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

Daniel Borkmann says:

====================
pull-request: bpf 2021-09-28

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

We've added 10 non-merge commits during the last 14 day(s) which contain
a total of 11 files changed, 139 insertions(+), 53 deletions(-).

The main changes are:

1) Fix MIPS JIT jump code emission for too large offsets, from Piotr Krysiuk.

2) Fix x86 JIT atomic/fetch emission when dst reg maps to rax, from Johan Almbladh.

3) Fix cgroup_sk_alloc corner case when called from interrupt, from Daniel Borkmann.

4) Fix segfault in libbpf's linker for objects without BTF, from Kumar Kartikeya Dwivedi.

5) Fix bpf_jit_charge_modmem for applications with CAP_BPF, from Lorenz Bauer.

6) Fix return value handling for struct_ops BPF programs, from Hou Tao.

7) Various fixes to BPF selftests, from Jiri Benc.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
,

MAINTAINERS

@@ -3384,9 +3384,11 @@ F:	Documentation/networking/filter.rst
 F:	Documentation/userspace-api/ebpf/
 F:	arch/*/net/*
 F:	include/linux/bpf*
+F:	include/linux/btf*
 F:	include/linux/filter.h
 F:	include/trace/events/xdp.h
 F:	include/uapi/linux/bpf*
+F:	include/uapi/linux/btf*
 F:	include/uapi/linux/filter.h
 F:	kernel/bpf/
 F:	kernel/trace/bpf_trace.c

arch/mips/net/bpf_jit.c

@@ -662,6 +662,11 @@ static void build_epilogue(struct jit_ctx *ctx)
 	((int)K < 0 ? ((int)K >= SKF_LL_OFF ? func##_negative : func) : \
 	 func##_positive)
 
