Merge branch 'bpf-unprivileged'

Alexei Starovoitov says:

====================
bpf: unprivileged

v1-v2:
- this set logically depends on cb patch
  "bpf: fix cb access in socket filter programs":
  http://patchwork.ozlabs.org/patch/527391/
  which is must have to allow unprivileged programs.
  Thanks Daniel for finding that issue.
- refactored sysctl to be similar to 'modules_disabled'
- dropped bpf_trace_printk
- split tests into separate patch and added more tests
  based on discussion

v1 cover letter:
I think it is time to liberate eBPF from CAP_SYS_ADMIN.
As was discussed when eBPF was first introduced two years ago
the only piece missing in eBPF verifier is 'pointer leak detection'
to make it available to non-root users.
Patch 1 adds this pointer analysis.
The eBPF programs, obviously, need to see and operate on kernel addresses,
but with these extra checks they won't be able to pass these addresses
to user space.
Patch 2 adds accounting of kernel memory used by programs and maps.
It changes behavoir for existing root users, but I think it needs
to be done consistently for both root and non-root, since today
programs and maps are only limited by number of open FDs (RLIMIT_NOFILE).
Patch 2 accounts program's and map's kernel memory as RLIMIT_MEMLOCK.

Unprivileged eBPF is only meaningful for 'socket filter'-like programs.
eBPF programs for tracing and TC classifiers/actions will stay root only.

In parallel the bpf fuzzing effort is ongoing and so far
we've found only one verifier bug and that was already fixed.
The 'constant blinding' pass also being worked on.
It will obfuscate constant-like values that are part of eBPF ISA
to make jit spraying attacks even harder.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/bpf.h

@@ -36,6 +36,8 @@ struct bpf_map {
 	u32 key_size;
 	u32 value_size;
 	u32 max_entries;
+	u32 pages;
+	struct user_struct *user;
 	const struct bpf_map_ops *ops;
 	struct work_struct work;
 };
@@ -128,6 +130,7 @@ struct bpf_prog_aux {
 	const struct bpf_verifier_ops *ops;
 	struct bpf_map **used_maps;
 	struct bpf_prog *prog;
+	struct user_struct *user;
 	union {
 		struct work_struct work;
 		struct rcu_head	rcu;
@@ -167,6 +170,8 @@ void bpf_prog_put_rcu(struct bpf_prog *prog);
 struct bpf_map *bpf_map_get(struct fd f);
 void bpf_map_put(struct bpf_map *map);
 
+extern int sysctl_unprivileged_bpf_disabled;
+
 /* verify correctness of eBPF program */
 int bpf_check(struct bpf_prog **fp, union bpf_attr *attr);
 #else

include/linux/sched.h

@@ -840,7 +840,7 @@ struct user_struct {
 	struct hlist_node uidhash_node;
 	kuid_t uid;
 
-#ifdef CONFIG_PERF_EVENTS
+#if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
 	atomic_long_t locked_vm;
 #endif
 };

kernel/bpf/arraymap.c

@@ -49,7 +49,7 @@ static struct bpf_map *array_map_alloc(union bpf_attr *attr)
 	array->map.key_size = attr->key_size;
 	array->map.value_size = attr->value_size;
 	array->map.max_entries = attr->max_entries;
-
+	array->map.pages = round_up(array_size, PAGE_SIZE) >> PAGE_SHIFT;
 	array->elem_size = elem_size;
 
 	return &array->map;

kernel/bpf/hashtab.c

@@ -88,6 +88,10 @@ static struct bpf_map *htab_map_alloc(union bpf_attr *attr)
 	htab->elem_size = sizeof(struct htab_elem) +
 			  round_up(htab->map.key_size, 8) +
 			  htab->map.value_size;
+
+	htab->map.pages = round_up(htab->n_buckets * sizeof(struct hlist_head) +
+				   htab->elem_size * htab->map.max_entries,
+				   PAGE_SIZE) >> PAGE_SHIFT;
 	return &htab->map;
 
 free_htab:

kernel/bpf/syscall.c

@@ -18,6 +18,8 @@
 #include <linux/filter.h>
 #include <linux/version.h>
 
+int sysctl_unprivileged_bpf_disabled __read_mostly;
+
 static LIST_HEAD(bpf_map_types);
 
 static struct bpf_map *find_and_alloc_map(union bpf_attr *attr)
@@ -44,11 +46,38 @@ void bpf_register_map_type(struct bpf_map_type_list *tl)
 	list_add(&tl->list_node, &bpf_map_types);
 }
 
