[PATCH v2 bpf-next 0/3] bpffs pretty print for cgroup local storage

[PATCH v2 bpf-next 0/3] bpffs pretty print for cgroup local storage

[Roman Gushchin]

This patchset implements bpffs pretty printing for cgroup local storage maps.

v2->v1:
  - fixed size checks in cgroup_storage_check_btf(),
  based on input from Yonghong and Martin
  - implemented and used btf_type_is_reg_int(), suggested by Martin

Roman Gushchin (3):
  bpf: pass struct btf pointer to the map_check_btf() callback
  bpf: add bpffs pretty print for cgroup local storage maps
  selftests/bpf: add btf annotations for cgroup_local_storage maps

 include/linux/bpf.h                       |  3 +
 include/linux/btf.h                       |  1 +
 kernel/bpf/arraymap.c                     |  1 +
 kernel/bpf/btf.c                          | 22 ++++++
 kernel/bpf/local_storage.c                | 93 ++++++++++++++++++++++-
 kernel/bpf/lpm_trie.c                     |  1 +
 kernel/bpf/syscall.c                      |  3 +-
 tools/testing/selftests/bpf/netcnt_prog.c |  6 ++
 8 files changed, 128 insertions(+), 2 deletions(-)

-- 
2.19.2

[PATCH v2 bpf-next 1/3] bpf: pass struct btf pointer to the map_check_btf() callback

[Roman Gushchin]

If key_type or value_type are of non-trivial data types
(e.g. structure or typedef), it's not possible to check them without
the additional information, which can't be obtained without a pointer
to the btf structure.

So, let's pass btf pointer to the map_check_btf() callbacks.

Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
---
 include/linux/bpf.h   | 3 +++
 kernel/bpf/arraymap.c | 1 +
 kernel/bpf/lpm_trie.c | 1 +
 kernel/bpf/syscall.c  | 3 ++-
 4 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 0c992b86eb2c..e734f163bd0b 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -23,6 +23,7 @@ struct bpf_prog;
 struct bpf_map;
 struct sock;
 struct seq_file;
+struct btf;
 struct btf_type;
 
 /* map is generic key/value storage optionally accesible by eBPF programs */
@@ -52,6 +53,7 @@ struct bpf_map_ops {
 	void (*map_seq_show_elem)(struct bpf_map *map, void *key,
 				  struct seq_file *m);
 	int (*map_check_btf)(const struct bpf_map *map,
+			     const struct btf *btf,
 			     const struct btf_type *key_type,
 			     const struct btf_type *value_type);
 };
@@ -126,6 +128,7 @@ static inline bool bpf_map_support_seq_show(const struct bpf_map *map)
 }
 
 int map_check_no_btf(const struct bpf_map *map,
+		     const struct btf *btf,
 		     const struct btf_type *key_type,
 		     const struct btf_type *value_type);
 
diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c
index 24583da9ffd1..25632a75d630 100644
--- a/kernel/bpf/arraymap.c
+++ b/kernel/bpf/arraymap.c
@@ -382,6 +382,7 @@ static void percpu_array_map_seq_show_elem(struct bpf_map *map, void *key,
 }
 
 static int array_map_check_btf(const struct bpf_map *map,
+			       const struct btf *btf,
 			       const struct btf_type *key_type,
 			       const struct btf_type *value_type)
 {
diff --git a/kernel/bpf/lpm_trie.c b/kernel/bpf/lpm_trie.c
index bfd4882e1106..abf1002080df 100644
--- a/kernel/bpf/lpm_trie.c
+++ b/kernel/bpf/lpm_trie.c
@@ -728,6 +728,7 @@ static int trie_get_next_key(struct bpf_map *map, void *_key, void *_next_key)
 }
 
 static int trie_check_btf(const struct bpf_map *map,
+			  const struct btf *btf,
 			  const struct btf_type *key_type,
 			  const struct btf_type *value_type)
 {
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index a99a23bf5910..4eb7750cb5c0 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -456,6 +456,7 @@ static int bpf_obj_name_cpy(char *dst, const char *src)
 }
 
 int map_check_no_btf(const struct bpf_map *map,
+		     const struct btf *btf,
 		     const struct btf_type *key_type,
 		     const struct btf_type *value_type)
 {
@@ -478,7 +479,7 @@ static int map_check_btf(const struct bpf_map *map, const struct btf *btf,
 		return -EINVAL;
 
 	if (map->ops->map_check_btf)
-		ret = map->ops->map_check_btf(map, key_type, value_type);
+		ret = map->ops->map_check_btf(map, btf, key_type, value_type);
 
 	return ret;
 }
-- 
2.19.2

[PATCH v2 bpf-next 2/3] bpf: add bpffs pretty print for cgroup local storage maps

