Merge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next

Daniel Borkmann says:

====================
pull-request: bpf-next 2017-12-18

The following pull-request contains BPF updates for your *net-next* tree.

The main changes are:

1) Allow arbitrary function calls from one BPF function to another BPF function.
   As of today when writing BPF programs, __always_inline had to be used in
   the BPF C programs for all functions, unnecessarily causing LLVM to inflate
   code size. Handle this more naturally with support for BPF to BPF calls
   such that this __always_inline restriction can be overcome. As a result,
   it allows for better optimized code and finally enables to introduce core
   BPF libraries in the future that can be reused out of different projects.
   x86 and arm64 JIT support was added as well, from Alexei.

2) Add infrastructure for tagging functions as error injectable and allow for
   BPF to return arbitrary error values when BPF is attached via kprobes on
   those. This way of injecting errors generically eases testing and debugging
   without having to recompile or restart the kernel. Tags for opting-in for
   this facility are added with BPF_ALLOW_ERROR_INJECTION(), from Josef.

3) For BPF offload via nfp JIT, add support for bpf_xdp_adjust_head() helper
   call for XDP programs. First part of this work adds handling of BPF
   capabilities included in the firmware, and the later patches add support
   to the nfp verifier part and JIT as well as some small optimizations,
   from Jakub.

4) The bpftool now also gets support for basic cgroup BPF operations such
   as attaching, detaching and listing current BPF programs. As a requirement
   for the attach part, bpftool can now also load object files through
   'bpftool prog load'. This reuses libbpf which we have in the kernel tree
   as well. bpftool-cgroup man page is added along with it, from Roman.

5) Back then commit e87c6bc3852b ("bpf: permit multiple bpf attachments for
   a single perf event") added support for attaching multiple BPF programs
   to a single perf event. Given they are configured through perf's ioctl()
   interface, the interface has been extended with a PERF_EVENT_IOC_QUERY_BPF
   command in this work in order to return an array of one or multiple BPF
   prog ids that are currently attached, from Yonghong.

6) Various minor fixes and cleanups to the bpftool's Makefile as well
   as a new 'uninstall' and 'doc-uninstall' target for removing bpftool
   itself or prior installed documentation related to it, from Quentin.

7) Add CONFIG_CGROUP_BPF=y to the BPF kernel selftest config file which is
   required for the test_dev_cgroup test case to run, from Naresh.

8) Fix reporting of XDP prog_flags for nfp driver, from Jakub.

9) Fix libbpf's exit code from the Makefile when libelf was not found in
   the system, also from Jakub.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

11 files changed, 1380 insertions, 205 deletions

diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
index 86b50aa26ee8..768e0a02d8c8 100644
--- a/kernel/bpf/core.c
+++ b/kernel/bpf/core.c
@@ -94,6 +94,7 @@ struct bpf_prog *bpf_prog_alloc(unsigned int size, gfp_t gfp_extra_flags)
 	fp->pages = size / PAGE_SIZE;
 	fp->aux = aux;
 	fp->aux->prog = fp;
+	fp->jit_requested = ebpf_jit_enabled();
 
 	INIT_LIST_HEAD_RCU(&fp->aux->ksym_lnode);
 
@@ -217,30 +218,40 @@ int bpf_prog_calc_tag(struct bpf_prog *fp)
 	return 0;
 }
 
-static bool bpf_is_jmp_and_has_target(const struct bpf_insn *insn)
-{
-	return BPF_CLASS(insn->code) == BPF_JMP  &&
-	       /* Call and Exit are both special jumps with no
-		* target inside the BPF instruction image.
-		*/
-	       BPF_OP(insn->code) != BPF_CALL &&
-	       BPF_OP(insn->code) != BPF_EXIT;
-}
-
 static void bpf_adj_branches(struct bpf_prog *prog, u32 pos, u32 delta)
 {
 	struct bpf_insn *insn = prog->insnsi;
 	u32 i, insn_cnt = prog->len;
+	bool pseudo_call;
+	u8 code;
+	int off;
 
 	for (i = 0; i < insn_cnt; i++, insn++) {
-		if (!bpf_is_jmp_and_has_target(insn))
+		code = insn->code;
+		if (BPF_CLASS(code) != BPF_JMP)
+			continue;
+		if (BPF_OP(code) == BPF_EXIT)
 			continue;
+		if (BPF_OP(code) == BPF_CALL) {
+			if (insn->src_reg == BPF_PSEUDO_CALL)
+				pseudo_call = true;
+			else
+				continue;
+		} else {
+			pseudo_call = false;
+		}
+		off = pseudo_call ? insn->imm : insn->off;
 
 		/* Adjust offset of jmps if we cross boundaries. */
-		if (i < pos && i + insn->off + 1 > pos)
-			insn->off += delta;
-		else if (i > pos + delta && i + insn->off + 1 <= pos + delta)
-			insn->off -= delta;
+		if (i < pos && i + off + 1 > pos)
+			off += delta;
+		else if (i > pos + delta && i + off + 1 <= pos + delta)
+			off -= delta;
+
+		if (pseudo_call)
+			insn->imm = off;
+		else
+			insn->off = off;
 	}
 }
 
@@ -711,7 +722,7 @@ struct bpf_prog *bpf_jit_blind_constants(struct bpf_prog *prog)
 	struct bpf_insn *insn;
 	int i, rewritten;
 
-	if (!bpf_jit_blinding_enabled())
+	if (!bpf_jit_blinding_enabled(prog) || prog->blinded)
 		return prog;
 
 	clone = bpf_prog_clone_create(prog, GFP_USER);
@@ -753,6 +764,7 @@ struct bpf_prog *bpf_jit_blind_constants(struct bpf_prog *prog)
 		i        += insn_delta;
 	}
 
+	clone->blinded = 1;
 	return clone;
 }
 #endif /* CONFIG_BPF_JIT */
@@ -774,8 +786,7 @@ EXPORT_SYMBOL_GPL(__bpf_call_base);
  *
  * Decode and execute eBPF instructions.
  */
-static unsigned int ___bpf_prog_run(u64 *regs, const struct bpf_insn *insn,
-				    u64 *stack)
+static u64 ___bpf_prog_run(u64 *regs, const struct bpf_insn *insn, u64 *stack)
 {
 	u64 tmp;
 	static const void *jumptable[256] = {
@@ -835,6 +846,7 @@ static unsigned int ___bpf_prog_run(u64 *regs, const struct bpf_insn *insn,
 		[BPF_ALU64 | BPF_NEG] = &&ALU64_NEG,
 		/* Call instruction */
 		[BPF_JMP | BPF_CALL] = &&JMP_CALL,
+		[BPF_JMP | BPF_CALL_ARGS] = &&JMP_CALL_ARGS,
 		[BPF_JMP | BPF_TAIL_CALL] = &&JMP_TAIL_CALL,
 		/* Jumps */
 		[BPF_JMP | BPF_JA] = &&JMP_JA,
@@ -1025,6 +1037,13 @@ select_insn:
 						       BPF_R4, BPF_R5);
 		CONT;
 
+	JMP_CALL_ARGS:
+		BPF_R0 = (__bpf_call_base_args + insn->imm)(BPF_R1, BPF_R2,
+							    BPF_R3, BPF_R4,
+							    BPF_R5,
+							    insn + insn->off + 1);
+		CONT;
+
 	JMP_TAIL_CALL: {
 		struct bpf_map *map = (struct bpf_map *) (unsigned long) BPF_R2;
 		struct bpf_array *array = container_of(map, struct bpf_array, map);
@@ -1297,6 +1316,23 @@ static unsigned int PROG_NAME(stack_size)(const void *ctx, const struct bpf_insn
 	return ___bpf_prog_run(regs, insn, stack); \
 }
 
+#define PROG_NAME_ARGS(stack_size) __bpf_prog_run_args##stack_size
+#define DEFINE_BPF_PROG_RUN_ARGS(stack_size) \
+static u64 PROG_NAME_ARGS(stack_size)(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5, \
+				      const struct bpf_insn *insn) \
+{ \
+	u64 stack[stack_size / sizeof(u64)]; \
+	u64 regs[MAX_BPF_REG]; \
+\
+	FP = (u64) (unsigned long) &stack[ARRAY_SIZE(stack)]; \
+	BPF_R1 = r1; \
+	BPF_R2 = r2; \
+	BPF_R3 = r3; \
+	BPF_R4 = r4; \
+	BPF_R5 = r5; \
+	return ___bpf_prog_run(regs, insn, stack); \
+}
+
 #define EVAL1(FN, X) FN(X)
 #define EVAL2(FN, X, Y...) FN(X) EVAL1(FN, Y)
 #define EVAL3(FN, X, Y...) FN(X) EVAL2(FN, Y)
@@ -1308,6 +1344,10 @@ EVAL6(DEFINE_BPF_PROG_RUN, 32, 64, 96, 128, 160, 192);
 EVAL6(DEFINE_BPF_PROG_RUN, 224, 256, 288, 320, 352, 384);
 EVAL4(DEFINE_BPF_PROG_RUN, 416, 448, 480, 512);
 
+EVAL6(DEFINE_BPF_PROG_RUN_ARGS, 32, 64, 96, 128, 160, 192);
+EVAL6(DEFINE_BPF_PROG_RUN_ARGS, 224, 256, 288, 320, 352, 384);
+EVAL4(DEFINE_BPF_PROG_RUN_ARGS, 416, 448, 480, 512);
+
 #define PROG_NAME_LIST(stack_size) PROG_NAME(stack_size),
 
 static unsigned int (*interpreters[])(const void *ctx,
@@ -1316,10 +1356,31 @@ EVAL6(PROG_NAME_LIST, 32, 64, 96, 128, 160, 192)
 EVAL6(PROG_NAME_LIST, 224, 256, 288, 320, 352, 384)
 EVAL4(PROG_NAME_LIST, 416, 448, 480, 512)
 };
+#undef PROG_NAME_LIST
+#define PROG_NAME_LIST(stack_size) PROG_NAME_ARGS(stack_size),
+static u64 (*interpreters_args[])(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5,
+				  const struct bpf_insn *insn) = {
+EVAL6(PROG_NAME_LIST, 32, 64, 96, 128, 160, 192)
+EVAL6(PROG_NAME_LIST, 224, 256, 288, 320, 352, 384)
+EVAL4(PROG_NAME_LIST, 416, 448, 480, 512)
+};
+#undef PROG_NAME_LIST
+
+void bpf_patch_call_args(struct bpf_insn *insn, u32 stack_depth)
+{
+	stack_depth = max_t(u32, stack_depth, 1);
+	insn->off = (s16) insn->imm;
+	insn->imm = interpreters_args[(round_up(stack_depth, 32) / 32) - 1] -
+		__bpf_call_base_args;
+	insn->code = BPF_JMP | BPF_CALL_ARGS;
+}
 
 bool bpf_prog_array_compatible(struct bpf_array *array,
 			       const struct bpf_prog *fp)
 {
+	if (fp->kprobe_override)
+		return false;
+
 	if (!array->owner_prog_type) {
 		/* There's no owner yet where we could check for
 		 * compatibility.
@@ -1462,6 +1523,8 @@ int bpf_prog_array_copy_to_user(struct bpf_prog_array __rcu *progs,
 	rcu_read_lock();
 	prog = rcu_dereference(progs)->progs;
 	for (; *prog; prog++) {
+		if (*prog == &dummy_bpf_prog.prog)
+			continue;
 		id = (*prog)->aux->id;
 		if (copy_to_user(prog_ids + i, &id, sizeof(id))) {
 			rcu_read_unlock();
@@ -1545,14 +1608,41 @@ int bpf_prog_array_copy(struct bpf_prog_array __rcu *old_array,
 	return 0;
 }
 
+int bpf_prog_array_copy_info(struct bpf_prog_array __rcu *array,
+			     __u32 __user *prog_ids, u32 request_cnt,
+			     __u32 __user *prog_cnt)
+{
+	u32 cnt = 0;
+
+	if (array)
+		cnt = bpf_prog_array_length(array);
+
+	if (copy_to_user(prog_cnt, &cnt, sizeof(cnt)))
+		return -EFAULT;
+
+	/* return early if user requested only program count or nothing to copy */
+	if (!request_cnt || !cnt)
+		return 0;
+
+	return bpf_prog_array_copy_to_user(array, prog_ids, request_cnt);
+}
+
 static void bpf_prog_free_deferred(struct work_struct *work)
 {
 	struct bpf_prog_aux *aux;
+	int i;
 
 	aux = container_of(work, struct bpf_prog_aux, work);
 	if (bpf_prog_is_dev_bound(aux))
 		bpf_prog_offload_destroy(aux->prog);
-	bpf_jit_free(aux->prog);
+	for (i = 0; i < aux->func_cnt; i++)
+		bpf_jit_free(aux->func[i]);
+	if (aux->func_cnt) {
+		kfree(aux->func);
+		bpf_prog_unlock_free(aux->prog);
+	} else {
+		bpf_jit_free(aux->prog);
+	}
 }
 