+static bool is_bad_offset(int b_off)
+{
+	return b_off > 0x1ffff || b_off < -0x20000;
+}
+
 static int build_body(struct jit_ctx *ctx)
 {
 	const struct bpf_prog *prog = ctx->skf;
@@ -728,7 +733,10 @@ static int build_body(struct jit_ctx *ctx)
 			/* Load return register on DS for failures */
 			emit_reg_move(r_ret, r_zero, ctx);
 			/* Return with error */
-			emit_b(b_imm(prog->len, ctx), ctx);
+			b_off = b_imm(prog->len, ctx);
+			if (is_bad_offset(b_off))
+				return -E2BIG;
+			emit_b(b_off, ctx);
 			emit_nop(ctx);
 			break;
 		case BPF_LD | BPF_W | BPF_IND:
@@ -775,8 +783,10 @@ static int build_body(struct jit_ctx *ctx)
 			emit_jalr(MIPS_R_RA, r_s0, ctx);
 			emit_reg_move(MIPS_R_A0, r_skb, ctx); /* delay slot */
 			/* Check the error value */
-			emit_bcond(MIPS_COND_NE, r_ret, 0,
-				   b_imm(prog->len, ctx), ctx);
+			b_off = b_imm(prog->len, ctx);
+			if (is_bad_offset(b_off))
+				return -E2BIG;
+			emit_bcond(MIPS_COND_NE, r_ret, 0, b_off, ctx);
 			emit_reg_move(r_ret, r_zero, ctx);
 			/* We are good */
 			/* X <- P[1:K] & 0xf */
@@ -855,8 +865,10 @@ static int build_body(struct jit_ctx *ctx)
 			/* A /= X */
 			ctx->flags |= SEEN_X | SEEN_A;
 			/* Check if r_X is zero */
-			emit_bcond(MIPS_COND_EQ, r_X, r_zero,
-				   b_imm(prog->len, ctx), ctx);
+			b_off = b_imm(prog->len, ctx);
+			if (is_bad_offset(b_off))
+				return -E2BIG;
+			emit_bcond(MIPS_COND_EQ, r_X, r_zero, b_off, ctx);
 			emit_load_imm(r_ret, 0, ctx); /* delay slot */
 			emit_div(r_A, r_X, ctx);
 			break;
@@ -864,8 +876,10 @@ static int build_body(struct jit_ctx *ctx)
 			/* A %= X */
 			ctx->flags |= SEEN_X | SEEN_A;
 			/* Check if r_X is zero */
-			emit_bcond(MIPS_COND_EQ, r_X, r_zero,
-				   b_imm(prog->len, ctx), ctx);
+			b_off = b_imm(prog->len, ctx);
+			if (is_bad_offset(b_off))
+				return -E2BIG;
+			emit_bcond(MIPS_COND_EQ, r_X, r_zero, b_off, ctx);
 			emit_load_imm(r_ret, 0, ctx); /* delay slot */
 			emit_mod(r_A, r_X, ctx);
 			break;
@@ -926,7 +940,10 @@ static int build_body(struct jit_ctx *ctx)
 			break;
 		case BPF_JMP | BPF_JA:
 			/* pc += K */
-			emit_b(b_imm(i + k + 1, ctx), ctx);
+			b_off = b_imm(i + k + 1, ctx);
+			if (is_bad_offset(b_off))
+				return -E2BIG;
+			emit_b(b_off, ctx);
 			emit_nop(ctx);
 			break;
 		case BPF_JMP | BPF_JEQ | BPF_K:
@@ -1056,12 +1073,16 @@ static int build_body(struct jit_ctx *ctx)
 			break;
 		case BPF_RET | BPF_A:
 			ctx->flags |= SEEN_A;
-			if (i != prog->len - 1)
+			if (i != prog->len - 1) {
 				/*
 				 * If this is not the last instruction
 				 * then jump to the epilogue
 				 */
-				emit_b(b_imm(prog->len, ctx), ctx);
+				b_off = b_imm(prog->len, ctx);
+				if (is_bad_offset(b_off))
+					return -E2BIG;
+				emit_b(b_off, ctx);
+			}
 			emit_reg_move(r_ret, r_A, ctx); /* delay slot */
 			break;
 		case BPF_RET | BPF_K:
@@ -1075,7 +1096,10 @@ static int build_body(struct jit_ctx *ctx)
 				 * If this is not the last instruction
 				 * then jump to the epilogue
 				 */
-				emit_b(b_imm(prog->len, ctx), ctx);
+				b_off = b_imm(prog->len, ctx);
+				if (is_bad_offset(b_off))
+					return -E2BIG;
+				emit_b(b_off, ctx);
 				emit_nop(ctx);
 			}
 			break;
@@ -1133,8 +1157,10 @@ static int build_body(struct jit_ctx *ctx)
 			/* Load *dev pointer */
 			emit_load_ptr(r_s0, r_skb, off, ctx);
 			/* error (0) in the delay slot */
-			emit_bcond(MIPS_COND_EQ, r_s0, r_zero,
-				   b_imm(prog->len, ctx), ctx);
+			b_off = b_imm(prog->len, ctx);
+			if (is_bad_offset(b_off))
+				return -E2BIG;
+			emit_bcond(MIPS_COND_EQ, r_s0, r_zero, b_off, ctx);
 			emit_reg_move(r_ret, r_zero, ctx);
 			if (code == (BPF_ANC | SKF_AD_IFINDEX)) {
 				BUILD_BUG_ON(sizeof_field(struct net_device, ifindex) != 4);
@@ -1244,7 +1270,10 @@ void bpf_jit_compile(struct bpf_prog *fp)
 
 	/* Generate the actual JIT code */
 	build_prologue(&ctx);
-	build_body(&ctx);
+	if (build_body(&ctx)) {
+		module_memfree(ctx.target);
+		goto out;
+	}
 	build_epilogue(&ctx);
 
 	/* Update the icache */

arch/x86/net/bpf_jit_comp.c

@@ -1341,9 +1341,10 @@ st:			if (is_imm8(insn->off))
 			if (insn->imm == (BPF_AND | BPF_FETCH) ||
 			    insn->imm == (BPF_OR | BPF_FETCH) ||
 			    insn->imm == (BPF_XOR | BPF_FETCH)) {
-				u8 *branch_target;
 				bool is64 = BPF_SIZE(insn->code) == BPF_DW;
 				u32 real_src_reg = src_reg;
+				u32 real_dst_reg = dst_reg;
+				u8 *branch_target;
 
