bpf, x86: Generate trampolines from bpf_tramp_links

Replace struct bpf_tramp_progs with struct bpf_tramp_links to collect
struct bpf_tramp_link(s) for a trampoline.  struct bpf_tramp_link
extends bpf_link to act as a linked list node.

arch_prepare_bpf_trampoline() accepts a struct bpf_tramp_links to
collects all bpf_tramp_link(s) that a trampoline should call.

Change BPF trampoline and bpf_struct_ops to pass bpf_tramp_links
instead of bpf_tramp_progs.

Signed-off-by: Kui-Feng Lee <kuifeng@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20220510205923.3206889-2-kuifeng@fb.com

Author: Kui-Feng Lee

arch/x86/net/bpf_jit_comp.c

@@ -1762,10 +1762,12 @@ 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 save_ret)
+			   struct bpf_tramp_link *l, int stack_size,
+			   bool save_ret)
 {
 	u8 *prog = *pprog;
 	u8 *jmp_insn;
+	struct bpf_prog *p = l->link.prog;
 
 	/* arg1: mov rdi, progs[i] */
 	emit_mov_imm64(&prog, BPF_REG_1, (long) p >> 32, (u32) (long) p);
@@ -1850,14 +1852,14 @@ 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_links *tl, 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,
+	for (i = 0; i < tl->nr_links; i++) {
+		if (invoke_bpf_prog(m, &prog, tl->links[i], stack_size,
 				    save_ret))
 			return -EINVAL;
 	}
@@ -1866,7 +1868,7 @@ static int invoke_bpf(const struct btf_func_model *m, u8 **pprog,
 }
 
 static int invoke_bpf_mod_ret(const struct btf_func_model *m, u8 **pprog,
-			      struct bpf_tramp_progs *tp, int stack_size,
+			      struct bpf_tramp_links *tl, int stack_size,
 			      u8 **branches)
 {
 	u8 *prog = *pprog;
@@ -1877,8 +1879,8 @@ static int invoke_bpf_mod_ret(const struct btf_func_model *m, u8 **pprog,
 	 */
 	emit_mov_imm32(&prog, false, BPF_REG_0, 0);
 	emit_stx(&prog, BPF_DW, BPF_REG_FP, BPF_REG_0, -8);
-	for (i = 0; i < tp->nr_progs; i++) {
-		if (invoke_bpf_prog(m, &prog, tp->progs[i], stack_size, true))
+	for (i = 0; i < tl->nr_links; i++) {
+		if (invoke_bpf_prog(m, &prog, tl->links[i], stack_size, true))
 			return -EINVAL;
 
 		/* mod_ret prog stored return value into [rbp - 8]. Emit:
@@ -1980,14 +1982,14 @@ static bool is_valid_bpf_tramp_flags(unsigned int flags)
  */
 int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *image_end,
 				const struct btf_func_model *m, u32 flags,
-				struct bpf_tramp_progs *tprogs,
+				struct bpf_tramp_links *tlinks,
 				void *orig_call)
 {
 	int ret, i, nr_args = m->nr_args;
 	int regs_off, ip_off, args_off, stack_size = nr_args * 8;
-	struct bpf_tramp_progs *fentry = &tprogs[BPF_TRAMP_FENTRY];
-	struct bpf_tramp_progs *fexit = &tprogs[BPF_TRAMP_FEXIT];
-	struct bpf_tramp_progs *fmod_ret = &tprogs[BPF_TRAMP_MODIFY_RETURN];
+	struct bpf_tramp_links *fentry = &tlinks[BPF_TRAMP_FENTRY];
+	struct bpf_tramp_links *fexit = &tlinks[BPF_TRAMP_FEXIT];
+	struct bpf_tramp_links *fmod_ret = &tlinks[BPF_TRAMP_MODIFY_RETURN];
 	u8 **branches = NULL;
 	u8 *prog;
 	bool save_ret;
@@ -2078,13 +2080,13 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 		}
 	}
 
-	if (fentry->nr_progs)
+	if (fentry->nr_links)
 		if (invoke_bpf(m, &prog, fentry, regs_off,
 			       flags & BPF_TRAMP_F_RET_FENTRY_RET))
 			return -EINVAL;
 
-	if (fmod_ret->nr_progs) {
-		branches = kcalloc(fmod_ret->nr_progs, sizeof(u8 *),
+	if (fmod_ret->nr_links) {
+		branches = kcalloc(fmod_ret->nr_links, sizeof(u8 *),
 				   GFP_KERNEL);
 		if (!branches)
 			return -ENOMEM;
@@ -2111,7 +2113,7 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 		prog += X86_PATCH_SIZE;
 	}
 