 /* Free internal BPF program */
diff --git a/kernel/bpf/disasm.c b/kernel/bpf/disasm.c
index e682850c9715..883f88fa5bfc 100644
--- a/kernel/bpf/disasm.c
+++ b/kernel/bpf/disasm.c
@@ -189,8 +189,12 @@ void print_bpf_insn(bpf_insn_print_cb verbose, struct bpf_verifier_env *env,
 		u8 opcode = BPF_OP(insn->code);
 
 		if (opcode == BPF_CALL) {
-			verbose(env, "(%02x) call %s#%d\n", insn->code,
-				func_id_name(insn->imm), insn->imm);
+			if (insn->src_reg == BPF_PSEUDO_CALL)
+				verbose(env, "(%02x) call pc%+d\n", insn->code,
+					insn->imm);
+			else
+				verbose(env, "(%02x) call %s#%d\n", insn->code,
+					func_id_name(insn->imm), insn->imm);
 		} else if (insn->code == (BPF_JMP | BPF_JA)) {
 			verbose(env, "(%02x) goto pc%+d\n",
 				insn->code, insn->off);
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 2c4cfeaa8d5e..e2e1c78ce1dc 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -1194,7 +1194,8 @@ static int bpf_prog_load(union bpf_attr *attr)
 		goto free_used_maps;
 
 	/* eBPF program is ready to be JITed */
-	prog = bpf_prog_select_runtime(prog, &err);
+	if (!prog->bpf_func)
+		prog = bpf_prog_select_runtime(prog, &err);
 	if (err < 0)
 		goto free_used_maps;
 
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 7afa92e9b409..48b2901cf483 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -20,6 +20,8 @@
 #include <linux/file.h>
 #include <linux/vmalloc.h>
 #include <linux/stringify.h>
+#include <linux/bsearch.h>
+#include <linux/sort.h>
 
 #include "disasm.h"
 
@@ -227,13 +229,23 @@ static void print_liveness(struct bpf_verifier_env *env,
 		verbose(env, "w");
 }
 
+static struct bpf_func_state *func(struct bpf_verifier_env *env,
+				   const struct bpf_reg_state *reg)
+{
+	struct bpf_verifier_state *cur = env->cur_state;
+
+	return cur->frame[reg->frameno];
+}
+
 static void print_verifier_state(struct bpf_verifier_env *env,
-				 struct bpf_verifier_state *state)
+				 const struct bpf_func_state *state)
 {
-	struct bpf_reg_state *reg;
+	const struct bpf_reg_state *reg;
 	enum bpf_reg_type t;
 	int i;
 
+	if (state->frameno)
+		verbose(env, " frame%d:", state->frameno);
 	for (i = 0; i < MAX_BPF_REG; i++) {
 		reg = &state->regs[i];
 		t = reg->type;
@@ -246,6 +258,8 @@ static void print_verifier_state(struct bpf_verifier_env *env,
 		    tnum_is_const(reg->var_off)) {
 			/* reg->off should be 0 for SCALAR_VALUE */
 			verbose(env, "%lld", reg->var_off.value + reg->off);
+			if (t == PTR_TO_STACK)
+				verbose(env, ",call_%d", func(env, reg)->callsite);
 		} else {
 			verbose(env, "(id=%d", reg->id);
 			if (t != SCALAR_VALUE)
@@ -297,12 +311,14 @@ static void print_verifier_state(struct bpf_verifier_env *env,
 			verbose(env, "=%s",
 				reg_type_str[state->stack[i].spilled_ptr.type]);
 		}
+		if (state->stack[i].slot_type[0] == STACK_ZERO)
+			verbose(env, " fp%d=0", (-i - 1) * BPF_REG_SIZE);
 	}
 	verbose(env, "\n");
 }
 
-static int copy_stack_state(struct bpf_verifier_state *dst,
-			    const struct bpf_verifier_state *src)
+static int copy_stack_state(struct bpf_func_state *dst,
+			    const struct bpf_func_state *src)
 {
 	if (!src->stack)
 		return 0;
@@ -318,13 +334,13 @@ static int copy_stack_state(struct bpf_verifier_state *dst,
 
 /* do_check() starts with zero-sized stack in struct bpf_verifier_state to
  * make it consume minimal amount of memory. check_stack_write() access from
- * the program calls into realloc_verifier_state() to grow the stack size.
+ * the program calls into realloc_func_state() to grow the stack size.
  * Note there is a non-zero 'parent' pointer inside bpf_verifier_state
  * which this function copies over. It points to previous bpf_verifier_state
  * which is never reallocated
  */
-static int realloc_verifier_state(struct bpf_verifier_state *state, int size,
-				  bool copy_old)
+static int realloc_func_state(struct bpf_func_state *state, int size,
+			      bool copy_old)
 {
 	u32 old_size = state->allocated_stack;
 	struct bpf_stack_state *new_stack;
@@ -357,10 +373,21 @@ static int realloc_verifier_state(struct bpf_verifier_state *state, int size,
 	return 0;
 }
 
+static void free_func_state(struct bpf_func_state *state)
+{
+	kfree(state->stack);
+	kfree(state);
+}
+
 static void free_verifier_state(struct bpf_verifier_state *state,
 				bool free_self)
 {
-	kfree(state->stack);
+	int i;
+
+	for (i = 0; i <= state->curframe; i++) {
+		free_func_state(state->frame[i]);
+		state->frame[i] = NULL;
+	}
 	if (free_self)
 		kfree(state);
 }
@@ -368,18 +395,46 @@ static void free_verifier_state(struct bpf_verifier_state *state,
 /* copy verifier state from src to dst growing dst stack space
  * when necessary to accommodate larger src stack
  */
-static int copy_verifier_state(struct bpf_verifier_state *dst,
-			       const struct bpf_verifier_state *src)
+static int copy_func_state(struct bpf_func_state *dst,
+			   const struct bpf_func_state *src)
 {
 	int err;
 
-	err = realloc_verifier_state(dst, src->allocated_stack, false);
+	err = realloc_func_state(dst, src->allocated_stack, false);
 	if (err)
 		return err;
-	memcpy(dst, src, offsetof(struct bpf_verifier_state, allocated_stack));
+	memcpy(dst, src, offsetof(struct bpf_func_state, allocated_stack));
 	return copy_stack_state(dst, src);
 }
 
+static int copy_verifier_state(struct bpf_verifier_state *dst_state,
+			       const struct bpf_verifier_state *src)
+{
+	struct bpf_func_state *dst;
+	int i, err;
+
+	/* if dst has more stack frames then src frame, free them */
+	for (i = src->curframe + 1; i <= dst_state->curframe; i++) {
+		free_func_state(dst_state->frame[i]);
+		dst_state->frame[i] = NULL;
+	}
+	dst_state->curframe = src->curframe;
+	dst_state->parent = src->parent;
+	for (i = 0; i <= src->curframe; i++) {
+		dst = dst_state->frame[i];
+		if (!dst) {
+			dst = kzalloc(sizeof(*dst), GFP_KERNEL);
+			if (!dst)
+				return -ENOMEM;
+			dst_state->frame[i] = dst;
+		}
+		err = copy_func_state(dst, src->frame[i]);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
 static int pop_stack(struct bpf_verifier_env *env, int *prev_insn_idx,
 		     int *insn_idx)
 {
@@ -441,6 +496,10 @@ err:
 static const int caller_saved[CALLER_SAVED_REGS] = {
 	BPF_REG_0, BPF_REG_1, BPF_REG_2, BPF_REG_3, BPF_REG_4, BPF_REG_5
 };
+#define CALLEE_SAVED_REGS 5
+static const int callee_saved[CALLEE_SAVED_REGS] = {
+	BPF_REG_6, BPF_REG_7, BPF_REG_8, BPF_REG_9
+};
 
 static void __mark_reg_not_init(struct bpf_reg_state *reg);
 
@@ -465,6 +524,13 @@ static void __mark_reg_known_zero(struct bpf_reg_state *reg)
 	__mark_reg_known(reg, 0);
 }
 
+static void __mark_reg_const_zero(struct bpf_reg_state *reg)
+{
+	__mark_reg_known(reg, 0);
+	reg->off = 0;
+	reg->type = SCALAR_VALUE;
+}
+
 static void mark_reg_known_zero(struct bpf_verifier_env *env,
 				struct bpf_reg_state *regs, u32 regno)
 {
@@ -576,6 +642,7 @@ static void __mark_reg_unknown(struct bpf_reg_state *reg)
 	reg->id = 0;
 	reg->off = 0;
 	reg->var_off = tnum_unknown;
+	reg->frameno = 0;
 	__mark_reg_unbounded(reg);
 }
 
@@ -612,8 +679,9 @@ static void mark_reg_not_init(struct bpf_verifier_env *env,
 }
 
 static void init_reg_state(struct bpf_verifier_env *env,
-			   struct bpf_reg_state *regs)
+			   struct bpf_func_state *state)
 {
+	struct bpf_reg_state *regs = state->regs;
 	int i;
 
 	for (i = 0; i < MAX_BPF_REG; i++) {
@@ -624,41 +692,217 @@ static void init_reg_state(struct bpf_verifier_env *env,
 	/* frame pointer */
 	regs[BPF_REG_FP].type = PTR_TO_STACK;
 	mark_reg_known_zero(env, regs, BPF_REG_FP);
+	regs[BPF_REG_FP].frameno = state->frameno;
 
 	/* 1st arg to a function */
 	regs[BPF_REG_1].type = PTR_TO_CTX;
 	mark_reg_known_zero(env, regs, BPF_REG_1);
 }
 
+#define BPF_MAIN_FUNC (-1)
+static void init_func_state(struct bpf_verifier_env *env,
+			    struct bpf_func_state *state,
+			    int callsite, int frameno, int subprogno)
+{
+	state->callsite = callsite;
+	state->frameno = frameno;
+	state->subprogno = subprogno;
+	init_reg_state(env, state);
+}
+
 enum reg_arg_type {
 	SRC_OP,		/* register is used as source operand */
 	DST_OP,		/* register is used as destination operand */
 	DST_OP_NO_MARK	/* same as above, check only, don't mark */
 };
 
-static void mark_reg_read(const struct bpf_verifier_state *state, u32 regno)
+static int cmp_subprogs(const void *a, const void *b)
+{
+	return *(int *)a - *(int *)b;
+}
+
+static int find_subprog(struct bpf_verifier_env *env, int off)
 {
-	struct bpf_verifier_state *parent = state->parent;
+	u32 *p;
+
+	p = bsearch(&off, env->subprog_starts, env->subprog_cnt,
+		    sizeof(env->subprog_starts[0]), cmp_subprogs);
+	if (!p)
+		return -ENOENT;
+	return p - env->subprog_starts;
+
+}
+
+static int add_subprog(struct bpf_verifier_env *env, int off)
+{
+	int insn_cnt = env->prog->len;
+	int ret;
+
+	if (off >= insn_cnt || off < 0) {
+		verbose(env, "call to invalid destination\n");
+		return -EINVAL;
+	}
+	ret = find_subprog(env, off);
+	if (ret >= 0)
+		return 0;
+	if (env->subprog_cnt >= BPF_MAX_SUBPROGS) {
+		verbose(env, "too many subprograms\n");
+		return -E2BIG;
+	}
+	env->subprog_starts[env->subprog_cnt++] = off;
+	sort(env->subprog_starts, env->subprog_cnt,
+	     sizeof(env->subprog_starts[0]), cmp_subprogs, NULL);
+	return 0;
+}
+
+static int check_subprogs(struct bpf_verifier_env *env)
+{
+	int i, ret, subprog_start, subprog_end, off, cur_subprog = 0;
+	struct bpf_insn *insn = env->prog->insnsi;
+	int insn_cnt = env->prog->len;
+
+	/* determine subprog starts. The end is one before the next starts */
+	for (i = 0; i < insn_cnt; i++) {
+		if (insn[i].code != (BPF_JMP | BPF_CALL))
+			continue;
+		if (insn[i].src_reg != BPF_PSEUDO_CALL)
+			continue;
+		if (!env->allow_ptr_leaks) {
+			verbose(env, "function calls to other bpf functions are allowed for root only\n");
+			return -EPERM;
+		}
+		if (bpf_prog_is_dev_bound(env->prog->aux)) {
+			verbose(env, "funcation calls in offloaded programs are not supported yet\n");
+			return -EINVAL;
+		}
+		ret = add_subprog(env, i + insn[i].imm + 1);
+		if (ret < 0)
+			return ret;
+	}
+
+	if (env->log.level > 1)
+		for (i = 0; i < env->subprog_cnt; i++)
+			verbose(env, "func#%d @%d\n", i, env->subprog_starts[i]);
+
+	/* now check that all jumps are within the same subprog */
+	subprog_start = 0;
+	if (env->subprog_cnt == cur_subprog)
+		subprog_end = insn_cnt;
+	else
+		subprog_end = env->subprog_starts[cur_subprog++];
+	for (i = 0; i < insn_cnt; i++) {
+		u8 code = insn[i].code;
+
+		if (BPF_CLASS(code) != BPF_JMP)
+			goto next;
+		if (BPF_OP(code) == BPF_EXIT || BPF_OP(code) == BPF_CALL)
+			goto next;
+		off = i + insn[i].off + 1;
+		if (off < subprog_start || off >= subprog_end) {
+			verbose(env, "jump out of range from insn %d to %d\n", i, off);
+			return -EINVAL;
+		}
+next:
+		if (i == subprog_end - 1) {
+			/* to avoid fall-through from one subprog into another
+			 * the last insn of the subprog should be either exit
+			 * or unconditional jump back
+			 */
+			if (code != (BPF_JMP | BPF_EXIT) &&
+			    code != (BPF_JMP | BPF_JA)) {
+				verbose(env, "last insn is not an exit or jmp\n");
+				return -EINVAL;
+			}
+			subprog_start = subprog_end;
+			if (env->subprog_cnt == cur_subprog)
+				subprog_end = insn_cnt;
+			else
+				subprog_end = env->subprog_starts[cur_subprog++];
+		}
+	}
+	return 0;
+}
+
+struct bpf_verifier_state *skip_callee(struct bpf_verifier_env *env,
+				       const struct bpf_verifier_state *state,
+				       struct bpf_verifier_state *parent,
+				       u32 regno)
+{
+	struct bpf_verifier_state *tmp = NULL;
+
+	/* 'parent' could be a state of caller and
+	 * 'state' could be a state of callee. In such case
+	 * parent->curframe < state->curframe
+	 * and it's ok for r1 - r5 registers
+	 *
+	 * 'parent' could be a callee's state after it bpf_exit-ed.
+	 * In such case parent->curframe > state->curframe
+	 * and it's ok for r0 only
+	 */
+	if (parent->curframe == state->curframe ||
+	    (parent->curframe < state->curframe &&
+	     regno >= BPF_REG_1 && regno <= BPF_REG_5) ||
+	    (parent->curframe > state->curframe &&
+	       regno == BPF_REG_0))
+		return parent;
+
+	if (parent->curframe > state->curframe &&
+	    regno >= BPF_REG_6) {
+		/* for callee saved regs we have to skip the whole chain
+		 * of states that belong to callee and mark as LIVE_READ
+		 * the registers before the call
+		 */
+		tmp = parent;
+		while (tmp && tmp->curframe != state->curframe) {
+			tmp = tmp->parent;
+		}
+		if (!tmp)
+			goto bug;
+		parent = tmp;
+	} else {
+		goto bug;
+	}
+	return parent;
+bug:
+	verbose(env, "verifier bug regno %d tmp %p\n", regno, tmp);
+	verbose(env, "regno %d parent frame %d current frame %d\n",
+		regno, parent->curframe, state->curframe);
+	return 0;
+}
+
+static int mark_reg_read(struct bpf_verifier_env *env,
+			 const struct bpf_verifier_state *state,
+			 struct bpf_verifier_state *parent,
+			 u32 regno)
+{
+	bool writes = parent == state->parent; /* Observe write marks */
 