 				/*
 				 * Can't be implemented with a single x86 insn.
@@ -1354,11 +1355,13 @@ st:			if (is_imm8(insn->off))
 				emit_mov_reg(&prog, true, BPF_REG_AX, BPF_REG_0);
 				if (src_reg == BPF_REG_0)
 					real_src_reg = BPF_REG_AX;
+				if (dst_reg == BPF_REG_0)
+					real_dst_reg = BPF_REG_AX;
 
 				branch_target = prog;
 				/* Load old value */
 				emit_ldx(&prog, BPF_SIZE(insn->code),
-					 BPF_REG_0, dst_reg, insn->off);
+					 BPF_REG_0, real_dst_reg, insn->off);
 				/*
 				 * Perform the (commutative) operation locally,
 				 * put the result in the AUX_REG.
@@ -1369,7 +1372,8 @@ st:			if (is_imm8(insn->off))
 				      add_2reg(0xC0, AUX_REG, real_src_reg));
 				/* Attempt to swap in new value */
 				err = emit_atomic(&prog, BPF_CMPXCHG,
-						  dst_reg, AUX_REG, insn->off,
+						  real_dst_reg, AUX_REG,
+						  insn->off,
 						  BPF_SIZE(insn->code));
 				if (WARN_ON(err))
 					return err;
@@ -1383,7 +1387,6 @@ st:			if (is_imm8(insn->off))
 				/* Restore R0 after clobbering RAX */
 				emit_mov_reg(&prog, true, BPF_REG_0, BPF_REG_AX);
 				break;
-
 			}
 
 			err = emit_atomic(&prog, insn->imm, dst_reg, src_reg,
@@ -1744,7 +1747,7 @@ static void restore_regs(const struct btf_func_model *m, u8 **prog, int nr_args,
 }
 
 static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog,
-			   struct bpf_prog *p, int stack_size, bool mod_ret)
+			   struct bpf_prog *p, int stack_size, bool save_ret)
 {
 	u8 *prog = *pprog;
 	u8 *jmp_insn;
@@ -1777,11 +1780,15 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog,
 	if (emit_call(&prog, p->bpf_func, prog))
 		return -EINVAL;
 
-	/* BPF_TRAMP_MODIFY_RETURN trampolines can modify the return
+	/*
+	 * BPF_TRAMP_MODIFY_RETURN trampolines can modify the return
 	 * of the previous call which is then passed on the stack to
 	 * the next BPF program.
+	 *
+	 * BPF_TRAMP_FENTRY trampoline may need to return the return
+	 * value of BPF_PROG_TYPE_STRUCT_OPS prog.
 	 */
-	if (mod_ret)
+	if (save_ret)
 		emit_stx(&prog, BPF_DW, BPF_REG_FP, BPF_REG_0, -8);
 
 	/* replace 2 nops with JE insn, since jmp target is known */
@@ -1828,13 +1835,15 @@ static int emit_cond_near_jump(u8 **pprog, void *func, void *ip, u8 jmp_cond)
 }
 
 static int invoke_bpf(const struct btf_func_model *m, u8 **pprog,
-		      struct bpf_tramp_progs *tp, int stack_size)
+		      struct bpf_tramp_progs *tp, int stack_size,
+		      bool save_ret)
 {
 	int i;
 	u8 *prog = *pprog;
 
 	for (i = 0; i < tp->nr_progs; i++) {
-		if (invoke_bpf_prog(m, &prog, tp->progs[i], stack_size, false))
+		if (invoke_bpf_prog(m, &prog, tp->progs[i], stack_size,
+				    save_ret))
 			return -EINVAL;
 	}
 	*pprog = prog;
@@ -1877,6 +1886,23 @@ static int invoke_bpf_mod_ret(const struct btf_func_model *m, u8 **pprog,
 	return 0;
 }
 