-	if (fmod_ret->nr_progs) {
+	if (fmod_ret->nr_links) {
 		/* From Intel 64 and IA-32 Architectures Optimization
 		 * Reference Manual, 3.4.1.4 Code Alignment, Assembly/Compiler
 		 * Coding Rule 11: All branch targets should be 16-byte
@@ -2121,12 +2123,12 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, void *i
 		/* Update the branches saved in invoke_bpf_mod_ret with the
 		 * aligned address of do_fexit.
 		 */
-		for (i = 0; i < fmod_ret->nr_progs; i++)
+		for (i = 0; i < fmod_ret->nr_links; i++)
 			emit_cond_near_jump(&branches[i], prog, branches[i],
 					    X86_JNE);
 	}
 
-	if (fexit->nr_progs)
+	if (fexit->nr_links)
 		if (invoke_bpf(m, &prog, fexit, regs_off, false)) {
 			ret = -EINVAL;
 			goto cleanup;

include/linux/bpf.h

@@ -723,11 +723,11 @@ struct btf_func_model {
 /* Each call __bpf_prog_enter + call bpf_func + call __bpf_prog_exit is ~50
  * bytes on x86.  Pick a number to fit into BPF_IMAGE_SIZE / 2
  */
-#define BPF_MAX_TRAMP_PROGS 38
+#define BPF_MAX_TRAMP_LINKS 38
 
-struct bpf_tramp_progs {
-	struct bpf_prog *progs[BPF_MAX_TRAMP_PROGS];
-	int nr_progs;
+struct bpf_tramp_links {
+	struct bpf_tramp_link *links[BPF_MAX_TRAMP_LINKS];
+	int nr_links;
 };
 
 /* Different use cases for BPF trampoline:
@@ -753,7 +753,7 @@ struct bpf_tramp_progs {
 struct bpf_tramp_image;
 int arch_prepare_bpf_trampoline(struct bpf_tramp_image *tr, void *image, void *image_end,
 				const struct btf_func_model *m, u32 flags,
-				struct bpf_tramp_progs *tprogs,
+				struct bpf_tramp_links *tlinks,
 				void *orig_call);
 /* these two functions are called from generated trampoline */
 u64 notrace __bpf_prog_enter(struct bpf_prog *prog);
@@ -852,9 +852,10 @@ static __always_inline __nocfi unsigned int bpf_dispatcher_nop_func(
 {
 	return bpf_func(ctx, insnsi);
 }
+
 #ifdef CONFIG_BPF_JIT
-int bpf_trampoline_link_prog(struct bpf_prog *prog, struct bpf_trampoline *tr);
-int bpf_trampoline_unlink_prog(struct bpf_prog *prog, struct bpf_trampoline *tr);
+int bpf_trampoline_link_prog(struct bpf_tramp_link *link, struct bpf_trampoline *tr);
+int bpf_trampoline_unlink_prog(struct bpf_tramp_link *link, struct bpf_trampoline *tr);
 struct bpf_trampoline *bpf_trampoline_get(u64 key,
 					  struct bpf_attach_target_info *tgt_info);
 void bpf_trampoline_put(struct bpf_trampoline *tr);
@@ -905,12 +906,12 @@ int bpf_jit_charge_modmem(u32 size);
 void bpf_jit_uncharge_modmem(u32 size);
 bool bpf_prog_has_trampoline(const struct bpf_prog *prog);
 #else
-static inline int bpf_trampoline_link_prog(struct bpf_prog *prog,
+static inline int bpf_trampoline_link_prog(struct bpf_tramp_link *link,
 					   struct bpf_trampoline *tr)
 {
 	return -ENOTSUPP;
 }
-static inline int bpf_trampoline_unlink_prog(struct bpf_prog *prog,
+static inline int bpf_trampoline_unlink_prog(struct bpf_tramp_link *link,
 					     struct bpf_trampoline *tr)
 {
 	return -ENOTSUPP;
@@ -1009,7 +1010,6 @@ struct bpf_prog_aux {
 	bool tail_call_reachable;
 	bool xdp_has_frags;
 	bool use_bpf_prog_pack;
-	struct hlist_node tramp_hlist;
 	/* BTF_KIND_FUNC_PROTO for valid attach_btf_id */
 	const struct btf_type *attach_func_proto;
 	/* function name for valid attach_btf_id */
@@ -1096,6 +1096,18 @@ struct bpf_link_ops {
 			      struct bpf_link_info *info);
 };
 