 	if (regno == BPF_REG_FP)
 		/* We don't need to worry about FP liveness because it's read-only */
-		return;
+		return 0;
 
 	while (parent) {
 		/* if read wasn't screened by an earlier write ... */
-		if (state->regs[regno].live & REG_LIVE_WRITTEN)
+		if (writes && state->frame[state->curframe]->regs[regno].live & REG_LIVE_WRITTEN)
 			break;
+		parent = skip_callee(env, state, parent, regno);
+		if (!parent)
+			return -EFAULT;
 		/* ... then we depend on parent's value */
-		parent->regs[regno].live |= REG_LIVE_READ;
+		parent->frame[parent->curframe]->regs[regno].live |= REG_LIVE_READ;
 		state = parent;
 		parent = state->parent;
+		writes = true;
 	}
+	return 0;
 }
 
 static int check_reg_arg(struct bpf_verifier_env *env, u32 regno,
 			 enum reg_arg_type t)
 {
-	struct bpf_reg_state *regs = env->cur_state->regs;
+	struct bpf_verifier_state *vstate = env->cur_state;
+	struct bpf_func_state *state = vstate->frame[vstate->curframe];
+	struct bpf_reg_state *regs = state->regs;
 
 	if (regno >= MAX_BPF_REG) {
 		verbose(env, "R%d is invalid\n", regno);
@@ -671,7 +915,7 @@ static int check_reg_arg(struct bpf_verifier_env *env, u32 regno,
 			verbose(env, "R%d !read_ok\n", regno);
 			return -EACCES;
 		}
-		mark_reg_read(env->cur_state, regno);
+		return mark_reg_read(env, vstate, vstate->parent, regno);
 	} else {
 		/* check whether register used as dest operand can be written to */
 		if (regno == BPF_REG_FP) {
@@ -702,17 +946,25 @@ static bool is_spillable_regtype(enum bpf_reg_type type)
 	}
 }
 
+/* Does this register contain a constant zero? */
+static bool register_is_null(struct bpf_reg_state *reg)
+{
+	return reg->type == SCALAR_VALUE && tnum_equals_const(reg->var_off, 0);
+}
+
 /* check_stack_read/write functions track spill/fill of registers,
  * stack boundary and alignment are checked in check_mem_access()
  */
 static int check_stack_write(struct bpf_verifier_env *env,
-			     struct bpf_verifier_state *state, int off,
-			     int size, int value_regno)
+			     struct bpf_func_state *state, /* func where register points to */
+			     int off, int size, int value_regno)
 {
+	struct bpf_func_state *cur; /* state of the current function */
 	int i, slot = -off - 1, spi = slot / BPF_REG_SIZE, err;
+	enum bpf_reg_type type;
 
-	err = realloc_verifier_state(state, round_up(slot + 1, BPF_REG_SIZE),
-				     true);
+	err = realloc_func_state(state, round_up(slot + 1, BPF_REG_SIZE),
+				 true);
 	if (err)
 		return err;
 	/* caller checked that off % size == 0 and -MAX_BPF_STACK <= off < 0,
@@ -725,8 +977,9 @@ static int check_stack_write(struct bpf_verifier_env *env,
 		return -EACCES;
 	}
 
+	cur = env->cur_state->frame[env->cur_state->curframe];
 	if (value_regno >= 0 &&
-	    is_spillable_regtype(state->regs[value_regno].type)) {
+	    is_spillable_regtype((type = cur->regs[value_regno].type))) {
 
 		/* register containing pointer is being spilled into stack */
 		if (size != BPF_REG_SIZE) {
@@ -734,51 +987,116 @@ static int check_stack_write(struct bpf_verifier_env *env,
 			return -EACCES;
 		}
 
+		if (state != cur && type == PTR_TO_STACK) {
+			verbose(env, "cannot spill pointers to stack into stack frame of the caller\n");
+			return -EINVAL;
+		}
+
 		/* save register state */
-		state->stack[spi].spilled_ptr = state->regs[value_regno];
+		state->stack[spi].spilled_ptr = cur->regs[value_regno];
 		state->stack[spi].spilled_ptr.live |= REG_LIVE_WRITTEN;
 
 		for (i = 0; i < BPF_REG_SIZE; i++)
 			state->stack[spi].slot_type[i] = STACK_SPILL;
 	} else {
+		u8 type = STACK_MISC;
+
 		/* regular write of data into stack */
 		state->stack[spi].spilled_ptr = (struct bpf_reg_state) {};
 
+		/* only mark the slot as written if all 8 bytes were written
+		 * otherwise read propagation may incorrectly stop too soon
+		 * when stack slots are partially written.
+		 * This heuristic means that read propagation will be
+		 * conservative, since it will add reg_live_read marks
+		 * to stack slots all the way to first state when programs
+		 * writes+reads less than 8 bytes
+		 */
+		if (size == BPF_REG_SIZE)
+			state->stack[spi].spilled_ptr.live |= REG_LIVE_WRITTEN;
+
+		/* when we zero initialize stack slots mark them as such */
+		if (value_regno >= 0 &&
+		    register_is_null(&cur->regs[value_regno]))
+			type = STACK_ZERO;
+
 		for (i = 0; i < size; i++)
 			state->stack[spi].slot_type[(slot - i) % BPF_REG_SIZE] =
-				STACK_MISC;
+				type;
 	}
 	return 0;
 }
 
-static void mark_stack_slot_read(const struct bpf_verifier_state *state, int slot)
+/* registers of every function are unique and mark_reg_read() propagates
+ * the liveness in the following cases:
+ * - from callee into caller for R1 - R5 that were used as arguments
+ * - from caller into callee for R0 that used as result of the call
+ * - from caller to the same caller skipping states of the callee for R6 - R9,
+ *   since R6 - R9 are callee saved by implicit function prologue and
+ *   caller's R6 != callee's R6, so when we propagate liveness up to
+ *   parent states we need to skip callee states for R6 - R9.
+ *
+ * stack slot marking is different, since stacks of caller and callee are
+ * accessible in both (since caller can pass a pointer to caller's stack to
+ * callee which can pass it to another function), hence mark_stack_slot_read()
+ * has to propagate the stack liveness to all parent states at given frame number.
+ * Consider code:
+ * f1() {
+ *   ptr = fp - 8;
+ *   *ptr = ctx;
+ *   call f2 {
+ *      .. = *ptr;
+ *   }
+ *   .. = *ptr;
+ * }
+ * First *ptr is reading from f1's stack and mark_stack_slot_read() has
+ * to mark liveness at the f1's frame and not f2's frame.
+ * Second *ptr is also reading from f1's stack and mark_stack_slot_read() has
+ * to propagate liveness to f2 states at f1's frame level and further into
+ * f1 states at f1's frame level until write into that stack slot
+ */
+static void mark_stack_slot_read(struct bpf_verifier_env *env,
+				 const struct bpf_verifier_state *state,
+				 struct bpf_verifier_state *parent,
+				 int slot, int frameno)
 {
-	struct bpf_verifier_state *parent = state->parent;
+	bool writes = parent == state->parent; /* Observe write marks */
 
 	while (parent) {
+		if (parent->frame[frameno]->allocated_stack <= slot * BPF_REG_SIZE)
+			/* since LIVE_WRITTEN mark is only done for full 8-byte
+			 * write the read marks are conservative and parent
+			 * state may not even have the stack allocated. In such case
+			 * end the propagation, since the loop reached beginning
+			 * of the function
+			 */
+			break;
 		/* if read wasn't screened by an earlier write ... */
-		if (state->stack[slot].spilled_ptr.live & REG_LIVE_WRITTEN)
+		if (writes && state->frame[frameno]->stack[slot].spilled_ptr.live & REG_LIVE_WRITTEN)
 			break;
 		/* ... then we depend on parent's value */
-		parent->stack[slot].spilled_ptr.live |= REG_LIVE_READ;
+		parent->frame[frameno]->stack[slot].spilled_ptr.live |= REG_LIVE_READ;
 		state = parent;
 		parent = state->parent;
+		writes = true;
 	}
 }
 
 static int check_stack_read(struct bpf_verifier_env *env,
-			    struct bpf_verifier_state *state, int off, int size,
-			    int value_regno)
+			    struct bpf_func_state *reg_state /* func where register points to */,
+			    int off, int size, int value_regno)
 {
+	struct bpf_verifier_state *vstate = env->cur_state;
+	struct bpf_func_state *state = vstate->frame[vstate->curframe];
 	int i, slot = -off - 1, spi = slot / BPF_REG_SIZE;
 	u8 *stype;
 
-	if (state->allocated_stack <= slot) {
+	if (reg_state->allocated_stack <= slot) {
 		verbose(env, "invalid read from stack off %d+0 size %d\n",
 			off, size);
 		return -EACCES;
 	}
-	stype = state->stack[spi].slot_type;
+	stype = reg_state->stack[spi].slot_type;
 
 	if (stype[0] == STACK_SPILL) {
 		if (size != BPF_REG_SIZE) {
@@ -794,26 +1112,44 @@ static int check_stack_read(struct bpf_verifier_env *env,
 
 		if (value_regno >= 0) {
 			/* restore register state from stack */
-			state->regs[value_regno] = state->stack[spi].spilled_ptr;
+			state->regs[value_regno] = reg_state->stack[spi].spilled_ptr;
 			/* mark reg as written since spilled pointer state likely
 			 * has its liveness marks cleared by is_state_visited()
 			 * which resets stack/reg liveness for state transitions
 			 */
 			state->regs[value_regno].live |= REG_LIVE_WRITTEN;
-			mark_stack_slot_read(state, spi);
 		}
+		mark_stack_slot_read(env, vstate, vstate->parent, spi,
+				     reg_state->frameno);
 		return 0;
 	} else {
+		int zeros = 0;
+
 		for (i = 0; i < size; i++) {
-			if (stype[(slot - i) % BPF_REG_SIZE] != STACK_MISC) {
-				verbose(env, "invalid read from stack off %d+%d size %d\n",
-					off, i, size);
-				return -EACCES;
+			if (stype[(slot - i) % BPF_REG_SIZE] == STACK_MISC)
+				continue;
+			if (stype[(slot - i) % BPF_REG_SIZE] == STACK_ZERO) {
+				zeros++;
+				continue;
 			}
+			verbose(env, "invalid read from stack off %d+%d size %d\n",
+				off, i, size);
+			return -EACCES;
+		}
+		mark_stack_slot_read(env, vstate, vstate->parent, spi,
+				     reg_state->frameno);
+		if (value_regno >= 0) {
+			if (zeros == size) {
+				/* any size read into register is zero extended,
+				 * so the whole register == const_zero
+				 */
+				__mark_reg_const_zero(&state->regs[value_regno]);
+			} else {
+				/* have read misc data from the stack */
+				mark_reg_unknown(env, state->regs, value_regno);
+			}
+			state->regs[value_regno].live |= REG_LIVE_WRITTEN;
 		}
-		if (value_regno >= 0)
-			/* have read misc data from the stack */
-			mark_reg_unknown(env, state->regs, value_regno);
 		return 0;
 	}
 }
@@ -838,7 +1174,8 @@ static int __check_map_access(struct bpf_verifier_env *env, u32 regno, int off,
 static int check_map_access(struct bpf_verifier_env *env, u32 regno,
 			    int off, int size, bool zero_size_allowed)
 {
-	struct bpf_verifier_state *state = env->cur_state;
+	struct bpf_verifier_state *vstate = env->cur_state;
+	struct bpf_func_state *state = vstate->frame[vstate->curframe];
 	struct bpf_reg_state *reg = &state->regs[regno];
 	int err;
 