+static int bpf_map_charge_memlock(struct bpf_map *map)
+{
+	struct user_struct *user = get_current_user();
+	unsigned long memlock_limit;
+
+	memlock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
+
+	atomic_long_add(map->pages, &user->locked_vm);
+
+	if (atomic_long_read(&user->locked_vm) > memlock_limit) {
+		atomic_long_sub(map->pages, &user->locked_vm);
+		free_uid(user);
+		return -EPERM;
+	}
+	map->user = user;
+	return 0;
+}
+
+static void bpf_map_uncharge_memlock(struct bpf_map *map)
+{
+	struct user_struct *user = map->user;
+
+	atomic_long_sub(map->pages, &user->locked_vm);
+	free_uid(user);
+}
+
 /* called from workqueue */
 static void bpf_map_free_deferred(struct work_struct *work)
 {
 	struct bpf_map *map = container_of(work, struct bpf_map, work);
 
+	bpf_map_uncharge_memlock(map);
 	/* implementation dependent freeing */
 	map->ops->map_free(map);
 }
@@ -108,6 +137,10 @@ static int map_create(union bpf_attr *attr)
 
 	atomic_set(&map->refcnt, 1);
 
+	err = bpf_map_charge_memlock(map);
+	if (err)
+		goto free_map;
+
 	err = anon_inode_getfd("bpf-map", &bpf_map_fops, map, O_RDWR | O_CLOEXEC);
 
 	if (err < 0)
@@ -440,11 +473,37 @@ static void free_used_maps(struct bpf_prog_aux *aux)
 	kfree(aux->used_maps);
 }
 
+static int bpf_prog_charge_memlock(struct bpf_prog *prog)
+{
+	struct user_struct *user = get_current_user();
+	unsigned long memlock_limit;
+
+	memlock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
+
+	atomic_long_add(prog->pages, &user->locked_vm);
+	if (atomic_long_read(&user->locked_vm) > memlock_limit) {
+		atomic_long_sub(prog->pages, &user->locked_vm);
+		free_uid(user);
+		return -EPERM;
+	}
+	prog->aux->user = user;
+	return 0;
+}
+
+static void bpf_prog_uncharge_memlock(struct bpf_prog *prog)
+{
+	struct user_struct *user = prog->aux->user;
+
+	atomic_long_sub(prog->pages, &user->locked_vm);
+	free_uid(user);
+}
+
 static void __prog_put_rcu(struct rcu_head *rcu)
 {
 	struct bpf_prog_aux *aux = container_of(rcu, struct bpf_prog_aux, rcu);
 
 	free_used_maps(aux);
+	bpf_prog_uncharge_memlock(aux->prog);
 	bpf_prog_free(aux->prog);
 }
 
@@ -544,11 +603,18 @@ static int bpf_prog_load(union bpf_attr *attr)
 	    attr->kern_version != LINUX_VERSION_CODE)
 		return -EINVAL;
 
+	if (type != BPF_PROG_TYPE_SOCKET_FILTER && !capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
 	/* plain bpf_prog allocation */
 	prog = bpf_prog_alloc(bpf_prog_size(attr->insn_cnt), GFP_USER);
 	if (!prog)
 		return -ENOMEM;
 
+	err = bpf_prog_charge_memlock(prog);
+	if (err)
+		goto free_prog_nouncharge;
+
 	prog->len = attr->insn_cnt;
 
 	err = -EFAULT;
@@ -590,6 +656,8 @@ static int bpf_prog_load(union bpf_attr *attr)
 free_used_maps:
 	free_used_maps(prog->aux);
 free_prog:
+	bpf_prog_uncharge_memlock(prog);
+free_prog_nouncharge:
 	bpf_prog_free(prog);
 	return err;
 }
@@ -599,11 +667,7 @@ SYSCALL_DEFINE3(bpf, int, cmd, union bpf_attr __user *, uattr, unsigned int, siz
 	union bpf_attr attr = {};
 	int err;
 