+struct bpf_tramp_link {
+	struct bpf_link link;
+	struct hlist_node tramp_hlist;
+};
+
+struct bpf_tracing_link {
+	struct bpf_tramp_link link;
+	enum bpf_attach_type attach_type;
+	struct bpf_trampoline *trampoline;
+	struct bpf_prog *tgt_prog;
+};
+
 struct bpf_link_primer {
 	struct bpf_link *link;
 	struct file *file;
@@ -1133,8 +1145,8 @@ bool bpf_struct_ops_get(const void *kdata);
 void bpf_struct_ops_put(const void *kdata);
 int bpf_struct_ops_map_sys_lookup_elem(struct bpf_map *map, void *key,
 				       void *value);
-int bpf_struct_ops_prepare_trampoline(struct bpf_tramp_progs *tprogs,
-				      struct bpf_prog *prog,
+int bpf_struct_ops_prepare_trampoline(struct bpf_tramp_links *tlinks,
+				      struct bpf_tramp_link *link,
 				      const struct btf_func_model *model,
 				      void *image, void *image_end);
 static inline bool bpf_try_module_get(const void *data, struct module *owner)

include/linux/bpf_types.h

@@ -141,3 +141,4 @@ BPF_LINK_TYPE(BPF_LINK_TYPE_XDP, xdp)
 BPF_LINK_TYPE(BPF_LINK_TYPE_PERF_EVENT, perf)
 #endif
 BPF_LINK_TYPE(BPF_LINK_TYPE_KPROBE_MULTI, kprobe_multi)
+BPF_LINK_TYPE(BPF_LINK_TYPE_STRUCT_OPS, struct_ops)

include/uapi/linux/bpf.h

@@ -1013,6 +1013,7 @@ enum bpf_link_type {
 	BPF_LINK_TYPE_XDP = 6,
 	BPF_LINK_TYPE_PERF_EVENT = 7,
 	BPF_LINK_TYPE_KPROBE_MULTI = 8,
+	BPF_LINK_TYPE_STRUCT_OPS = 9,
 
 	MAX_BPF_LINK_TYPE,
 };

kernel/bpf/bpf_struct_ops.c

@@ -33,15 +33,15 @@ struct bpf_struct_ops_map {
 	const struct bpf_struct_ops *st_ops;
 	/* protect map_update */
 	struct mutex lock;
-	/* progs has all the bpf_prog that is populated
+	/* link has all the bpf_links that is populated
 	 * to the func ptr of the kernel's struct
 	 * (in kvalue.data).
 	 */
-	struct bpf_prog **progs;
+	struct bpf_link **links;
 	/* image is a page that has all the trampolines
 	 * that stores the func args before calling the bpf_prog.
 	 * A PAGE_SIZE "image" is enough to store all trampoline for
-	 * "progs[]".
+	 * "links[]".
 	 */
 	void *image;
 	/* uvalue->data stores the kernel struct
@@ -283,9 +283,9 @@ static void bpf_struct_ops_map_put_progs(struct bpf_struct_ops_map *st_map)
 	u32 i;
 
 	for (i = 0; i < btf_type_vlen(t); i++) {
-		if (st_map->progs[i]) {
-			bpf_prog_put(st_map->progs[i]);
-			st_map->progs[i] = NULL;
+		if (st_map->links[i]) {
+			bpf_link_put(st_map->links[i]);
+			st_map->links[i] = NULL;
 		}
 	}
 }
@@ -316,18 +316,34 @@ static int check_zero_holes(const struct btf_type *t, void *data)
 	return 0;
 }
 
-int bpf_struct_ops_prepare_trampoline(struct bpf_tramp_progs *tprogs,
-				      struct bpf_prog *prog,
+static void bpf_struct_ops_link_release(struct bpf_link *link)
+{
+}
+
+static void bpf_struct_ops_link_dealloc(struct bpf_link *link)
+{
+	struct bpf_tramp_link *tlink = container_of(link, struct bpf_tramp_link, link);
+
+	kfree(tlink);
+}
+
+const struct bpf_link_ops bpf_struct_ops_link_lops = {
+	.release = bpf_struct_ops_link_release,
+	.dealloc = bpf_struct_ops_link_dealloc,
+};
+
+int bpf_struct_ops_prepare_trampoline(struct bpf_tramp_links *tlinks,
+				      struct bpf_tramp_link *link,
 				      const struct btf_func_model *model,
 				      void *image, void *image_end)
 {
 	u32 flags;
 