@@ -1088,6 +1425,54 @@ static int check_ptr_alignment(struct bpf_verifier_env *env,
 					   strict);
 }
 
+static int update_stack_depth(struct bpf_verifier_env *env,
+			      const struct bpf_func_state *func,
+			      int off)
+{
+	u16 stack = env->subprog_stack_depth[func->subprogno], total = 0;
+	struct bpf_verifier_state *cur = env->cur_state;
+	int i;
+
+	if (stack >= -off)
+		return 0;
+
+	/* update known max for given subprogram */
+	env->subprog_stack_depth[func->subprogno] = -off;
+
+	/* compute the total for current call chain */
+	for (i = 0; i <= cur->curframe; i++) {
+		u32 depth = env->subprog_stack_depth[cur->frame[i]->subprogno];
+
+		/* round up to 32-bytes, since this is granularity
+		 * of interpreter stack sizes
+		 */
+		depth = round_up(depth, 32);
+		total += depth;
+	}
+
+	if (total > MAX_BPF_STACK) {
+		verbose(env, "combined stack size of %d calls is %d. Too large\n",
+			cur->curframe, total);
+		return -EACCES;
+	}
+	return 0;
+}
+
+static int get_callee_stack_depth(struct bpf_verifier_env *env,
+				  const struct bpf_insn *insn, int idx)
+{
+	int start = idx + insn->imm + 1, subprog;
+
+	subprog = find_subprog(env, start);
+	if (subprog < 0) {
+		WARN_ONCE(1, "verifier bug. No program starts at insn %d\n",
+			  start);
+		return -EFAULT;
+	}
+	subprog++;
+	return env->subprog_stack_depth[subprog];
+}
+
 /* check whether memory at (regno + off) is accessible for t = (read | write)
  * if t==write, value_regno is a register which value is stored into memory
  * if t==read, value_regno is a register which will receive the value from memory
@@ -1098,9 +1483,9 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, u32 regn
 			    int bpf_size, enum bpf_access_type t,
 			    int value_regno)
 {
-	struct bpf_verifier_state *state = env->cur_state;
 	struct bpf_reg_state *regs = cur_regs(env);
 	struct bpf_reg_state *reg = regs + regno;
+	struct bpf_func_state *state;
 	int size, err = 0;
 
 	size = bpf_size_to_bytes(bpf_size);
@@ -1189,8 +1574,10 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, u32 regn
 			return -EACCES;
 		}
 
-		if (env->prog->aux->stack_depth < -off)
-			env->prog->aux->stack_depth = -off;
+		state = func(env, reg);
+		err = update_stack_depth(env, state, off);
+		if (err)
+			return err;
 
 		if (t == BPF_WRITE)
 			err = check_stack_write(env, state, off, size,
@@ -1264,12 +1651,6 @@ static int check_xadd(struct bpf_verifier_env *env, int insn_idx, struct bpf_ins
 				BPF_SIZE(insn->code), BPF_WRITE, -1);
 }
 
-/* Does this register contain a constant zero? */
-static bool register_is_null(struct bpf_reg_state *reg)
-{
-	return reg->type == SCALAR_VALUE && tnum_equals_const(reg->var_off, 0);
-}
-
 /* when register 'regno' is passed into function that will read 'access_size'
  * bytes from that pointer, make sure that it's within stack boundary
  * and all elements of stack are initialized.
@@ -1281,7 +1662,7 @@ static int check_stack_boundary(struct bpf_verifier_env *env, int regno,
 				struct bpf_call_arg_meta *meta)
 {
 	struct bpf_reg_state *reg = cur_regs(env) + regno;
-	struct bpf_verifier_state *state = env->cur_state;
+	struct bpf_func_state *state = func(env, reg);
 	int off, i, slot, spi;
 
 	if (reg->type != PTR_TO_STACK) {
@@ -1312,9 +1693,6 @@ static int check_stack_boundary(struct bpf_verifier_env *env, int regno,
 		return -EACCES;
 	}
 
-	if (env->prog->aux->stack_depth < -off)
-		env->prog->aux->stack_depth = -off;
-
 	if (meta && meta->raw_mode) {
 		meta->access_size = access_size;
 		meta->regno = regno;
@@ -1322,17 +1700,32 @@ static int check_stack_boundary(struct bpf_verifier_env *env, int regno,
 	}
 
 	for (i = 0; i < access_size; i++) {
+		u8 *stype;
+
 		slot = -(off + i) - 1;
 		spi = slot / BPF_REG_SIZE;
-		if (state->allocated_stack <= slot ||
-		    state->stack[spi].slot_type[slot % BPF_REG_SIZE] !=
-			STACK_MISC) {
-			verbose(env, "invalid indirect read from stack off %d+%d size %d\n",
-				off, i, access_size);
-			return -EACCES;
+		if (state->allocated_stack <= slot)
+			goto err;
+		stype = &state->stack[spi].slot_type[slot % BPF_REG_SIZE];
+		if (*stype == STACK_MISC)
+			goto mark;
+		if (*stype == STACK_ZERO) {
+			/* helper can write anything into the stack */
+			*stype = STACK_MISC;
+			goto mark;
 		}
+err:
+		verbose(env, "invalid indirect read from stack off %d+%d size %d\n",
+			off, i, access_size);
+		return -EACCES;
+mark:
+		/* reading any byte out of 8-byte 'spill_slot' will cause
+		 * the whole slot to be marked as 'read'
+		 */
+		mark_stack_slot_read(env, env->cur_state, env->cur_state->parent,
+				     spi, state->frameno);
 	}
-	return 0;
+	return update_stack_depth(env, state, off);
 }
 
 static int check_helper_mem_access(struct bpf_verifier_env *env, int regno,
@@ -1585,6 +1978,10 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
 	case BPF_FUNC_tail_call:
 		if (map->map_type != BPF_MAP_TYPE_PROG_ARRAY)
 			goto error;
+		if (env->subprog_cnt) {
+			verbose(env, "tail_calls are not allowed in programs with bpf-to-bpf calls\n");
+			return -EINVAL;
+		}
 		break;
 	case BPF_FUNC_perf_event_read:
 	case BPF_FUNC_perf_event_output:
@@ -1646,9 +2043,9 @@ static int check_raw_mode(const struct bpf_func_proto *fn)
 /* Packet data might have moved, any old PTR_TO_PACKET[_META,_END]
  * are now invalid, so turn them into unknown SCALAR_VALUE.
  */
-static void clear_all_pkt_pointers(struct bpf_verifier_env *env)
+static void __clear_all_pkt_pointers(struct bpf_verifier_env *env,
+				     struct bpf_func_state *state)
 {
-	struct bpf_verifier_state *state = env->cur_state;
 	struct bpf_reg_state *regs = state->regs, *reg;
 	int i;
 
@@ -1665,7 +2062,121 @@ static void clear_all_pkt_pointers(struct bpf_verifier_env *env)
 	}
 }
 
-static int check_call(struct bpf_verifier_env *env, int func_id, int insn_idx)
+static void clear_all_pkt_pointers(struct bpf_verifier_env *env)
+{
+	struct bpf_verifier_state *vstate = env->cur_state;
+	int i;
+
+	for (i = 0; i <= vstate->curframe; i++)
+		__clear_all_pkt_pointers(env, vstate->frame[i]);
+}
+
+static int check_func_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
+			   int *insn_idx)
+{
+	struct bpf_verifier_state *state = env->cur_state;
+	struct bpf_func_state *caller, *callee;
+	int i, subprog, target_insn;
+
+	if (state->curframe >= MAX_CALL_FRAMES) {
+		verbose(env, "the call stack of %d frames is too deep\n",
+			state->curframe);
+		return -E2BIG;
+	}
+
+	target_insn = *insn_idx + insn->imm;
+	subprog = find_subprog(env, target_insn + 1);
+	if (subprog < 0) {
+		verbose(env, "verifier bug. No program starts at insn %d\n",
+			target_insn + 1);
+		return -EFAULT;
+	}
+
+	caller = state->frame[state->curframe];
+	if (state->frame[state->curframe + 1]) {
+		verbose(env, "verifier bug. Frame %d already allocated\n",
+			state->curframe + 1);
+		return -EFAULT;
+	}
+
+	callee = kzalloc(sizeof(*callee), GFP_KERNEL);
+	if (!callee)
+		return -ENOMEM;
+	state->frame[state->curframe + 1] = callee;
+
+	/* callee cannot access r0, r6 - r9 for reading and has to write
+	 * into its own stack before reading from it.
+	 * callee can read/write into caller's stack
+	 */
+	init_func_state(env, callee,
+			/* remember the callsite, it will be used by bpf_exit */
+			*insn_idx /* callsite */,
+			state->curframe + 1 /* frameno within this callchain */,
+			subprog + 1 /* subprog number within this prog */);
+
+	/* copy r1 - r5 args that callee can access */
+	for (i = BPF_REG_1; i <= BPF_REG_5; i++)
+		callee->regs[i] = caller->regs[i];
+
+	/* after the call regsiters r0 - r5 were scratched */
+	for (i = 0; i < CALLER_SAVED_REGS; i++) {
+		mark_reg_not_init(env, caller->regs, caller_saved[i]);
+		check_reg_arg(env, caller_saved[i], DST_OP_NO_MARK);
+	}
+
+	/* only increment it after check_reg_arg() finished */
+	state->curframe++;
+
+	/* and go analyze first insn of the callee */
+	*insn_idx = target_insn;
+
+	if (env->log.level) {
+		verbose(env, "caller:\n");
+		print_verifier_state(env, caller);
+		verbose(env, "callee:\n");
+		print_verifier_state(env, callee);
+	}
+	return 0;
+}
+
+static int prepare_func_exit(struct bpf_verifier_env *env, int *insn_idx)
+{
+	struct bpf_verifier_state *state = env->cur_state;
+	struct bpf_func_state *caller, *callee;
+	struct bpf_reg_state *r0;
+
+	callee = state->frame[state->curframe];
+	r0 = &callee->regs[BPF_REG_0];
+	if (r0->type == PTR_TO_STACK) {
+		/* technically it's ok to return caller's stack pointer
+		 * (or caller's caller's pointer) back to the caller,
+		 * since these pointers are valid. Only current stack
+		 * pointer will be invalid as soon as function exits,
+		 * but let's be conservative
+		 */
+		verbose(env, "cannot return stack pointer to the caller\n");
+		return -EINVAL;
+	}
+
+	state->curframe--;
+	caller = state->frame[state->curframe];
+	/* return to the caller whatever r0 had in the callee */
+	caller->regs[BPF_REG_0] = *r0;
+
+	*insn_idx = callee->callsite + 1;
+	if (env->log.level) {
+		verbose(env, "returning from callee:\n");
+		print_verifier_state(env, callee);
+		verbose(env, "to caller at %d:\n", *insn_idx);
+		print_verifier_state(env, caller);
+	}
+	/* clear everything in the callee */
+	free_func_state(callee);
+	state->frame[state->curframe + 1] = NULL;
+	return 0;
+}
+
+static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn_idx)
 {
 	const struct bpf_func_proto *fn = NULL;
 	struct bpf_reg_state *regs;
@@ -1825,7 +2336,9 @@ static int adjust_ptr_min_max_vals(struct bpf_verifier_env *env,
 				   const struct bpf_reg_state *ptr_reg,
 				   const struct bpf_reg_state *off_reg)
 {
-	struct bpf_reg_state *regs = cur_regs(env), *dst_reg;
+	struct bpf_verifier_state *vstate = env->cur_state;
+	struct bpf_func_state *state = vstate->frame[vstate->curframe];
+	struct bpf_reg_state *regs = state->regs, *dst_reg;
 	bool known = tnum_is_const(off_reg->var_off);
 	s64 smin_val = off_reg->smin_value, smax_val = off_reg->smax_value,
 	    smin_ptr = ptr_reg->smin_value, smax_ptr = ptr_reg->smax_value;
@@ -1837,13 +2350,13 @@ static int adjust_ptr_min_max_vals(struct bpf_verifier_env *env,
 	dst_reg = &regs[dst];
 
 	if (WARN_ON_ONCE(known && (smin_val != smax_val))) {
-		print_verifier_state(env, env->cur_state);
+		print_verifier_state(env, state);
 		verbose(env,
 			"verifier internal error: known but bad sbounds\n");
 		return -EINVAL;
 	}
 	if (WARN_ON_ONCE(known && (umin_val != umax_val))) {
-		print_verifier_state(env, env->cur_state);
+		print_verifier_state(env, state);
 		verbose(env,
 			"verifier internal error: known but bad ubounds\n");
 		return -EINVAL;
@@ -2245,7 +2758,9 @@ static int adjust_scalar_min_max_vals(struct bpf_verifier_env *env,
 static int adjust_reg_min_max_vals(struct bpf_verifier_env *env,
 				   struct bpf_insn *insn)
 {
-	struct bpf_reg_state *regs = cur_regs(env), *dst_reg, *src_reg;
+	struct bpf_verifier_state *vstate = env->cur_state;
+	struct bpf_func_state *state = vstate->frame[vstate->curframe];
+	struct bpf_reg_state *regs = state->regs, *dst_reg, *src_reg;
 	struct bpf_reg_state *ptr_reg = NULL, off_reg = {0};
 	u8 opcode = BPF_OP(insn->code);
 	int rc;
@@ -2319,12 +2834,12 @@ static int adjust_reg_min_max_vals(struct bpf_verifier_env *env,
 