+static bool is_valid_bpf_tramp_flags(unsigned int flags)
+{
+	if ((flags & BPF_TRAMP_F_RESTORE_REGS) &&
+	    (flags & BPF_TRAMP_F_SKIP_FRAME))
+		return false;
+
+	/*
+	 * BPF_TRAMP_F_RET_FENTRY_RET is only used by bpf_struct_ops,
+	 * and it must be used alone.
+	 */
+	if ((flags & BPF_TRAMP_F_RET_FENTRY_RET) &&
+	    (flags & ~BPF_TRAMP_F_RET_FENTRY_RET))
+		return false;
+
+	return true;
+}
+
 /* Example:
  * __be16 eth_type_trans(struct sk_buff *skb, struct net_device *dev);
  * its 'struct btf_func_model' will be nr_args=2
@@ -1949,17 +1975,19 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 	struct bpf_tramp_progs *fmod_ret = &tprogs[BPF_TRAMP_MODIFY_RETURN];
 	u8 **branches = NULL;
 	u8 *prog;
+	bool save_ret;
 
 	/* x86-64 supports up to 6 arguments. 7+ can be added in the future */
 	if (nr_args > 6)
 		return -ENOTSUPP;
 
-	if ((flags & BPF_TRAMP_F_RESTORE_REGS) &&
-	    (flags & BPF_TRAMP_F_SKIP_FRAME))
+	if (!is_valid_bpf_tramp_flags(flags))
 		return -EINVAL;
 
-	if (flags & BPF_TRAMP_F_CALL_ORIG)
-		stack_size += 8; /* room for return value of orig_call */
+	/* room for return value of orig_call or fentry prog */
+	save_ret = flags & (BPF_TRAMP_F_CALL_ORIG | BPF_TRAMP_F_RET_FENTRY_RET);
+	if (save_ret)
+		stack_size += 8;
 
 	if (flags & BPF_TRAMP_F_IP_ARG)
 		stack_size += 8; /* room for IP address argument */
@@ -2005,7 +2033,8 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 	}
 
 	if (fentry->nr_progs)
-		if (invoke_bpf(m, &prog, fentry, stack_size))
+		if (invoke_bpf(m, &prog, fentry, stack_size,
+			       flags & BPF_TRAMP_F_RET_FENTRY_RET))
 			return -EINVAL;
 
 	if (fmod_ret->nr_progs) {
@@ -2052,7 +2081,7 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 	}
 
 	if (fexit->nr_progs)
-		if (invoke_bpf(m, &prog, fexit, stack_size)) {
+		if (invoke_bpf(m, &prog, fexit, stack_size, false)) {
 			ret = -EINVAL;
 			goto cleanup;
 		}
@@ -2072,9 +2101,10 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 			ret = -EINVAL;
 			goto cleanup;
 		}
-		/* restore original return value back into RAX */
-		emit_ldx(&prog, BPF_DW, BPF_REG_0, BPF_REG_FP, -8);
 	}
+	/* restore return value of orig_call or fentry prog back into RAX */
+	if (save_ret)
+		emit_ldx(&prog, BPF_DW, BPF_REG_0, BPF_REG_FP, -8);
 
 	EMIT1(0x5B); /* pop rbx */
 	EMIT1(0xC9); /* leave */

include/linux/bpf.h

@@ -578,11 +578,12 @@ struct btf_func_model {
  * programs only. Should not be used with normal calls and indirect calls.
  */
 #define BPF_TRAMP_F_SKIP_FRAME		BIT(2)
-
 /* Store IP address of the caller on the trampoline stack,
  * so it's available for trampoline's programs.
  */
 #define BPF_TRAMP_F_IP_ARG		BIT(3)
+/* Return the return value of fentry prog. Only used by bpf_struct_ops. */
+#define BPF_TRAMP_F_RET_FENTRY_RET	BIT(4)
 