-	tprogs[BPF_TRAMP_FENTRY].progs[0] = prog;
-	tprogs[BPF_TRAMP_FENTRY].nr_progs = 1;
+	tlinks[BPF_TRAMP_FENTRY].links[0] = link;
+	tlinks[BPF_TRAMP_FENTRY].nr_links = 1;
 	flags = model->ret_size > 0 ? BPF_TRAMP_F_RET_FENTRY_RET : 0;
 	return arch_prepare_bpf_trampoline(NULL, image, image_end,
-					   model, flags, tprogs, NULL);
+					   model, flags, tlinks, NULL);
 }
 
 static int bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
@@ -338,7 +354,7 @@ static int bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 	struct bpf_struct_ops_value *uvalue, *kvalue;
 	const struct btf_member *member;
 	const struct btf_type *t = st_ops->type;
-	struct bpf_tramp_progs *tprogs = NULL;
+	struct bpf_tramp_links *tlinks = NULL;
 	void *udata, *kdata;
 	int prog_fd, err = 0;
 	void *image, *image_end;
@@ -362,8 +378,8 @@ static int bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 	if (uvalue->state || refcount_read(&uvalue->refcnt))
 		return -EINVAL;
 
-	tprogs = kcalloc(BPF_TRAMP_MAX, sizeof(*tprogs), GFP_KERNEL);
-	if (!tprogs)
+	tlinks = kcalloc(BPF_TRAMP_MAX, sizeof(*tlinks), GFP_KERNEL);
+	if (!tlinks)
 		return -ENOMEM;
 
 	uvalue = (struct bpf_struct_ops_value *)st_map->uvalue;
@@ -386,6 +402,7 @@ static int bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 	for_each_member(i, t, member) {
 		const struct btf_type *mtype, *ptype;
 		struct bpf_prog *prog;
+		struct bpf_tramp_link *link;
 		u32 moff;
 
 		moff = __btf_member_bit_offset(t, member) / 8;
@@ -439,16 +456,26 @@ static int bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 			err = PTR_ERR(prog);
 			goto reset_unlock;
 		}
-		st_map->progs[i] = prog;
 
 		if (prog->type != BPF_PROG_TYPE_STRUCT_OPS ||
 		    prog->aux->attach_btf_id != st_ops->type_id ||
 		    prog->expected_attach_type != i) {
+			bpf_prog_put(prog);
 			err = -EINVAL;
 			goto reset_unlock;
 		}
 
-		err = bpf_struct_ops_prepare_trampoline(tprogs, prog,
+		link = kzalloc(sizeof(*link), GFP_USER);
+		if (!link) {
+			bpf_prog_put(prog);
+			err = -ENOMEM;
+			goto reset_unlock;
+		}
+		bpf_link_init(&link->link, BPF_LINK_TYPE_STRUCT_OPS,
+			      &bpf_struct_ops_link_lops, prog);
+		st_map->links[i] = &link->link;
+
+		err = bpf_struct_ops_prepare_trampoline(tlinks, link,
 							&st_ops->func_models[i],
 							image, image_end);
 		if (err < 0)
@@ -491,7 +518,7 @@ static int bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
 	memset(uvalue, 0, map->value_size);
 	memset(kvalue, 0, map->value_size);
 unlock:
-	kfree(tprogs);
+	kfree(tlinks);
 	mutex_unlock(&st_map->lock);
 	return err;
 }
@@ -546,9 +573,9 @@ static void bpf_struct_ops_map_free(struct bpf_map *map)
 {
 	struct bpf_struct_ops_map *st_map = (struct bpf_struct_ops_map *)map;
 
-	if (st_map->progs)
+	if (st_map->links)
 		bpf_struct_ops_map_put_progs(st_map);
-	bpf_map_area_free(st_map->progs);
+	bpf_map_area_free(st_map->links);
 	bpf_jit_free_exec(st_map->image);
 	bpf_map_area_free(st_map->uvalue);
 	bpf_map_area_free(st_map);
@@ -597,11 +624,11 @@ static struct bpf_map *bpf_struct_ops_map_alloc(union bpf_attr *attr)
 	map = &st_map->map;
 
 	st_map->uvalue = bpf_map_area_alloc(vt->size, NUMA_NO_NODE);
-	st_map->progs =
-		bpf_map_area_alloc(btf_type_vlen(t) * sizeof(struct bpf_prog *),
+	st_map->links =
+		bpf_map_area_alloc(btf_type_vlen(t) * sizeof(struct bpf_links *),
 				   NUMA_NO_NODE);
 	st_map->image = bpf_jit_alloc_exec(PAGE_SIZE);
-	if (!st_map->uvalue || !st_map->progs || !st_map->image) {
+	if (!st_map->uvalue || !st_map->links || !st_map->image) {
 		bpf_struct_ops_map_free(map);
 		return ERR_PTR(-ENOMEM);
 	}