 	/* Got here implies adding two SCALAR_VALUEs */
 	if (WARN_ON_ONCE(ptr_reg)) {
-		print_verifier_state(env, env->cur_state);
+		print_verifier_state(env, state);
 		verbose(env, "verifier internal error: unexpected ptr_reg\n");
 		return -EINVAL;
 	}
 	if (WARN_ON(!src_reg)) {
-		print_verifier_state(env, env->cur_state);
+		print_verifier_state(env, state);
 		verbose(env, "verifier internal error: no src_reg\n");
 		return -EINVAL;
 	}
@@ -2478,14 +2993,15 @@ static int check_alu_op(struct bpf_verifier_env *env, struct bpf_insn *insn)
 	return 0;
 }
 
-static void find_good_pkt_pointers(struct bpf_verifier_state *state,
+static void find_good_pkt_pointers(struct bpf_verifier_state *vstate,
 				   struct bpf_reg_state *dst_reg,
 				   enum bpf_reg_type type,
 				   bool range_right_open)
 {
+	struct bpf_func_state *state = vstate->frame[vstate->curframe];
 	struct bpf_reg_state *regs = state->regs, *reg;
 	u16 new_range;
-	int i;
+	int i, j;
 
 	if (dst_reg->off < 0 ||
 	    (dst_reg->off == 0 && range_right_open))
@@ -2555,12 +3071,15 @@ static void find_good_pkt_pointers(struct bpf_verifier_state *state,
 			/* keep the maximum range already checked */
 			regs[i].range = max(regs[i].range, new_range);
 
-	for (i = 0; i < state->allocated_stack / BPF_REG_SIZE; i++) {
-		if (state->stack[i].slot_type[0] != STACK_SPILL)
-			continue;
-		reg = &state->stack[i].spilled_ptr;
-		if (reg->type == type && reg->id == dst_reg->id)
-			reg->range = max(reg->range, new_range);
+	for (j = 0; j <= vstate->curframe; j++) {
+		state = vstate->frame[j];
+		for (i = 0; i < state->allocated_stack / BPF_REG_SIZE; i++) {
+			if (state->stack[i].slot_type[0] != STACK_SPILL)
+				continue;
+			reg = &state->stack[i].spilled_ptr;
+			if (reg->type == type && reg->id == dst_reg->id)
+				reg->range = max(reg->range, new_range);
+		}
 	}
 }
 
@@ -2798,20 +3317,24 @@ static void mark_map_reg(struct bpf_reg_state *regs, u32 regno, u32 id,
 /* The logic is similar to find_good_pkt_pointers(), both could eventually
  * be folded together at some point.
  */
-static void mark_map_regs(struct bpf_verifier_state *state, u32 regno,
+static void mark_map_regs(struct bpf_verifier_state *vstate, u32 regno,
 			  bool is_null)
 {
+	struct bpf_func_state *state = vstate->frame[vstate->curframe];
 	struct bpf_reg_state *regs = state->regs;
 	u32 id = regs[regno].id;
-	int i;
+	int i, j;
 
 	for (i = 0; i < MAX_BPF_REG; i++)
 		mark_map_reg(regs, i, id, is_null);
 
-	for (i = 0; i < state->allocated_stack / BPF_REG_SIZE; i++) {
-		if (state->stack[i].slot_type[0] != STACK_SPILL)
-			continue;
-		mark_map_reg(&state->stack[i].spilled_ptr, 0, id, is_null);
+	for (j = 0; j <= vstate->curframe; j++) {
+		state = vstate->frame[j];
+		for (i = 0; i < state->allocated_stack / BPF_REG_SIZE; i++) {
+			if (state->stack[i].slot_type[0] != STACK_SPILL)
+				continue;
+			mark_map_reg(&state->stack[i].spilled_ptr, 0, id, is_null);
+		}
 	}
 }
 
@@ -2911,8 +3434,10 @@ static bool try_match_pkt_pointers(const struct bpf_insn *insn,
 static int check_cond_jmp_op(struct bpf_verifier_env *env,
 			     struct bpf_insn *insn, int *insn_idx)
 {
-	struct bpf_verifier_state *other_branch, *this_branch = env->cur_state;
-	struct bpf_reg_state *regs = this_branch->regs, *dst_reg;
+	struct bpf_verifier_state *this_branch = env->cur_state;
+	struct bpf_verifier_state *other_branch;
+	struct bpf_reg_state *regs = this_branch->frame[this_branch->curframe]->regs;
+	struct bpf_reg_state *dst_reg, *other_branch_regs;
 	u8 opcode = BPF_OP(insn->code);
 	int err;
 
@@ -2975,6 +3500,7 @@ static int check_cond_jmp_op(struct bpf_verifier_env *env,
 	other_branch = push_stack(env, *insn_idx + insn->off + 1, *insn_idx);
 	if (!other_branch)
 		return -EFAULT;
+	other_branch_regs = other_branch->frame[other_branch->curframe]->regs;
 
 	/* detect if we are comparing against a constant value so we can adjust
 	 * our min/max values for our dst register.
@@ -2987,22 +3513,22 @@ static int check_cond_jmp_op(struct bpf_verifier_env *env,
 		if (dst_reg->type == SCALAR_VALUE &&
 		    regs[insn->src_reg].type == SCALAR_VALUE) {
 			if (tnum_is_const(regs[insn->src_reg].var_off))
-				reg_set_min_max(&other_branch->regs[insn->dst_reg],
+				reg_set_min_max(&other_branch_regs[insn->dst_reg],
 						dst_reg, regs[insn->src_reg].var_off.value,
 						opcode);
 			else if (tnum_is_const(dst_reg->var_off))
-				reg_set_min_max_inv(&other_branch->regs[insn->src_reg],
+				reg_set_min_max_inv(&other_branch_regs[insn->src_reg],
 						    &regs[insn->src_reg],
 						    dst_reg->var_off.value, opcode);
 			else if (opcode == BPF_JEQ || opcode == BPF_JNE)
 				/* Comparing for equality, we can combine knowledge */
-				reg_combine_min_max(&other_branch->regs[insn->src_reg],
-						    &other_branch->regs[insn->dst_reg],
+				reg_combine_min_max(&other_branch_regs[insn->src_reg],
+						    &other_branch_regs[insn->dst_reg],
 						    &regs[insn->src_reg],
 						    &regs[insn->dst_reg], opcode);
 		}
 	} else if (dst_reg->type == SCALAR_VALUE) {
-		reg_set_min_max(&other_branch->regs[insn->dst_reg],
+		reg_set_min_max(&other_branch_regs[insn->dst_reg],
 					dst_reg, insn->imm, opcode);
 	}
 
@@ -3023,7 +3549,7 @@ static int check_cond_jmp_op(struct bpf_verifier_env *env,
 		return -EACCES;
 	}
 	if (env->log.level)
-		print_verifier_state(env, this_branch);
+		print_verifier_state(env, this_branch->frame[this_branch->curframe]);
 	return 0;
 }
 
@@ -3108,6 +3634,18 @@ static int check_ld_abs(struct bpf_verifier_env *env, struct bpf_insn *insn)
 		return -EINVAL;
 	}
 
+	if (env->subprog_cnt) {
+		/* when program has LD_ABS insn JITs and interpreter assume
+		 * that r1 == ctx == skb which is not the case for callees
+		 * that can have arbitrary arguments. It's problematic
+		 * for main prog as well since JITs would need to analyze
+		 * all functions in order to make proper register save/restore
+		 * decisions in the main prog. Hence disallow LD_ABS with calls
+		 */
+		verbose(env, "BPF_LD_[ABS|IND] instructions cannot be mixed with bpf-to-bpf calls\n");
+		return -EINVAL;
+	}
+
 	if (insn->dst_reg != BPF_REG_0 || insn->off != 0 ||
 	    BPF_SIZE(insn->code) == BPF_DW ||
 	    (mode == BPF_ABS && insn->src_reg != BPF_REG_0)) {
@@ -3284,6 +3822,10 @@ static int check_cfg(struct bpf_verifier_env *env)
 	int ret = 0;
 	int i, t;
 
+	ret = check_subprogs(env);
+	if (ret < 0)
+		return ret;
+
 	insn_state = kcalloc(insn_cnt, sizeof(int), GFP_KERNEL);
 	if (!insn_state)
 		return -ENOMEM;
@@ -3316,6 +3858,14 @@ peek_stack:
 				goto err_free;
 			if (t + 1 < insn_cnt)
 				env->explored_states[t + 1] = STATE_LIST_MARK;
+			if (insns[t].src_reg == BPF_PSEUDO_CALL) {
+				env->explored_states[t] = STATE_LIST_MARK;
+				ret = push_insn(t, t + insns[t].imm + 1, BRANCH, env);
+				if (ret == 1)
+					goto peek_stack;
+				else if (ret < 0)
+					goto err_free;
+			}
 		} else if (opcode == BPF_JA) {
 			if (BPF_SRC(insns[t].code) != BPF_K) {
 				ret = -EINVAL;
@@ -3434,11 +3984,21 @@ static bool check_ids(u32 old_id, u32 cur_id, struct idpair *idmap)
 static bool regsafe(struct bpf_reg_state *rold, struct bpf_reg_state *rcur,
 		    struct idpair *idmap)
 {
+	bool equal;
+
 	if (!(rold->live & REG_LIVE_READ))
 		/* explored state didn't use this */
 		return true;
 
-	if (memcmp(rold, rcur, offsetof(struct bpf_reg_state, live)) == 0)
+	equal = memcmp(rold, rcur, offsetof(struct bpf_reg_state, frameno)) == 0;
+
+	if (rold->type == PTR_TO_STACK)
+		/* two stack pointers are equal only if they're pointing to
+		 * the same stack frame, since fp-8 in foo != fp-8 in bar
+		 */
+		return equal && rold->frameno == rcur->frameno;
+
+	if (equal)
 		return true;
 
 	if (rold->type == NOT_INIT)
@@ -3511,7 +4071,6 @@ static bool regsafe(struct bpf_reg_state *rold, struct bpf_reg_state *rcur,
 		       tnum_in(rold->var_off, rcur->var_off);
 	case PTR_TO_CTX:
 	case CONST_PTR_TO_MAP:
-	case PTR_TO_STACK:
 	case PTR_TO_PACKET_END:
 		/* Only valid matches are exact, which memcmp() above
 		 * would have accepted
@@ -3526,8 +4085,8 @@ static bool regsafe(struct bpf_reg_state *rold, struct bpf_reg_state *rcur,
 	return false;
 }
 
-static bool stacksafe(struct bpf_verifier_state *old,
-		      struct bpf_verifier_state *cur,
+static bool stacksafe(struct bpf_func_state *old,
+		      struct bpf_func_state *cur,
 		      struct idpair *idmap)
 {
 	int i, spi;
@@ -3545,8 +4104,19 @@ static bool stacksafe(struct bpf_verifier_state *old,
 	for (i = 0; i < old->allocated_stack; i++) {
 		spi = i / BPF_REG_SIZE;
 
+		if (!(old->stack[spi].spilled_ptr.live & REG_LIVE_READ))
+			/* explored state didn't use this */
+			return true;
+
 		if (old->stack[spi].slot_type[i % BPF_REG_SIZE] == STACK_INVALID)
 			continue;
+		/* if old state was safe with misc data in the stack
+		 * it will be safe with zero-initialized stack.
+		 * The opposite is not true
+		 */
+		if (old->stack[spi].slot_type[i % BPF_REG_SIZE] == STACK_MISC &&
+		    cur->stack[spi].slot_type[i % BPF_REG_SIZE] == STACK_ZERO)
+			continue;
 		if (old->stack[spi].slot_type[i % BPF_REG_SIZE] !=
 		    cur->stack[spi].slot_type[i % BPF_REG_SIZE])
 			/* Ex: old explored (safe) state has STACK_SPILL in
@@ -3603,9 +4173,8 @@ static bool stacksafe(struct bpf_verifier_state *old,
  * whereas register type in current state is meaningful, it means that
  * the current state will reach 'bpf_exit' instruction safely
  */
-static bool states_equal(struct bpf_verifier_env *env,
-			 struct bpf_verifier_state *old,
-			 struct bpf_verifier_state *cur)
+static bool func_states_equal(struct bpf_func_state *old,
+			      struct bpf_func_state *cur)
 {
 	struct idpair *idmap;
 	bool ret = false;
@@ -3629,71 +4198,72 @@ out_free:
 	return ret;
 }
 
