[PATCH v2 bpf-next 0/5] bpftool: Support dumping a map with btf_vmlinux_value_type_id

[PATCH v2 bpf-next 0/5] bpftool: Support dumping a map with btf_vmlinux_value_type_id

[Martin KaFai Lau]

When a map is storing a kernel's struct, its
map_info->btf_vmlinux_value_type_id is set.  The first map type
supporting it is BPF_MAP_TYPE_STRUCT_OPS.

This series adds support to dump this kind of map with BTF.
The first two patches are bug fixes which only applicable to
in bpf-next.

Please see individual patches for details.

v2:
- Expose bpf_find_kernel_btf() as a LIBBPF_API in patch 3 (Andrii)
- Cache btf_vmlinux in bpftool/map.c (Andrii)

Martin KaFai Lau (5):
  bpftool: Fix a leak of btf object
  bpftool: Fix missing BTF output for json during map dump
  libbpf: Expose bpf_find_kernel_btf as a LIBBPF_API
  bpftool: Add struct_ops map name
  bpftool: Support dumping a map with btf_vmlinux_value_type_id

 tools/bpf/bpftool/map.c  | 103 ++++++++++++++++++++++++---------------
 tools/lib/bpf/btf.c      | 102 +++++++++++++++++++++++++++++++++++---
 tools/lib/bpf/btf.h      |   2 +
 tools/lib/bpf/libbpf.c   |  93 ++---------------------------------
 tools/lib/bpf/libbpf.map |   1 +
 5 files changed, 167 insertions(+), 134 deletions(-)

-- 
2.17.1

[PATCH v2 bpf-next 1/5] bpftool: Fix a leak of btf object

[Martin KaFai Lau]

When testing a map has btf or not, maps_have_btf() tests it by actually
getting a btf_fd from sys_bpf(BPF_BTF_GET_FD_BY_ID). However, it
forgot to btf__free() it.

In maps_have_btf() stage, there is no need to test it by really
calling sys_bpf(BPF_BTF_GET_FD_BY_ID). Testing non zero
info.btf_id is good enough.

Also, the err_close case is unnecessary, and also causes double
close() because the calling func do_dump() will close() all fds again.

Fixes: 99f9863a0c45 ("bpftool: Match maps by name")
Cc: Paul Chaignon <paul.chaignon@orange.com>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 tools/bpf/bpftool/map.c | 16 ++--------------
 1 file changed, 2 insertions(+), 14 deletions(-)

diff --git a/tools/bpf/bpftool/map.c b/tools/bpf/bpftool/map.c
index c01f76fa6876..e00e9e19d6b7 100644
--- a/tools/bpf/bpftool/map.c
+++ b/tools/bpf/bpftool/map.c
@@ -915,32 +915,20 @@ static int maps_have_btf(int *fds, int nb_fds)
 {
 	struct bpf_map_info info = {};
 	__u32 len = sizeof(info);
-	struct btf *btf = NULL;
 	int err, i;
 
 	for (i = 0; i < nb_fds; i++) {
 		err = bpf_obj_get_info_by_fd(fds[i], &info, &len);
 		if (err) {
 			p_err("can't get map info: %s", strerror(errno));
-			goto err_close;
-		}
-
-		err = btf__get_from_id(info.btf_id, &btf);
-		if (err) {
-			p_err("failed to get btf");
-			goto err_close;
+			return -1;
 		}
 
-		if (!btf)
+		if (!info.btf_id)
 			return 0;
 	}
 
 	return 1;
-
-err_close:
-	for (; i < nb_fds; i++)
-		close(fds[i]);
-	return -1;
 }
 
 static int
-- 
2.17.1

[PATCH v2 bpf-next 2/5] bpftool: Fix missing BTF output for json during map dump

[Martin KaFai Lau]

The btf availability check is only done for plain text output.
It causes the whole BTF output went missing when json_output
is used.

This patch simplifies the logic a little by avoiding passing "int btf" to
map_dump().