-	/* the syscall is limited to root temporarily. This restriction will be
-	 * lifted when security audit is clean. Note that eBPF+tracing must have
-	 * this restriction, since it may pass kernel data to user space
-	 */
-	if (!capable(CAP_SYS_ADMIN))
+	if (!capable(CAP_SYS_ADMIN) && sysctl_unprivileged_bpf_disabled)
 		return -EPERM;
 
 	if (!access_ok(VERIFY_READ, uattr, 1))

kernel/bpf/verifier.c

@@ -199,6 +199,7 @@ struct verifier_env {
 	struct verifier_state_list **explored_states; /* search pruning optimization */
 	struct bpf_map *used_maps[MAX_USED_MAPS]; /* array of map's used by eBPF program */
 	u32 used_map_cnt;		/* number of used maps */
+	bool allow_ptr_leaks;
 };
 
 /* verbose verifier prints what it's seeing
@@ -538,6 +539,21 @@ static int bpf_size_to_bytes(int bpf_size)
 		return -EINVAL;
 }
 
+static bool is_spillable_regtype(enum bpf_reg_type type)
+{
+	switch (type) {
+	case PTR_TO_MAP_VALUE:
+	case PTR_TO_MAP_VALUE_OR_NULL:
+	case PTR_TO_STACK:
+	case PTR_TO_CTX:
+	case FRAME_PTR:
+	case CONST_PTR_TO_MAP:
+		return true;
+	default:
+		return false;
+	}
+}
+
 /* check_stack_read/write functions track spill/fill of registers,
  * stack boundary and alignment are checked in check_mem_access()
  */
@@ -550,9 +566,7 @@ static int check_stack_write(struct verifier_state *state, int off, int size,
 	 */
 
 	if (value_regno >= 0 &&
-	    (state->regs[value_regno].type == PTR_TO_MAP_VALUE ||
-	     state->regs[value_regno].type == PTR_TO_STACK ||
-	     state->regs[value_regno].type == PTR_TO_CTX)) {
+	    is_spillable_regtype(state->regs[value_regno].type)) {
 
 		/* register containing pointer is being spilled into stack */
 		if (size != BPF_REG_SIZE) {
@@ -643,6 +657,20 @@ static int check_ctx_access(struct verifier_env *env, int off, int size,
 	return -EACCES;
 }
 
+static bool is_pointer_value(struct verifier_env *env, int regno)
+{
+	if (env->allow_ptr_leaks)
+		return false;
+
+	switch (env->cur_state.regs[regno].type) {
+	case UNKNOWN_VALUE:
+	case CONST_IMM:
+		return false;
+	default:
+		return true;
+	}
+}
+
 /* 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
@@ -669,11 +697,21 @@ static int check_mem_access(struct verifier_env *env, u32 regno, int off,
 	}
 
 	if (state->regs[regno].type == PTR_TO_MAP_VALUE) {
+		if (t == BPF_WRITE && value_regno >= 0 &&
+		    is_pointer_value(env, value_regno)) {
+			verbose("R%d leaks addr into map\n", value_regno);
+			return -EACCES;
+		}
 		err = check_map_access(env, regno, off, size);
 		if (!err && t == BPF_READ && value_regno >= 0)
 			mark_reg_unknown_value(state->regs, value_regno);
 