+static bool states_equal(struct bpf_verifier_env *env,
+			 struct bpf_verifier_state *old,
+			 struct bpf_verifier_state *cur)
+{
+	int i;
+
+	if (old->curframe != cur->curframe)
+		return false;
+
+	/* for states to be equal callsites have to be the same
+	 * and all frame states need to be equivalent
+	 */
+	for (i = 0; i <= old->curframe; i++) {
+		if (old->frame[i]->callsite != cur->frame[i]->callsite)
+			return false;
+		if (!func_states_equal(old->frame[i], cur->frame[i]))
+			return false;
+	}
+	return true;
+}
+
 /* A write screens off any subsequent reads; but write marks come from the
- * straight-line code between a state and its parent.  When we arrive at a
- * jump target (in the first iteration of the propagate_liveness() loop),
- * we didn't arrive by the straight-line code, so read marks in state must
- * propagate to parent regardless of state's write marks.
+ * straight-line code between a state and its parent.  When we arrive at an
+ * equivalent state (jump target or such) we didn't arrive by the straight-line
+ * code, so read marks in the state must propagate to the parent regardless
+ * of the state's write marks. That's what 'parent == state->parent' comparison
+ * in mark_reg_read() and mark_stack_slot_read() is for.
  */
-static bool do_propagate_liveness(const struct bpf_verifier_state *state,
-				  struct bpf_verifier_state *parent)
+static int propagate_liveness(struct bpf_verifier_env *env,
+			      const struct bpf_verifier_state *vstate,
+			      struct bpf_verifier_state *vparent)
 {
-	bool writes = parent == state->parent; /* Observe write marks */
-	bool touched = false; /* any changes made? */
-	int i;
+	int i, frame, err = 0;
+	struct bpf_func_state *state, *parent;
 
-	if (!parent)
-		return touched;
+	if (vparent->curframe != vstate->curframe) {
+		WARN(1, "propagate_live: parent frame %d current frame %d\n",
+		     vparent->curframe, vstate->curframe);
+		return -EFAULT;
+	}
 	/* Propagate read liveness of registers... */
 	BUILD_BUG_ON(BPF_REG_FP + 1 != MAX_BPF_REG);
 	/* We don't need to worry about FP liveness because it's read-only */
 	for (i = 0; i < BPF_REG_FP; i++) {
-		if (parent->regs[i].live & REG_LIVE_READ)
+		if (vparent->frame[vparent->curframe]->regs[i].live & REG_LIVE_READ)
 			continue;
-		if (writes && (state->regs[i].live & REG_LIVE_WRITTEN))
-			continue;
-		if (state->regs[i].live & REG_LIVE_READ) {
-			parent->regs[i].live |= REG_LIVE_READ;
-			touched = true;
+		if (vstate->frame[vstate->curframe]->regs[i].live & REG_LIVE_READ) {
+			err = mark_reg_read(env, vstate, vparent, i);
+			if (err)
+				return err;
 		}
 	}
+
 	/* ... and stack slots */
-	for (i = 0; i < state->allocated_stack / BPF_REG_SIZE &&
-		    i < parent->allocated_stack / BPF_REG_SIZE; i++) {
-		if (parent->stack[i].slot_type[0] != STACK_SPILL)
-			continue;
-		if (state->stack[i].slot_type[0] != STACK_SPILL)
-			continue;
-		if (parent->stack[i].spilled_ptr.live & REG_LIVE_READ)
-			continue;
-		if (writes &&
-		    (state->stack[i].spilled_ptr.live & REG_LIVE_WRITTEN))
-			continue;
-		if (state->stack[i].spilled_ptr.live & REG_LIVE_READ) {
-			parent->stack[i].spilled_ptr.live |= REG_LIVE_READ;
-			touched = true;
+	for (frame = 0; frame <= vstate->curframe; frame++) {
+		state = vstate->frame[frame];
+		parent = vparent->frame[frame];
+		for (i = 0; i < state->allocated_stack / BPF_REG_SIZE &&
+			    i < parent->allocated_stack / BPF_REG_SIZE; i++) {
+			if (parent->stack[i].spilled_ptr.live & REG_LIVE_READ)
+				continue;
+			if (state->stack[i].spilled_ptr.live & REG_LIVE_READ)
+				mark_stack_slot_read(env, vstate, vparent, i, frame);
 		}
 	}
-	return touched;
-}
-
-/* "parent" is "a state from which we reach the current state", but initially
- * it is not the state->parent (i.e. "the state whose straight-line code leads
- * to the current state"), instead it is the state that happened to arrive at
- * a (prunable) equivalent of the current state.  See comment above
- * do_propagate_liveness() for consequences of this.
- * This function is just a more efficient way of calling mark_reg_read() or
- * mark_stack_slot_read() on each reg in "parent" that is read in "state",
- * though it requires that parent != state->parent in the call arguments.
- */
-static void propagate_liveness(const struct bpf_verifier_state *state,
-			       struct bpf_verifier_state *parent)
-{
-	while (do_propagate_liveness(state, parent)) {
-		/* Something changed, so we need to feed those changes onward */
-		state = parent;
-		parent = state->parent;
-	}
+	return err;
 }
 
 static int is_state_visited(struct bpf_verifier_env *env, int insn_idx)
@@ -3701,7 +4271,7 @@ static int is_state_visited(struct bpf_verifier_env *env, int insn_idx)
 	struct bpf_verifier_state_list *new_sl;
 	struct bpf_verifier_state_list *sl;
 	struct bpf_verifier_state *cur = env->cur_state;
-	int i, err;
+	int i, j, err;
 
 	sl = env->explored_states[insn_idx];
 	if (!sl)
@@ -3722,7 +4292,9 @@ static int is_state_visited(struct bpf_verifier_env *env, int insn_idx)
 			 * they'll be immediately forgotten as we're pruning
 			 * this state and will pop a new one.
 			 */
-			propagate_liveness(&sl->state, cur);
+			err = propagate_liveness(env, &sl->state, cur);
+			if (err)
+				return err;
 			return 1;
 		}
 		sl = sl->next;
@@ -3730,9 +4302,10 @@ static int is_state_visited(struct bpf_verifier_env *env, int insn_idx)
 
 	/* there were no equivalent states, remember current one.
 	 * technically the current state is not proven to be safe yet,
-	 * but it will either reach bpf_exit (which means it's safe) or
-	 * it will be rejected. Since there are no loops, we won't be
-	 * seeing this 'insn_idx' instruction again on the way to bpf_exit
+	 * but it will either reach outer most bpf_exit (which means it's safe)
+	 * or it will be rejected. Since there are no loops, we won't be
+	 * seeing this tuple (frame[0].callsite, frame[1].callsite, .. insn_idx)
+	 * again on the way to bpf_exit
 	 */
 	new_sl = kzalloc(sizeof(struct bpf_verifier_state_list), GFP_KERNEL);
 	if (!new_sl)
@@ -3756,10 +4329,15 @@ static int is_state_visited(struct bpf_verifier_env *env, int insn_idx)
 	 * explored_states can get read marks.)
 	 */
 	for (i = 0; i < BPF_REG_FP; i++)
-		cur->regs[i].live = REG_LIVE_NONE;
-	for (i = 0; i < cur->allocated_stack / BPF_REG_SIZE; i++)
-		if (cur->stack[i].slot_type[0] == STACK_SPILL)
-			cur->stack[i].spilled_ptr.live = REG_LIVE_NONE;
+		cur->frame[cur->curframe]->regs[i].live = REG_LIVE_NONE;
+
+	/* all stack frames are accessible from callee, clear them all */
+	for (j = 0; j <= cur->curframe; j++) {
+		struct bpf_func_state *frame = cur->frame[j];
+
+		for (i = 0; i < frame->allocated_stack / BPF_REG_SIZE; i++)
+			frame->stack[i].spilled_ptr.live = REG_LIVE_NONE;
+	}
 	return 0;
 }
 
@@ -3777,7 +4355,7 @@ static int do_check(struct bpf_verifier_env *env)
 	struct bpf_verifier_state *state;
 	struct bpf_insn *insns = env->prog->insnsi;
 	struct bpf_reg_state *regs;
-	int insn_cnt = env->prog->len;
+	int insn_cnt = env->prog->len, i;
 	int insn_idx, prev_insn_idx = 0;
 	int insn_processed = 0;
 	bool do_print_state = false;
@@ -3785,9 +4363,18 @@ static int do_check(struct bpf_verifier_env *env)
 	state = kzalloc(sizeof(struct bpf_verifier_state), GFP_KERNEL);
 	if (!state)
 		return -ENOMEM;
-	env->cur_state = state;
-	init_reg_state(env, state->regs);
+	state->curframe = 0;
 	state->parent = NULL;
+	state->frame[0] = kzalloc(sizeof(struct bpf_func_state), GFP_KERNEL);
+	if (!state->frame[0]) {
+		kfree(state);
+		return -ENOMEM;
+	}
+	env->cur_state = state;
+	init_func_state(env, state->frame[0],
+			BPF_MAIN_FUNC /* callsite */,
+			0 /* frameno */,
+			0 /* subprogno, zero == main subprog */);
 	insn_idx = 0;
 	for (;;) {
 		struct bpf_insn *insn;
@@ -3834,7 +4421,7 @@ static int do_check(struct bpf_verifier_env *env)
 			else
 				verbose(env, "\nfrom %d to %d:",
 					prev_insn_idx, insn_idx);
-			print_verifier_state(env, state);
+			print_verifier_state(env, state->frame[state->curframe]);
 			do_print_state = false;
 		}
 
@@ -3967,13 +4554,17 @@ static int do_check(struct bpf_verifier_env *env)
 			if (opcode == BPF_CALL) {
 				if (BPF_SRC(insn->code) != BPF_K ||
 				    insn->off != 0 ||
-				    insn->src_reg != BPF_REG_0 ||
+				    (insn->src_reg != BPF_REG_0 &&
+				     insn->src_reg != BPF_PSEUDO_CALL) ||
 				    insn->dst_reg != BPF_REG_0) {
 					verbose(env, "BPF_CALL uses reserved fields\n");
 					return -EINVAL;
 				}
 
-				err = check_call(env, insn->imm, insn_idx);
+				if (insn->src_reg == BPF_PSEUDO_CALL)
+					err = check_func_call(env, insn, &insn_idx);
+				else
+					err = check_helper_call(env, insn->imm, insn_idx);
 				if (err)
 					return err;
 
@@ -3998,6 +4589,16 @@ static int do_check(struct bpf_verifier_env *env)
 					return -EINVAL;
 				}
 
+				if (state->curframe) {
+					/* exit from nested function */
+					prev_insn_idx = insn_idx;
+					err = prepare_func_exit(env, &insn_idx);
+					if (err)
+						return err;
+					do_print_state = true;
+					continue;
+				}
+
 				/* eBPF calling convetion is such that R0 is used
 				 * to return the value from eBPF program.
 				 * Make sure that it's readable at this time
@@ -4058,8 +4659,16 @@ process_bpf_exit:
 		insn_idx++;
 	}
 
-	verbose(env, "processed %d insns, stack depth %d\n", insn_processed,
-		env->prog->aux->stack_depth);
+	verbose(env, "processed %d insns, stack depth ", insn_processed);
+	for (i = 0; i < env->subprog_cnt + 1; i++) {
+		u32 depth = env->subprog_stack_depth[i];
+
+		verbose(env, "%d", depth);
+		if (i + 1 < env->subprog_cnt + 1)
+			verbose(env, "+");
+	}
+	verbose(env, "\n");
+	env->prog->aux->stack_depth = env->subprog_stack_depth[0];
 	return 0;
 }
 
@@ -4245,6 +4854,19 @@ static int adjust_insn_aux_data(struct bpf_verifier_env *env, u32 prog_len,
 	return 0;
 }
 
+static void adjust_subprog_starts(struct bpf_verifier_env *env, u32 off, u32 len)
+{
+	int i;
+
+	if (len == 1)
+		return;
+	for (i = 0; i < env->subprog_cnt; i++) {
+		if (env->subprog_starts[i] < off)
+			continue;
+		env->subprog_starts[i] += len - 1;
+	}
+}
+
 static struct bpf_prog *bpf_patch_insn_data(struct bpf_verifier_env *env, u32 off,
 					    const struct bpf_insn *patch, u32 len)
 {
@@ -4255,6 +4877,7 @@ static struct bpf_prog *bpf_patch_insn_data(struct bpf_verifier_env *env, u32 of
 		return NULL;
 	if (adjust_insn_aux_data(env, new_prog->len, off, len))
 		return NULL;
+	adjust_subprog_starts(env, off, len);
 	return new_prog;
 }
 
@@ -4389,6 +5012,150 @@ static int convert_ctx_accesses(struct bpf_verifier_env *env)
 	return 0;
 }
 
