	bpf_map_lookup_elem(bpf_map__fd(skel->maps.hash2), &key, 0);
+	ASSERT_EQ(skel->bss->pass2, 1, "pass2 == 1");
+	bpf_map_lookup_elem(bpf_map__fd(skel->maps.hash2), &key, 0);
+	ASSERT_EQ(skel->bss->pass2, 2, "pass2 == 2");
+out:
+	recursion__destroy(skel);
+}
diff --git a/tools/testing/selftests/bpf/progs/recursion.c b/tools/testing/selftests/bpf/progs/recursion.c
new file mode 100644
index 000000000000..49f679375b9d
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/recursion.c
@@ -0,0 +1,46 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2021 Facebook */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+struct {
+	__uint(type, BPF_MAP_TYPE_HASH);
+	__uint(max_entries, 1);
+	__type(key, int);
+	__type(value, long);
+} hash1 SEC(".maps");
+
+struct {
+	__uint(type, BPF_MAP_TYPE_HASH);
+	__uint(max_entries, 1);
+	__type(key, int);
+	__type(value, long);
+} hash2 SEC(".maps");
+
+int pass1 = 0;
+int pass2 = 0;
+
+SEC("fentry/__htab_map_lookup_elem")
+int BPF_PROG(on_lookup, struct bpf_map *map)
+{
+	int key = 0;
+
+	if (map == (void *)&hash1) {
+		pass1++;
+		return 0;
+	}
+	if (map == (void *)&hash2) {
+		pass2++;
+		/* htab_map_gen_lookup() will inline below call
+		 * into direct call to __htab_map_lookup_elem()
+		 */
+		bpf_map_lookup_elem(&hash2, &key);
+		return 0;
+	}
+
+	return 0;
+}
-- 
2.24.1

[PATCH bpf-next 5/5] bpf: Count the number of times recursion was prevented

[Alexei Starovoitov]

From: Alexei Starovoitov <ast@kernel.org>

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 <ast@kernel.org>
---
 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       |  5 +++++
 tools/include/uapi/linux/bpf.h |  1 +
 6 files changed, 34 insertions(+), 6 deletions(-)

diff --git a/include/linux/filter.h b/include/linux/filter.h
index 9927e14ce021..707f86560a43 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..a5b8af378c33 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 += misses;
 	}
 	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 226f613ab289..83b77883bd77 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -381,6 +381,16 @@ void bpf_trampoline_put(struct bpf_trampoline *tr)
 	mutex_unlock(&trampoline_mutex);
 }
 
+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
@@ -396,8 +406,10 @@ 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;
+	}
 	if (static_branch_unlikely(&bpf_stats_enabled_key)) {
 		start = sched_clock();
 		if (unlikely(!start))
@@ -442,8 +454,10 @@ 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;
+	}
 	if (static_branch_unlikely(&bpf_stats_enabled_key)) {
 		start = sched_clock();
 		if (unlikely(!start))
diff --git a/tools/bpf/bpftool/prog.c b/tools/bpf/bpftool/prog.c
index 1fe3ba255bad..2e1cd12589c5 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 +448,9 @@ 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 {
-- 
2.24.1

Re: [PATCH bpf-next 5/5] bpf: Count the number of times recursion was prevented

[kernel test robot]

[-- Attachment #1: Type: text/plain, Size: 2395 bytes --]

Hi Alexei,

I love your patch! Perhaps something to improve:

[auto build test WARNING on bpf-next/master]

url:    https://github.com/0day-ci/linux/commits/Alexei-Starovoitov/bpf-Misc-improvements/20210206-150251
base:   https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git master
config: openrisc-randconfig-p001-20210206 (attached as .config)
compiler: or1k-linux-gcc (GCC) 9.3.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/8d71f0650d2551940d9fc39a723d6595ab604715
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Alexei-Starovoitov/bpf-Misc-improvements/20210206-150251
        git checkout 8d71f0650d2551940d9fc39a723d6595ab604715
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=openrisc 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All warnings (new ones prefixed by >>):

   kernel/bpf/syscall.c: In function 'bpf_prog_get_stats':
>> kernel/bpf/syscall.c:1740:21: warning: variable 'tmisses' set but not used [-Wunused-but-set-variable]
    1740 |   u64 tnsecs, tcnt, tmisses;
         |                     ^~~~~~~


vim +/tmisses +1740 kernel/bpf/syscall.c

  1730	
  1731	static void bpf_prog_get_stats(const struct bpf_prog *prog,
  1732				       struct bpf_prog_stats *stats)
  1733	{
  1734		u64 nsecs = 0, cnt = 0, misses = 0;
  1735		int cpu;
  1736	
  1737		for_each_possible_cpu(cpu) {
  1738			const struct bpf_prog_stats *st;
  1739			unsigned int start;
> 1740			u64 tnsecs, tcnt, tmisses;
  1741	
  1742			st = per_cpu_ptr(prog->stats, cpu);
  1743			do {
  1744				start = u64_stats_fetch_begin_irq(&st->syncp);
  1745				tnsecs = st->nsecs;
  1746				tcnt = st->cnt;
  1747				tmisses = st->misses;
  1748			} while (u64_stats_fetch_retry_irq(&st->syncp, start));
  1749			nsecs += tnsecs;
  1750			cnt += tcnt;
  1751			misses += misses;
  1752		}
  1753		stats->nsecs = nsecs;
  1754		stats->cnt = cnt;
  1755		stats->misses = misses;
  1756	}
  1757	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 21845 bytes --]