 	} else if (state->regs[regno].type == PTR_TO_CTX) {
+		if (t == BPF_WRITE && value_regno >= 0 &&
+		    is_pointer_value(env, value_regno)) {
+			verbose("R%d leaks addr into ctx\n", value_regno);
+			return -EACCES;
+		}
 		err = check_ctx_access(env, off, size, t);
 		if (!err && t == BPF_READ && value_regno >= 0)
 			mark_reg_unknown_value(state->regs, value_regno);
@@ -684,10 +722,17 @@ static int check_mem_access(struct verifier_env *env, u32 regno, int off,
 			verbose("invalid stack off=%d size=%d\n", off, size);
 			return -EACCES;
 		}
-		if (t == BPF_WRITE)
+		if (t == BPF_WRITE) {
+			if (!env->allow_ptr_leaks &&
+			    state->stack_slot_type[MAX_BPF_STACK + off] == STACK_SPILL &&
+			    size != BPF_REG_SIZE) {
+				verbose("attempt to corrupt spilled pointer on stack\n");
+				return -EACCES;
+			}
 			err = check_stack_write(state, off, size, value_regno);
-		else
+		} else {
 			err = check_stack_read(state, off, size, value_regno);
+		}
 	} else {
 		verbose("R%d invalid mem access '%s'\n",
 			regno, reg_type_str[state->regs[regno].type]);
@@ -775,8 +820,13 @@ static int check_func_arg(struct verifier_env *env, u32 regno,
 		return -EACCES;
 	}
 
-	if (arg_type == ARG_ANYTHING)
+	if (arg_type == ARG_ANYTHING) {
+		if (is_pointer_value(env, regno)) {
+			verbose("R%d leaks addr into helper function\n", regno);
+			return -EACCES;
+		}
 		return 0;
+	}
 
 	if (arg_type == ARG_PTR_TO_STACK || arg_type == ARG_PTR_TO_MAP_KEY ||
 	    arg_type == ARG_PTR_TO_MAP_VALUE) {
@@ -950,8 +1000,9 @@ static int check_call(struct verifier_env *env, int func_id)
 }
 
 /* check validity of 32-bit and 64-bit arithmetic operations */
-static int check_alu_op(struct reg_state *regs, struct bpf_insn *insn)
+static int check_alu_op(struct verifier_env *env, struct bpf_insn *insn)
 {
+	struct reg_state *regs = env->cur_state.regs;
 	u8 opcode = BPF_OP(insn->code);
 	int err;
 
@@ -976,6 +1027,12 @@ static int check_alu_op(struct reg_state *regs, struct bpf_insn *insn)
 		if (err)
 			return err;
 
+		if (is_pointer_value(env, insn->dst_reg)) {
+			verbose("R%d pointer arithmetic prohibited\n",
+				insn->dst_reg);
+			return -EACCES;
+		}
+
 		/* check dest operand */
 		err = check_reg_arg(regs, insn->dst_reg, DST_OP);
 		if (err)
@@ -1012,6 +1069,11 @@ static int check_alu_op(struct reg_state *regs, struct bpf_insn *insn)
 				 */
 				regs[insn->dst_reg] = regs[insn->src_reg];
 			} else {
+				if (is_pointer_value(env, insn->src_reg)) {
+					verbose("R%d partial copy of pointer\n",
+						insn->src_reg);
+					return -EACCES;
+				}
 				regs[insn->dst_reg].type = UNKNOWN_VALUE;
 				regs[insn->dst_reg].map_ptr = NULL;
 			}
@@ -1061,8 +1123,18 @@ static int check_alu_op(struct reg_state *regs, struct bpf_insn *insn)
 		/* pattern match 'bpf_add Rx, imm' instruction */
 		if (opcode == BPF_ADD && BPF_CLASS(insn->code) == BPF_ALU64 &&
 		    regs[insn->dst_reg].type == FRAME_PTR &&
-		    BPF_SRC(insn->code) == BPF_K)
+		    BPF_SRC(insn->code) == BPF_K) {
 			stack_relative = true;
+		} else if (is_pointer_value(env, insn->dst_reg)) {
+			verbose("R%d pointer arithmetic prohibited\n",
+				insn->dst_reg);
+			return -EACCES;
+		} else if (BPF_SRC(insn->code) == BPF_X &&
+			   is_pointer_value(env, insn->src_reg)) {
+			verbose("R%d pointer arithmetic prohibited\n",
+				insn->src_reg);
+			return -EACCES;
+		}
 