 /* Each call __bpf_prog_enter + call bpf_func + call __bpf_prog_exit is ~50
  * bytes on x86.  Pick a number to fit into BPF_IMAGE_SIZE / 2

kernel/bpf/bpf_struct_ops.c

@@ -368,6 +368,7 @@ static int bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 		const struct btf_type *mtype, *ptype;
 		struct bpf_prog *prog;
 		u32 moff;
+		u32 flags;
 
 		moff = btf_member_bit_offset(t, member) / 8;
 		ptype = btf_type_resolve_ptr(btf_vmlinux, member->type, NULL);
@@ -431,10 +432,12 @@ static int bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 
 		tprogs[BPF_TRAMP_FENTRY].progs[0] = prog;
 		tprogs[BPF_TRAMP_FENTRY].nr_progs = 1;
+		flags = st_ops->func_models[i].ret_size > 0 ?
+			BPF_TRAMP_F_RET_FENTRY_RET : 0;
 		err = arch_prepare_bpf_trampoline(NULL, image,
 						  st_map->image + PAGE_SIZE,
-						  &st_ops->func_models[i], 0,
-						  tprogs, NULL);
+						  &st_ops->func_models[i],
+						  flags, tprogs, NULL);
 		if (err < 0)
 			goto reset_unlock;
 

kernel/bpf/core.c

@@ -827,7 +827,7 @@ int bpf_jit_charge_modmem(u32 pages)
 {
 	if (atomic_long_add_return(pages, &bpf_jit_current) >
 	    (bpf_jit_limit >> PAGE_SHIFT)) {
-		if (!capable(CAP_SYS_ADMIN)) {
+		if (!bpf_capable()) {
 			atomic_long_sub(pages, &bpf_jit_current);
 			return -EPERM;
 		}

kernel/cgroup/cgroup.c

@@ -6574,22 +6574,29 @@ int cgroup_parse_float(const char *input, unsigned dec_shift, s64 *v)
 
 void cgroup_sk_alloc(struct sock_cgroup_data *skcd)
 {
-	/* Don't associate the sock with unrelated interrupted task's cgroup. */
-	if (in_interrupt())
-		return;
+	struct cgroup *cgroup;
 
 	rcu_read_lock();
+	/* Don't associate the sock with unrelated interrupted task's cgroup. */
+	if (in_interrupt()) {
+		cgroup = &cgrp_dfl_root.cgrp;
+		cgroup_get(cgroup);
+		goto out;
+	}
+
 	while (true) {
 		struct css_set *cset;
 
 		cset = task_css_set(current);
 		if (likely(cgroup_tryget(cset->dfl_cgrp))) {
-			skcd->cgroup = cset->dfl_cgrp;
-			cgroup_bpf_get(cset->dfl_cgrp);
+			cgroup = cset->dfl_cgrp;
 			break;
 		}
 		cpu_relax();
 	}
+out:
+	skcd->cgroup = cgroup;
+	cgroup_bpf_get(cgroup);
 	rcu_read_unlock();
 }
 

net/bpf/test_run.c

@@ -552,6 +552,12 @@ static void convert_skb_to___skb(struct sk_buff *skb, struct __sk_buff *__skb)
 	__skb->gso_segs = skb_shinfo(skb)->gso_segs;
 }
 
+static struct proto bpf_dummy_proto = {
+	.name   = "bpf_dummy",
+	.owner  = THIS_MODULE,
+	.obj_size = sizeof(struct sock),
+};
+
 int bpf_prog_test_run_skb(struct bpf_prog *prog, const union bpf_attr *kattr,
 			  union bpf_attr __user *uattr)
 {
@@ -596,20 +602,19 @@ int bpf_prog_test_run_skb(struct bpf_prog *prog, const union bpf_attr *kattr,
 		break;
 	}
 
-	sk = kzalloc(sizeof(struct sock), GFP_USER);
+	sk = sk_alloc(net, AF_UNSPEC, GFP_USER, &bpf_dummy_proto, 1);
 	if (!sk) {
 		kfree(data);
 		kfree(ctx);
 		return -ENOMEM;
 	}
-	sock_net_set(sk, net);
 	sock_init_data(NULL, sk);
 