+static int jit_subprogs(struct bpf_verifier_env *env)
+{
+	struct bpf_prog *prog = env->prog, **func, *tmp;
+	int i, j, subprog_start, subprog_end = 0, len, subprog;
+	struct bpf_insn *insn = prog->insnsi;
+	void *old_bpf_func;
+	int err = -ENOMEM;
+
+	if (env->subprog_cnt == 0)
+		return 0;
+
+	for (i = 0; i < prog->len; i++, insn++) {
+		if (insn->code != (BPF_JMP | BPF_CALL) ||
+		    insn->src_reg != BPF_PSEUDO_CALL)
+			continue;
+		subprog = find_subprog(env, i + insn->imm + 1);
+		if (subprog < 0) {
+			WARN_ONCE(1, "verifier bug. No program starts at insn %d\n",
+				  i + insn->imm + 1);
+			return -EFAULT;
+		}
+		/* temporarily remember subprog id inside insn instead of
+		 * aux_data, since next loop will split up all insns into funcs
+		 */
+		insn->off = subprog + 1;
+		/* remember original imm in case JIT fails and fallback
+		 * to interpreter will be needed
+		 */
+		env->insn_aux_data[i].call_imm = insn->imm;
+		/* point imm to __bpf_call_base+1 from JITs point of view */
+		insn->imm = 1;
+	}
+
+	func = kzalloc(sizeof(prog) * (env->subprog_cnt + 1), GFP_KERNEL);
+	if (!func)
+		return -ENOMEM;
+
+	for (i = 0; i <= env->subprog_cnt; i++) {
+		subprog_start = subprog_end;
+		if (env->subprog_cnt == i)
+			subprog_end = prog->len;
+		else
+			subprog_end = env->subprog_starts[i];
+
+		len = subprog_end - subprog_start;
+		func[i] = bpf_prog_alloc(bpf_prog_size(len), GFP_USER);
+		if (!func[i])
+			goto out_free;
+		memcpy(func[i]->insnsi, &prog->insnsi[subprog_start],
+		       len * sizeof(struct bpf_insn));
+		func[i]->len = len;
+		func[i]->is_func = 1;
+		/* Use bpf_prog_F_tag to indicate functions in stack traces.
+		 * Long term would need debug info to populate names
+		 */
+		func[i]->aux->name[0] = 'F';
+		func[i]->aux->stack_depth = env->subprog_stack_depth[i];
+		func[i]->jit_requested = 1;
+		func[i] = bpf_int_jit_compile(func[i]);
+		if (!func[i]->jited) {
+			err = -ENOTSUPP;
+			goto out_free;
+		}
+		cond_resched();
+	}
+	/* at this point all bpf functions were successfully JITed
+	 * now populate all bpf_calls with correct addresses and
+	 * run last pass of JIT
+	 */
+	for (i = 0; i <= env->subprog_cnt; i++) {
+		insn = func[i]->insnsi;
+		for (j = 0; j < func[i]->len; j++, insn++) {
+			if (insn->code != (BPF_JMP | BPF_CALL) ||
+			    insn->src_reg != BPF_PSEUDO_CALL)
+				continue;
+			subprog = insn->off;
+			insn->off = 0;
+			insn->imm = (u64 (*)(u64, u64, u64, u64, u64))
+				func[subprog]->bpf_func -
+				__bpf_call_base;
+		}
+	}
+	for (i = 0; i <= env->subprog_cnt; i++) {
+		old_bpf_func = func[i]->bpf_func;
+		tmp = bpf_int_jit_compile(func[i]);
+		if (tmp != func[i] || func[i]->bpf_func != old_bpf_func) {
+			verbose(env, "JIT doesn't support bpf-to-bpf calls\n");
+			err = -EFAULT;
+			goto out_free;
+		}
+		cond_resched();
+	}
+
+	/* finally lock prog and jit images for all functions and
+	 * populate kallsysm
+	 */
+	for (i = 0; i <= env->subprog_cnt; i++) {
+		bpf_prog_lock_ro(func[i]);
+		bpf_prog_kallsyms_add(func[i]);
+	}
+	prog->jited = 1;
+	prog->bpf_func = func[0]->bpf_func;
+	prog->aux->func = func;
+	prog->aux->func_cnt = env->subprog_cnt + 1;
+	return 0;
+out_free:
+	for (i = 0; i <= env->subprog_cnt; i++)
+		if (func[i])
+			bpf_jit_free(func[i]);
+	kfree(func);
+	/* cleanup main prog to be interpreted */
+	prog->jit_requested = 0;
+	for (i = 0, insn = prog->insnsi; i < prog->len; i++, insn++) {
+		if (insn->code != (BPF_JMP | BPF_CALL) ||
+		    insn->src_reg != BPF_PSEUDO_CALL)
+			continue;
+		insn->off = 0;
+		insn->imm = env->insn_aux_data[i].call_imm;
+	}
+	return err;
+}
+
+static int fixup_call_args(struct bpf_verifier_env *env)
+{
+	struct bpf_prog *prog = env->prog;
+	struct bpf_insn *insn = prog->insnsi;
+	int i, depth;
+
+	if (env->prog->jit_requested)
+		if (jit_subprogs(env) == 0)
+			return 0;
+
+	for (i = 0; i < prog->len; i++, insn++) {
+		if (insn->code != (BPF_JMP | BPF_CALL) ||
+		    insn->src_reg != BPF_PSEUDO_CALL)
+			continue;
+		depth = get_callee_stack_depth(env, insn, i);
+		if (depth < 0)
+			return depth;
+		bpf_patch_call_args(insn, depth);
+	}
+	return 0;
+}
+
 /* fixup insn->imm field of bpf_call instructions
  * and inline eligible helpers as explicit sequence of BPF instructions
  *
@@ -4408,11 +5175,15 @@ static int fixup_bpf_calls(struct bpf_verifier_env *env)
 	for (i = 0; i < insn_cnt; i++, insn++) {
 		if (insn->code != (BPF_JMP | BPF_CALL))
 			continue;
+		if (insn->src_reg == BPF_PSEUDO_CALL)
+			continue;
 
 		if (insn->imm == BPF_FUNC_get_route_realm)
 			prog->dst_needed = 1;
 		if (insn->imm == BPF_FUNC_get_prandom_u32)
 			bpf_user_rnd_init_once();
+		if (insn->imm == BPF_FUNC_override_return)
+			prog->kprobe_override = 1;
 		if (insn->imm == BPF_FUNC_tail_call) {
 			/* If we tail call into other programs, we
 			 * cannot make any assumptions since they can
@@ -4435,7 +5206,7 @@ static int fixup_bpf_calls(struct bpf_verifier_env *env)
 		/* BPF_EMIT_CALL() assumptions in some of the map_gen_lookup
 		 * handlers are currently limited to 64 bit only.
 		 */
-		if (ebpf_jit_enabled() && BITS_PER_LONG == 64 &&
+		if (prog->jit_requested && BITS_PER_LONG == 64 &&
 		    insn->imm == BPF_FUNC_map_lookup_elem) {
 			map_ptr = env->insn_aux_data[i + delta].map_ptr;
 			if (map_ptr == BPF_MAP_PTR_POISON ||
@@ -4587,12 +5358,12 @@ int bpf_check(struct bpf_prog **prog, union bpf_attr *attr)
 	if (!env->explored_states)
 		goto skip_full_check;
 
+	env->allow_ptr_leaks = capable(CAP_SYS_ADMIN);
+
 	ret = check_cfg(env);
 	if (ret < 0)
 		goto skip_full_check;
 
-	env->allow_ptr_leaks = capable(CAP_SYS_ADMIN);
-
 	ret = do_check(env);
 	if (env->cur_state) {
 		free_verifier_state(env->cur_state, true);
@@ -4613,6 +5384,9 @@ skip_full_check:
 	if (ret == 0)
 		ret = fixup_bpf_calls(env);
 
+	if (ret == 0)
+		ret = fixup_call_args(env);
+
 	if (log->level && bpf_verifier_log_full(log))
 		ret = -ENOSPC;
 	if (log->level && !log->ubuf) {
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 4df5b695bf0d..878d86c513d6 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -4723,6 +4723,9 @@ static long _perf_ioctl(struct perf_event *event, unsigned int cmd, unsigned lon
 		rcu_read_unlock();
 		return 0;
 	}
+
+	case PERF_EVENT_IOC_QUERY_BPF:
+		return perf_event_query_prog_array(event, (void __user *)arg);
 	default:
 		return -ENOTTY;
 	}
@@ -8080,6 +8083,13 @@ static int perf_event_set_bpf_prog(struct perf_event *event, u32 prog_fd)
 		return -EINVAL;
 	}
 
+	/* Kprobe override only works for kprobes, not uprobes. */
+	if (prog->kprobe_override &&
+	    !(event->tp_event->flags & TRACE_EVENT_FL_KPROBE)) {
+		bpf_prog_put(prog);
+		return -EINVAL;
+	}
+
 	if (is_tracepoint || is_syscall_tp) {
 		int off = trace_event_get_offsets(event->tp_event);
 
diff --git a/kernel/kprobes.c b/kernel/kprobes.c
index da2ccf142358..b4aab48ad258 100644
--- a/kernel/kprobes.c
+++ b/kernel/kprobes.c
@@ -83,6 +83,16 @@ static raw_spinlock_t *kretprobe_table_lock_ptr(unsigned long hash)
 	return &(kretprobe_table_locks[hash].lock);
 }
 
+/* List of symbols that can be overriden for error injection. */
+static LIST_HEAD(kprobe_error_injection_list);
+static DEFINE_MUTEX(kprobe_ei_mutex);
+struct kprobe_ei_entry {
+	struct list_head list;
+	unsigned long start_addr;
+	unsigned long end_addr;
+	void *priv;
+};
+
 /* Blacklist -- list of struct kprobe_blacklist_entry */
 static LIST_HEAD(kprobe_blacklist);
 
@@ -1394,6 +1404,17 @@ bool within_kprobe_blacklist(unsigned long addr)
 	return false;
 }
 
+bool within_kprobe_error_injection_list(unsigned long addr)
+{
+	struct kprobe_ei_entry *ent;
+
+	list_for_each_entry(ent, &kprobe_error_injection_list, list) {
+		if (addr >= ent->start_addr && addr < ent->end_addr)
+			return true;
+	}
+	return false;
+}
+
 /*
  * If we have a symbol_name argument, look it up and add the offset field
  * to it. This way, we can specify a relative address to a symbol.
@@ -2168,6 +2189,86 @@ static int __init populate_kprobe_blacklist(unsigned long *start,
 	return 0;
 }
 
+#ifdef CONFIG_BPF_KPROBE_OVERRIDE
+/* Markers of the _kprobe_error_inject_list section */
+extern unsigned long __start_kprobe_error_inject_list[];
+extern unsigned long __stop_kprobe_error_inject_list[];
+
+/*
+ * Lookup and populate the kprobe_error_injection_list.
+ *
+ * For safety reasons we only allow certain functions to be overriden with
+ * bpf_error_injection, so we need to populate the list of the symbols that have
+ * been marked as safe for overriding.
+ */
+static void populate_kprobe_error_injection_list(unsigned long *start,
+						 unsigned long *end,
+						 void *priv)
+{
+	unsigned long *iter;
+	struct kprobe_ei_entry *ent;
+	unsigned long entry, offset = 0, size = 0;
+
+	mutex_lock(&kprobe_ei_mutex);
+	for (iter = start; iter < end; iter++) {
+		entry = arch_deref_entry_point((void *)*iter);
+
+		if (!kernel_text_address(entry) ||
+		    !kallsyms_lookup_size_offset(entry, &size, &offset)) {
+			pr_err("Failed to find error inject entry at %p\n",
+				(void *)entry);
+			continue;
+		}
+
+		ent = kmalloc(sizeof(*ent), GFP_KERNEL);
+		if (!ent)
+			break;
+		ent->start_addr = entry;
+		ent->end_addr = entry + size;
+		ent->priv = priv;
+		INIT_LIST_HEAD(&ent->list);
+		list_add_tail(&ent->list, &kprobe_error_injection_list);
+	}
+	mutex_unlock(&kprobe_ei_mutex);
+}
+
+static void __init populate_kernel_kprobe_ei_list(void)
+{
+	populate_kprobe_error_injection_list(__start_kprobe_error_inject_list,
+					     __stop_kprobe_error_inject_list,
+					     NULL);
+}
+
+static void module_load_kprobe_ei_list(struct module *mod)
+{
+	if (!mod->num_kprobe_ei_funcs)
+		return;
+	populate_kprobe_error_injection_list(mod->kprobe_ei_funcs,
+					     mod->kprobe_ei_funcs +
+					     mod->num_kprobe_ei_funcs, mod);
+}
+
+static void module_unload_kprobe_ei_list(struct module *mod)
+{
+	struct kprobe_ei_entry *ent, *n;
+	if (!mod->num_kprobe_ei_funcs)
+		return;
+
+	mutex_lock(&kprobe_ei_mutex);
+	list_for_each_entry_safe(ent, n, &kprobe_error_injection_list, list) {
+		if (ent->priv == mod) {
+			list_del_init(&ent->list);
+			kfree(ent);
+		}
+	}
+	mutex_unlock(&kprobe_ei_mutex);
+}
+#else
+static inline void __init populate_kernel_kprobe_ei_list(void) {}
+static inline void module_load_kprobe_ei_list(struct module *m) {}
+static inline void module_unload_kprobe_ei_list(struct module *m) {}
+#endif
+
 /* Module notifier call back, checking kprobes on the module */
 static int kprobes_module_callback(struct notifier_block *nb,
 				   unsigned long val, void *data)
@@ -2178,6 +2279,11 @@ static int kprobes_module_callback(struct notifier_block *nb,
 	unsigned int i;
 	int checkcore = (val == MODULE_STATE_GOING);
 
+	if (val == MODULE_STATE_COMING)
+		module_load_kprobe_ei_list(mod);
+	else if (val == MODULE_STATE_GOING)
+		module_unload_kprobe_ei_list(mod);
+
 	if (val != MODULE_STATE_GOING && val != MODULE_STATE_LIVE)
 		return NOTIFY_DONE;
 
@@ -2240,6 +2346,8 @@ static int __init init_kprobes(void)
 		pr_err("Please take care of using kprobes.\n");
 	}
 