[Roman Gushchin]

Implement bpffs pretty printing for cgroup local storage maps
(both shared and per-cpu).
Output example (captured for tools/testing/selftests/bpf/netcnt_prog.c):

Shared:
  $ cat /sys/fs/bpf/map_2
  # WARNING!! The output is for debug purpose only
  # WARNING!! The output format will change
  {4294968594,1}: {9999,1039896}

Per-cpu:
  $ cat /sys/fs/bpf/map_1
  # WARNING!! The output is for debug purpose only
  # WARNING!! The output format will change
  {4294968594,1}: {
  	cpu0: {0,0,0,0,0}
  	cpu1: {0,0,0,0,0}
  	cpu2: {1,104,0,0,0}
  	cpu3: {0,0,0,0,0}
  }

Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
---
 include/linux/btf.h        |  1 +
 kernel/bpf/btf.c           | 22 +++++++++
 kernel/bpf/local_storage.c | 93 +++++++++++++++++++++++++++++++++++++-
 3 files changed, 115 insertions(+), 1 deletion(-)

diff --git a/include/linux/btf.h b/include/linux/btf.h
index b98405a56383..a4cf075b89eb 100644
--- a/include/linux/btf.h
+++ b/include/linux/btf.h
@@ -47,6 +47,7 @@ void btf_type_seq_show(const struct btf *btf, u32 type_id, void *obj,
 int btf_get_fd_by_id(u32 id);
 u32 btf_id(const struct btf *btf);
 bool btf_name_offset_valid(const struct btf *btf, u32 offset);
+bool btf_type_is_reg_int(const struct btf_type *t, u32 expected_size);
 
 #ifdef CONFIG_BPF_SYSCALL
 const struct btf_type *btf_type_by_id(const struct btf *btf, u32 type_id);
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index e0a827f95e19..804ab24283f6 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -514,6 +514,28 @@ static bool btf_type_int_is_regular(const struct btf_type *t)
 	return true;
 }
 
+/*
+ * Check that given type is a regular int and has the expected size.
+ */
+bool btf_type_is_reg_int(const struct btf_type *t, u32 expected_size)
+{
+	u8 nr_bits, nr_bytes;
+	u32 int_data;
+
+	if (!btf_type_is_int(t))
+		return false;
+
+	int_data = btf_type_int(t);
+	nr_bits = BTF_INT_BITS(int_data);
+	nr_bytes = BITS_ROUNDUP_BYTES(nr_bits);
+	if (BITS_PER_BYTE_MASKED(nr_bits) ||
+	    BTF_INT_OFFSET(int_data) ||
+	    nr_bytes != expected_size)
+		return false;
+
+	return true;
+}
+
 __printf(2, 3) static void __btf_verifier_log(struct bpf_verifier_log *log,
 					      const char *fmt, ...)
 {
diff --git a/kernel/bpf/local_storage.c b/kernel/bpf/local_storage.c
index b65017dead44..5eca03da0989 100644
--- a/kernel/bpf/local_storage.c
+++ b/kernel/bpf/local_storage.c
@@ -1,11 +1,13 @@
 //SPDX-License-Identifier: GPL-2.0
 #include <linux/bpf-cgroup.h>
 #include <linux/bpf.h>
+#include <linux/btf.h>
 #include <linux/bug.h>
 #include <linux/filter.h>
 #include <linux/mm.h>
 #include <linux/rbtree.h>
 #include <linux/slab.h>
+#include <uapi/linux/btf.h>
 
 DEFINE_PER_CPU(struct bpf_cgroup_storage*, bpf_cgroup_storage[MAX_BPF_CGROUP_STORAGE_TYPE]);
 
@@ -308,6 +310,94 @@ static int cgroup_storage_delete_elem(struct bpf_map *map, void *key)
 	return -EINVAL;
 }
 