 	skb = build_skb(data, 0);
 	if (!skb) {
 		kfree(data);
 		kfree(ctx);
-		kfree(sk);
+		sk_free(sk);
 		return -ENOMEM;
 	}
 	skb->sk = sk;
@@ -682,8 +687,7 @@ int bpf_prog_test_run_skb(struct bpf_prog *prog, const union bpf_attr *kattr,
 	if (dev && dev != net->loopback_dev)
 		dev_put(dev);
 	kfree_skb(skb);
-	bpf_sk_storage_free(sk);
-	kfree(sk);
+	sk_free(sk);
 	kfree(ctx);
 	return ret;
 }

tools/lib/bpf/linker.c

@@ -1649,11 +1649,17 @@ static bool btf_is_non_static(const struct btf_type *t)
 static int find_glob_sym_btf(struct src_obj *obj, Elf64_Sym *sym, const char *sym_name,
 			     int *out_btf_sec_id, int *out_btf_id)
 {
-	int i, j, n = btf__get_nr_types(obj->btf), m, btf_id = 0;
+	int i, j, n, m, btf_id = 0;
 	const struct btf_type *t;
 	const struct btf_var_secinfo *vi;
 	const char *name;
 
+	if (!obj->btf) {
+		pr_warn("failed to find BTF info for object '%s'\n", obj->filename);
+		return -EINVAL;
+	}
+
+	n = btf__get_nr_types(obj->btf);
 	for (i = 1; i <= n; i++) {
 		t = btf__type_by_id(obj->btf, i);
 

tools/testing/selftests/bpf/Makefile

@@ -375,7 +375,8 @@ $(TRUNNER_BPF_OBJS): $(TRUNNER_OUTPUT)/%.o:				\
 		     $(TRUNNER_BPF_PROGS_DIR)/%.c			\
 		     $(TRUNNER_BPF_PROGS_DIR)/*.h			\
 		     $$(INCLUDE_DIR)/vmlinux.h				\
-		     $(wildcard $(BPFDIR)/bpf_*.h) | $(TRUNNER_OUTPUT)
+		     $(wildcard $(BPFDIR)/bpf_*.h)			\
+		     | $(TRUNNER_OUTPUT) $$(BPFOBJ)
 	$$(call $(TRUNNER_BPF_BUILD_RULE),$$<,$$@,			\
 					  $(TRUNNER_BPF_CFLAGS))
 

tools/testing/selftests/bpf/test_lwt_ip_encap.sh

@@ -112,6 +112,14 @@ setup()
 	ip netns add "${NS2}"
 	ip netns add "${NS3}"
 
+	# rp_filter gets confused by what these tests are doing, so disable it
+	ip netns exec ${NS1} sysctl -wq net.ipv4.conf.all.rp_filter=0
+	ip netns exec ${NS2} sysctl -wq net.ipv4.conf.all.rp_filter=0
+	ip netns exec ${NS3} sysctl -wq net.ipv4.conf.all.rp_filter=0
+	ip netns exec ${NS1} sysctl -wq net.ipv4.conf.default.rp_filter=0
+	ip netns exec ${NS2} sysctl -wq net.ipv4.conf.default.rp_filter=0
+	ip netns exec ${NS3} sysctl -wq net.ipv4.conf.default.rp_filter=0
+
 	ip link add veth1 type veth peer name veth2
 	ip link add veth3 type veth peer name veth4
 	ip link add veth5 type veth peer name veth6
@@ -236,11 +244,6 @@ setup()
 	ip -netns ${NS1} -6 route add ${IPv6_GRE}/128 dev veth5 via ${IPv6_6} ${VRF}
 	ip -netns ${NS2} -6 route add ${IPv6_GRE}/128 dev veth7 via ${IPv6_8} ${VRF}
 
-	# rp_filter gets confused by what these tests are doing, so disable it
-	ip netns exec ${NS1} sysctl -wq net.ipv4.conf.all.rp_filter=0
-	ip netns exec ${NS2} sysctl -wq net.ipv4.conf.all.rp_filter=0
-	ip netns exec ${NS3} sysctl -wq net.ipv4.conf.all.rp_filter=0
-
 	TMPFILE=$(mktemp /tmp/test_lwt_ip_encap.XXXXXX)
 
 	sleep 1  # reduce flakiness