+	populate_kernel_kprobe_ei_list();
+
 	if (kretprobe_blacklist_size) {
 		/* lookup the function address from its name */
 		for (i = 0; kretprobe_blacklist[i].name != NULL; i++) {
@@ -2407,6 +2515,56 @@ static const struct file_operations debugfs_kprobe_blacklist_ops = {
 	.release        = seq_release,
 };
 
+/*
+ * kprobes/error_injection_list -- shows which functions can be overriden for
+ * error injection.
+ * */
+static void *kprobe_ei_seq_start(struct seq_file *m, loff_t *pos)
+{
+	mutex_lock(&kprobe_ei_mutex);
+	return seq_list_start(&kprobe_error_injection_list, *pos);
+}
+
+static void kprobe_ei_seq_stop(struct seq_file *m, void *v)
+{
+	mutex_unlock(&kprobe_ei_mutex);
+}
+
+static void *kprobe_ei_seq_next(struct seq_file *m, void *v, loff_t *pos)
+{
+	return seq_list_next(v, &kprobe_error_injection_list, pos);
+}
+
+static int kprobe_ei_seq_show(struct seq_file *m, void *v)
+{
+	char buffer[KSYM_SYMBOL_LEN];
+	struct kprobe_ei_entry *ent =
+		list_entry(v, struct kprobe_ei_entry, list);
+
+	sprint_symbol(buffer, ent->start_addr);
+	seq_printf(m, "%s\n", buffer);
+	return 0;
+}
+
+static const struct seq_operations kprobe_ei_seq_ops = {
+	.start = kprobe_ei_seq_start,
+	.next  = kprobe_ei_seq_next,
+	.stop  = kprobe_ei_seq_stop,
+	.show  = kprobe_ei_seq_show,
+};
+
+static int kprobe_ei_open(struct inode *inode, struct file *filp)
+{
+	return seq_open(filp, &kprobe_ei_seq_ops);
+}
+
+static const struct file_operations debugfs_kprobe_ei_ops = {
+	.open           = kprobe_ei_open,
+	.read           = seq_read,
+	.llseek         = seq_lseek,
+	.release        = seq_release,
+};
+
 static void arm_all_kprobes(void)
 {
 	struct hlist_head *head;
@@ -2548,6 +2706,11 @@ static int __init debugfs_kprobe_init(void)
 	if (!file)
 		goto error;
 
+	file = debugfs_create_file("error_injection_list", 0444, dir, NULL,
+				  &debugfs_kprobe_ei_ops);
+	if (!file)
+		goto error;
+
 	return 0;
 
 error:
diff --git a/kernel/module.c b/kernel/module.c
index dea01ac9cb74..bd695bfdc5c4 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -3118,7 +3118,11 @@ static int find_module_sections(struct module *mod, struct load_info *info)
 					     sizeof(*mod->ftrace_callsites),
 					     &mod->num_ftrace_callsites);
 #endif
-
+#ifdef CONFIG_BPF_KPROBE_OVERRIDE
+	mod->kprobe_ei_funcs = section_objs(info, "_kprobe_error_inject_list",
+					    sizeof(*mod->kprobe_ei_funcs),
+					    &mod->num_kprobe_ei_funcs);
+#endif
 	mod->extable = section_objs(info, "__ex_table",
 				    sizeof(*mod->extable), &mod->num_exentries);
 
diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
index 904c952ac383..ae3a2d519e50 100644
--- a/kernel/trace/Kconfig
+++ b/kernel/trace/Kconfig
@@ -530,6 +530,17 @@ config FUNCTION_PROFILER
 
 	  If in doubt, say N.
 
+config BPF_KPROBE_OVERRIDE
+	bool "Enable BPF programs to override a kprobed function"
+	depends on BPF_EVENTS
+	depends on KPROBES_ON_FTRACE
+	depends on HAVE_KPROBE_OVERRIDE
+	depends on DYNAMIC_FTRACE_WITH_REGS
+	default n
+	help
+	 Allows BPF to override the execution of a probed function and
+	 set a different return value.  This is used for error injection.
+
 config FTRACE_MCOUNT_RECORD
 	def_bool y
 	depends on DYNAMIC_FTRACE
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index 40207c2a4113..f6d2327ecb59 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -13,6 +13,10 @@
 #include <linux/filter.h>
 #include <linux/uaccess.h>
 #include <linux/ctype.h>
+#include <linux/kprobes.h>
+#include <asm/kprobes.h>
+
+#include "trace_probe.h"
 #include "trace.h"
 
 u64 bpf_get_stackid(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5);
@@ -76,6 +80,24 @@ unsigned int trace_call_bpf(struct trace_event_call *call, void *ctx)
 }
 EXPORT_SYMBOL_GPL(trace_call_bpf);
 
+#ifdef CONFIG_BPF_KPROBE_OVERRIDE
+BPF_CALL_2(bpf_override_return, struct pt_regs *, regs, unsigned long, rc)
+{
+	__this_cpu_write(bpf_kprobe_override, 1);
+	regs_set_return_value(regs, rc);
+	arch_ftrace_kprobe_override_function(regs);
+	return 0;
+}
+
+static const struct bpf_func_proto bpf_override_return_proto = {
+	.func		= bpf_override_return,
+	.gpl_only	= true,
+	.ret_type	= RET_INTEGER,
+	.arg1_type	= ARG_PTR_TO_CTX,
+	.arg2_type	= ARG_ANYTHING,
+};
+#endif
+
 BPF_CALL_3(bpf_probe_read, void *, dst, u32, size, const void *, unsafe_ptr)
 {
 	int ret;
@@ -556,6 +578,10 @@ static const struct bpf_func_proto *kprobe_prog_func_proto(enum bpf_func_id func
 		return &bpf_get_stackid_proto;
 	case BPF_FUNC_perf_event_read_value:
 		return &bpf_perf_event_read_value_proto;
+#ifdef CONFIG_BPF_KPROBE_OVERRIDE
+	case BPF_FUNC_override_return:
+		return &bpf_override_return_proto;
+#endif
 	default:
 		return tracing_func_proto(func_id);
 	}
@@ -773,6 +799,15 @@ int perf_event_attach_bpf_prog(struct perf_event *event,
 	struct bpf_prog_array *new_array;
 	int ret = -EEXIST;
 
+	/*
+	 * Kprobe override only works for ftrace based kprobes, and only if they
+	 * are on the opt-in list.
+	 */
+	if (prog->kprobe_override &&
+	    (!trace_kprobe_ftrace(event->tp_event) ||
+	     !trace_kprobe_error_injectable(event->tp_event)))
+		return -EINVAL;
+
 	mutex_lock(&bpf_event_mutex);
 
 	if (event->prog)
@@ -825,3 +860,26 @@ void perf_event_detach_bpf_prog(struct perf_event *event)
 unlock:
 	mutex_unlock(&bpf_event_mutex);
 }
+
+int perf_event_query_prog_array(struct perf_event *event, void __user *info)
+{
+	struct perf_event_query_bpf __user *uquery = info;
+	struct perf_event_query_bpf query = {};
+	int ret;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+	if (event->attr.type != PERF_TYPE_TRACEPOINT)
+		return -EINVAL;
+	if (copy_from_user(&query, uquery, sizeof(query)))
+		return -EFAULT;
+
+	mutex_lock(&bpf_event_mutex);
+	ret = bpf_prog_array_copy_info(event->tp_event->prog_array,
+				       uquery->ids,
+				       query.ids_len,
+				       &uquery->prog_cnt);
+	mutex_unlock(&bpf_event_mutex);
+
+	return ret;
+}
diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index 492700c5fb4d..91f4b57dab82 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -42,6 +42,7 @@ struct trace_kprobe {
 	(offsetof(struct trace_kprobe, tp.args) +	\
 	(sizeof(struct probe_arg) * (n)))
 
+DEFINE_PER_CPU(int, bpf_kprobe_override);
 
 static nokprobe_inline bool trace_kprobe_is_return(struct trace_kprobe *tk)
 {
@@ -87,6 +88,27 @@ static nokprobe_inline unsigned long trace_kprobe_nhit(struct trace_kprobe *tk)
 	return nhit;
 }
 
+int trace_kprobe_ftrace(struct trace_event_call *call)
+{
+	struct trace_kprobe *tk = (struct trace_kprobe *)call->data;
+	return kprobe_ftrace(&tk->rp.kp);
+}
+
+int trace_kprobe_error_injectable(struct trace_event_call *call)
+{
+	struct trace_kprobe *tk = (struct trace_kprobe *)call->data;
+	unsigned long addr;
+
+	if (tk->symbol) {
+		addr = (unsigned long)
+			kallsyms_lookup_name(trace_kprobe_symbol(tk));
+		addr += tk->rp.kp.offset;
+	} else {
+		addr = (unsigned long)tk->rp.kp.addr;
+	}
+	return within_kprobe_error_injection_list(addr);
+}
+
 static int register_kprobe_event(struct trace_kprobe *tk);
 static int unregister_kprobe_event(struct trace_kprobe *tk);
 
@@ -1170,7 +1192,7 @@ static int kretprobe_event_define_fields(struct trace_event_call *event_call)
 #ifdef CONFIG_PERF_EVENTS
 
 /* Kprobe profile handler */
-static void
+static int
 kprobe_perf_func(struct trace_kprobe *tk, struct pt_regs *regs)
 {
 	struct trace_event_call *call = &tk->tp.call;
@@ -1179,12 +1201,29 @@ kprobe_perf_func(struct trace_kprobe *tk, struct pt_regs *regs)
 	int size, __size, dsize;
 	int rctx;
 
-	if (bpf_prog_array_valid(call) && !trace_call_bpf(call, regs))
-		return;
+	if (bpf_prog_array_valid(call)) {
+		int ret;
+
+		ret = trace_call_bpf(call, regs);
+
+		/*
+		 * We need to check and see if we modified the pc of the
+		 * pt_regs, and if so clear the kprobe and return 1 so that we
+		 * don't do the instruction skipping.  Also reset our state so
+		 * we are clean the next pass through.
+		 */
+		if (__this_cpu_read(bpf_kprobe_override)) {
+			__this_cpu_write(bpf_kprobe_override, 0);
+			reset_current_kprobe();
+			return 1;
+		}
+		if (!ret)
+			return 0;
+	}
 
 	head = this_cpu_ptr(call->perf_events);
 	if (hlist_empty(head))
-		return;
+		return 0;
 
 	dsize = __get_data_size(&tk->tp, regs);
 	__size = sizeof(*entry) + tk->tp.size + dsize;
@@ -1193,13 +1232,14 @@ kprobe_perf_func(struct trace_kprobe *tk, struct pt_regs *regs)
 
 	entry = perf_trace_buf_alloc(size, NULL, &rctx);
 	if (!entry)
-		return;
+		return 0;
 
 	entry->ip = (unsigned long)tk->rp.kp.addr;
 	memset(&entry[1], 0, dsize);
 	store_trace_args(sizeof(*entry), &tk->tp, regs, (u8 *)&entry[1], dsize);
 	perf_trace_buf_submit(entry, size, rctx, call->event.type, 1, regs,
 			      head, NULL);
+	return 0;
 }
 NOKPROBE_SYMBOL(kprobe_perf_func);
 
@@ -1275,16 +1315,24 @@ static int kprobe_register(struct trace_event_call *event,
 static int kprobe_dispatcher(struct kprobe *kp, struct pt_regs *regs)
 {
 	struct trace_kprobe *tk = container_of(kp, struct trace_kprobe, rp.kp);
+	int ret = 0;
 
 	raw_cpu_inc(*tk->nhit);
 
 	if (tk->tp.flags & TP_FLAG_TRACE)
 		kprobe_trace_func(tk, regs);
 #ifdef CONFIG_PERF_EVENTS
-	if (tk->tp.flags & TP_FLAG_PROFILE)
-		kprobe_perf_func(tk, regs);
+	if (tk->tp.flags & TP_FLAG_PROFILE) {
+		ret = kprobe_perf_func(tk, regs);
+		/*
+		 * The ftrace kprobe handler leaves it up to us to re-enable
+		 * preemption here before returning if we've modified the ip.
+		 */
+		if (ret)
+			preempt_enable_no_resched();
+	}
 #endif
-	return 0;	/* We don't tweek kernel, so just return 0 */
+	return ret;
 }
 NOKPROBE_SYMBOL(kprobe_dispatcher);
 
diff --git a/kernel/trace/trace_probe.h b/kernel/trace/trace_probe.h
index fb66e3eaa192..5e54d748c84c 100644
--- a/kernel/trace/trace_probe.h
+++ b/kernel/trace/trace_probe.h
@@ -252,6 +252,8 @@ struct symbol_cache;
 unsigned long update_symbol_cache(struct symbol_cache *sc);
 void free_symbol_cache(struct symbol_cache *sc);
 struct symbol_cache *alloc_symbol_cache(const char *sym, long offset);
+int trace_kprobe_ftrace(struct trace_event_call *call);
+int trace_kprobe_error_injectable(struct trace_event_call *call);
 #else
 /* uprobes do not support symbol fetch methods */
 #define fetch_symbol_u8			NULL
@@ -277,6 +279,16 @@ alloc_symbol_cache(const char *sym, long offset)
 {
 	return NULL;
 }
+
+static inline int trace_kprobe_ftrace(struct trace_event_call *call)
+{
+	return 0;
+}
+
+static inline int trace_kprobe_error_injectable(struct trace_event_call *call)
+{
+	return 0;
+}
 #endif /* CONFIG_KPROBE_EVENTS */
 
 struct probe_arg {