+static int cgroup_storage_check_btf(const struct bpf_map *map,
+				    const struct btf *btf,
+				    const struct btf_type *key_type,
+				    const struct btf_type *value_type)
+{
+	const struct btf_type *t;
+	struct btf_member *m;
+	u32 id, size;
+
+	/* Key is expected to be of struct bpf_cgroup_storage_key type,
+	 * which is:
+	 * struct bpf_cgroup_storage_key {
+	 *	__u64	cgroup_inode_id;
+	 *	__u32	attach_type;
+	 * };
+	 */
+
+	/*
+	 * Key_type must be a structure with two fields.
+	 */
+	if (BTF_INFO_KIND(key_type->info) != BTF_KIND_STRUCT ||
+	    BTF_INFO_VLEN(key_type->info) != 2)
+		return -EINVAL;
+
+	/*
+	 * The first field must be a 64 bit integer at 0 offset.
+	 */
+	m = (struct btf_member *)(key_type + 1);
+	if (m->offset)
+		return -EINVAL;
+	id = m->type;
+	t = btf_type_id_size(btf, &id, NULL);
+	size = FIELD_SIZEOF(struct bpf_cgroup_storage_key, cgroup_inode_id);
+	if (!t || !btf_type_is_reg_int(t, size))
+		return -EINVAL;
+
+	/*
+	 * The second field must be a 32 bit integer at 64 bit offset.
+	 */
+	m++;
+	if (m->offset != offsetof(struct bpf_cgroup_storage_key, attach_type) *
+	    BITS_PER_BYTE)
+		return -EINVAL;
+	id = m->type;
+	t = btf_type_id_size(btf, &id, NULL);
+	size = FIELD_SIZEOF(struct bpf_cgroup_storage_key, attach_type);
+	if (!t || !btf_type_is_reg_int(t, size))
+		return -EINVAL;
+
+	return 0;
+}
+
+static void cgroup_storage_seq_show_elem(struct bpf_map *map, void *_key,
+					 struct seq_file *m)
+{
+	enum bpf_cgroup_storage_type stype = cgroup_storage_type(map);
+	struct bpf_cgroup_storage_key *key = _key;
+	struct bpf_cgroup_storage *storage;
+	int cpu;
+
+	rcu_read_lock();
+	storage = cgroup_storage_lookup(map_to_storage(map), key, false);
+	if (!storage) {
+		rcu_read_unlock();
+		return;
+	}
+
+	btf_type_seq_show(map->btf, map->btf_key_type_id, key, m);
+	stype = cgroup_storage_type(map);
+	if (stype == BPF_CGROUP_STORAGE_SHARED) {
+		seq_puts(m, ": ");
+		btf_type_seq_show(map->btf, map->btf_value_type_id,
+				  &READ_ONCE(storage->buf)->data[0], m);
+		seq_puts(m, "\n");
+	} else {
+		seq_puts(m, ": {\n");
+		for_each_possible_cpu(cpu) {
+			seq_printf(m, "\tcpu%d: ", cpu);
+			btf_type_seq_show(map->btf, map->btf_value_type_id,
+					  per_cpu_ptr(storage->percpu_buf, cpu),
+					  m);
+			seq_puts(m, "\n");
+		}
+		seq_puts(m, "}\n");
+	}
+	rcu_read_unlock();
+}
+
 const struct bpf_map_ops cgroup_storage_map_ops = {
 	.map_alloc = cgroup_storage_map_alloc,
 	.map_free = cgroup_storage_map_free,
@@ -315,7 +405,8 @@ const struct bpf_map_ops cgroup_storage_map_ops = {
 	.map_lookup_elem = cgroup_storage_lookup_elem,
 	.map_update_elem = cgroup_storage_update_elem,
 	.map_delete_elem = cgroup_storage_delete_elem,
-	.map_check_btf = map_check_no_btf,
+	.map_check_btf = cgroup_storage_check_btf,
+	.map_seq_show_elem = cgroup_storage_seq_show_elem,
 };
 
 int bpf_cgroup_storage_assign(struct bpf_prog *prog, struct bpf_map *_map)
-- 
2.19.2

Re: [PATCH v2 bpf-next 2/3] bpf: add bpffs pretty print for cgroup local storage maps

[Martin Lau]