 		/* check dest operand */
 		err = check_reg_arg(regs, insn->dst_reg, DST_OP);
@@ -1101,6 +1173,12 @@ static int check_cond_jmp_op(struct verifier_env *env,
 		err = check_reg_arg(regs, insn->src_reg, SRC_OP);
 		if (err)
 			return err;
+
+		if (is_pointer_value(env, insn->src_reg)) {
+			verbose("R%d pointer comparison prohibited\n",
+				insn->src_reg);
+			return -EACCES;
+		}
 	} else {
 		if (insn->src_reg != BPF_REG_0) {
 			verbose("BPF_JMP uses reserved fields\n");
@@ -1155,6 +1233,9 @@ static int check_cond_jmp_op(struct verifier_env *env,
 			regs[insn->dst_reg].type = CONST_IMM;
 			regs[insn->dst_reg].imm = 0;
 		}
+	} else if (is_pointer_value(env, insn->dst_reg)) {
+		verbose("R%d pointer comparison prohibited\n", insn->dst_reg);
+		return -EACCES;
 	} else if (BPF_SRC(insn->code) == BPF_K &&
 		   (opcode == BPF_JEQ || opcode == BPF_JNE)) {
 
@@ -1658,7 +1739,7 @@ static int do_check(struct verifier_env *env)
 		}
 
 		if (class == BPF_ALU || class == BPF_ALU64) {
-			err = check_alu_op(regs, insn);
+			err = check_alu_op(env, insn);
 			if (err)
 				return err;
 
@@ -1816,6 +1897,11 @@ static int do_check(struct verifier_env *env)
 				if (err)
 					return err;
 
+				if (is_pointer_value(env, BPF_REG_0)) {
+					verbose("R0 leaks addr as return value\n");
+					return -EACCES;
+				}
+
 process_bpf_exit:
 				insn_idx = pop_stack(env, &prev_insn_idx);
 				if (insn_idx < 0) {
@@ -2144,6 +2230,8 @@ int bpf_check(struct bpf_prog **prog, union bpf_attr *attr)
 	if (ret < 0)
 		goto skip_full_check;
 
+	env->allow_ptr_leaks = capable(CAP_SYS_ADMIN);
+
 	ret = do_check(env);
 
 skip_full_check:

kernel/sysctl.c

@@ -64,6 +64,7 @@
 #include <linux/binfmts.h>
 #include <linux/sched/sysctl.h>
 #include <linux/kexec.h>
+#include <linux/bpf.h>
 
 #include <asm/uaccess.h>
 #include <asm/processor.h>
@@ -1138,6 +1139,18 @@ static struct ctl_table kern_table[] = {
 		.mode		= 0644,
 		.proc_handler	= timer_migration_handler,
 	},
+#endif
+#ifdef CONFIG_BPF_SYSCALL
+	{
+		.procname	= "unprivileged_bpf_disabled",
+		.data		= &sysctl_unprivileged_bpf_disabled,
+		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
+		.mode		= 0644,
+		/* only handle a transition from default "0" to "1" */
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &one,
+		.extra2		= &one,
+	},
 #endif
 	{ }
 };

net/core/filter.c

@@ -1640,7 +1640,8 @@ sk_filter_func_proto(enum bpf_func_id func_id)
 	case BPF_FUNC_ktime_get_ns:
 		return &bpf_ktime_get_ns_proto;
 	case BPF_FUNC_trace_printk:
-		return bpf_get_trace_printk_proto();
+		if (capable(CAP_SYS_ADMIN))
+			return bpf_get_trace_printk_proto();
 	default:
 		return NULL;
 	}

samples/bpf/libbpf.h

@@ -64,6 +64,14 @@ extern char bpf_log_buf[LOG_BUF_SIZE];
 		.off   = 0,					\
 		.imm   = 0 })
 
+#define BPF_MOV32_REG(DST, SRC)					\
+	((struct bpf_insn) {					\
+		.code  = BPF_ALU | BPF_MOV | BPF_X,		\
+		.dst_reg = DST,					\
+		.src_reg = SRC,					\
+		.off   = 0,					\
+		.imm   = 0 })
+
 /* Short form of mov, dst_reg = imm32 */
 
 #define BPF_MOV64_IMM(DST, IMM)					\