On Mon, Dec 10, 2018 at 03:43:01PM -0800, Roman Gushchin wrote:
> Implement bpffs pretty printing for cgroup local storage maps
> (both shared and per-cpu).
> Output example (captured for tools/testing/selftests/bpf/netcnt_prog.c):
> 
> Shared:
>   $ cat /sys/fs/bpf/map_2
>   # WARNING!! The output is for debug purpose only
>   # WARNING!! The output format will change
>   {4294968594,1}: {9999,1039896}
> 
> Per-cpu:
>   $ cat /sys/fs/bpf/map_1
>   # WARNING!! The output is for debug purpose only
>   # WARNING!! The output format will change
>   {4294968594,1}: {
>   	cpu0: {0,0,0,0,0}
>   	cpu1: {0,0,0,0,0}
>   	cpu2: {1,104,0,0,0}
>   	cpu3: {0,0,0,0,0}
>   }
Acked-by: Martin KaFai Lau <kafai@fb.com>

[PATCH v2 bpf-next 3/3] selftests/bpf: add btf annotations for cgroup_local_storage maps

[Roman Gushchin]

Add btf annotations to cgroup local storage maps (per-cpu and shared)
in the network packet counting example.

Signed-off-by: Roman Gushchin <guro@fb.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
---
 tools/testing/selftests/bpf/netcnt_prog.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/tools/testing/selftests/bpf/netcnt_prog.c b/tools/testing/selftests/bpf/netcnt_prog.c
index 1198abca1360..9f741e69cebe 100644
--- a/tools/testing/selftests/bpf/netcnt_prog.c
+++ b/tools/testing/selftests/bpf/netcnt_prog.c
@@ -16,12 +16,18 @@ struct bpf_map_def SEC("maps") percpu_netcnt = {
 	.value_size = sizeof(struct percpu_net_cnt),
 };
 
+BPF_ANNOTATE_KV_PAIR(percpu_netcnt, struct bpf_cgroup_storage_key,
+		     struct percpu_net_cnt);
+
 struct bpf_map_def SEC("maps") netcnt = {
 	.type = BPF_MAP_TYPE_CGROUP_STORAGE,
 	.key_size = sizeof(struct bpf_cgroup_storage_key),
 	.value_size = sizeof(struct net_cnt),
 };
 
+BPF_ANNOTATE_KV_PAIR(netcnt, struct bpf_cgroup_storage_key,
+		     struct net_cnt);
+
 SEC("cgroup/skb")
 int bpf_nextcnt(struct __sk_buff *skb)
 {
-- 
2.19.2

Re: [PATCH v2 bpf-next 3/3] selftests/bpf: add btf annotations for cgroup_local_storage maps

[Martin Lau]

On Mon, Dec 10, 2018 at 03:43:02PM -0800, Roman Gushchin wrote:
> Add btf annotations to cgroup local storage maps (per-cpu and shared)
> in the network packet counting example.
Acked-by: Martin KaFai Lau <kafai@fb.com>

Re: [PATCH v2 bpf-next 0/3] bpffs pretty print for cgroup local storage

[Daniel Borkmann]

On 12/11/2018 12:42 AM, Roman Gushchin wrote:
> This patchset implements bpffs pretty printing for cgroup local storage maps.
> 
> v2->v1:
>   - fixed size checks in cgroup_storage_check_btf(),
>   based on input from Yonghong and Martin
>   - implemented and used btf_type_is_reg_int(), suggested by Martin
> 
> Roman Gushchin (3):
>   bpf: pass struct btf pointer to the map_check_btf() callback
>   bpf: add bpffs pretty print for cgroup local storage maps
>   selftests/bpf: add btf annotations for cgroup_local_storage maps
> 
>  include/linux/bpf.h                       |  3 +
>  include/linux/btf.h                       |  1 +
>  kernel/bpf/arraymap.c                     |  1 +
>  kernel/bpf/btf.c                          | 22 ++++++
>  kernel/bpf/local_storage.c                | 93 ++++++++++++++++++++++-
>  kernel/bpf/lpm_trie.c                     |  1 +
>  kernel/bpf/syscall.c                      |  3 +-
>  tools/testing/selftests/bpf/netcnt_prog.c |  6 ++
>  8 files changed, 128 insertions(+), 2 deletions(-)
> 

Applied, thanks!