For plain text output, the btf_wtr is only created when the map has
BTF (i.e. info->btf_id != 0).  The nullness of "json_writer_t *wtr"
in map_dump() alone can decide if dumping BTF output is needed.
As long as wtr is not NULL, map_dump() will print out the BTF-described
data whenever a map has BTF available (i.e. info->btf_id != 0)
regardless of json or plain-text output.

In do_dump(), the "int btf" is also renamed to "int do_plain_btf".

Fixes: 99f9863a0c45 ("bpftool: Match maps by name")
Cc: Paul Chaignon <paul.chaignon@orange.com>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 tools/bpf/bpftool/map.c | 42 ++++++++++++++++++++---------------------
 1 file changed, 20 insertions(+), 22 deletions(-)

diff --git a/tools/bpf/bpftool/map.c b/tools/bpf/bpftool/map.c
index e00e9e19d6b7..45c1eda6512c 100644
--- a/tools/bpf/bpftool/map.c
+++ b/tools/bpf/bpftool/map.c
@@ -933,7 +933,7 @@ static int maps_have_btf(int *fds, int nb_fds)
 
 static int
 map_dump(int fd, struct bpf_map_info *info, json_writer_t *wtr,
-	 bool enable_btf, bool show_header)
+	 bool show_header)
 {
 	void *key, *value, *prev_key;
 	unsigned int num_elems = 0;
@@ -950,18 +950,16 @@ map_dump(int fd, struct bpf_map_info *info, json_writer_t *wtr,
 
 	prev_key = NULL;
 
-	if (enable_btf) {
-		err = btf__get_from_id(info->btf_id, &btf);
-		if (err || !btf) {
-			/* enable_btf is true only if we've already checked
-			 * that all maps have BTF information.
-			 */
-			p_err("failed to get btf");
-			goto exit_free;
+	if (wtr) {
+		if (info->btf_id) {
+			err = btf__get_from_id(info->btf_id, &btf);
+			if (err || !btf) {
+				err = err ? : -ESRCH;
+				p_err("failed to get btf");
+				goto exit_free;
+			}
 		}
-	}
 
-	if (wtr) {
 		if (show_header) {
 			jsonw_start_object(wtr);	/* map object */
 			show_map_header_json(info, wtr);
@@ -1009,7 +1007,7 @@ static int do_dump(int argc, char **argv)
 {
 	json_writer_t *wtr = NULL, *btf_wtr = NULL;
 	struct bpf_map_info info = {};
-	int nb_fds, i = 0, btf = 0;
+	int nb_fds, i = 0;
 	__u32 len = sizeof(info);
 	int *fds = NULL;
 	int err = -1;
@@ -1029,17 +1027,17 @@ static int do_dump(int argc, char **argv)
 	if (json_output) {
 		wtr = json_wtr;
 	} else {
-		btf = maps_have_btf(fds, nb_fds);
-		if (btf < 0)
+		int do_plain_btf;
+
+		do_plain_btf = maps_have_btf(fds, nb_fds);
+		if (do_plain_btf < 0)
 			goto exit_close;
-		if (btf) {
+
+		if (do_plain_btf) {
 			btf_wtr = get_btf_writer();
-			if (btf_wtr) {
-				wtr = btf_wtr;
-			} else {
+			wtr = btf_wtr;
+			if (!btf_wtr)
 				p_info("failed to create json writer for btf. falling back to plain output");
-				btf = 0;
-			}
 		}
 	}
 
@@ -1050,7 +1048,7 @@ static int do_dump(int argc, char **argv)
 			p_err("can't get map info: %s", strerror(errno));
 			break;
 		}
-		err = map_dump(fds[i], &info, wtr, btf, nb_fds > 1);
+		err = map_dump(fds[i], &info, wtr, nb_fds > 1);
 		if (!wtr && i != nb_fds - 1)
 			printf("\n");
 
@@ -1061,7 +1059,7 @@ static int do_dump(int argc, char **argv)
 	if (wtr && nb_fds > 1)
 		jsonw_end_array(wtr);	/* root array */
 
-	if (btf)
+	if (btf_wtr)
 		jsonw_destroy(&btf_wtr);
 exit_close:
 	for (; i < nb_fds; i++)
-- 
2.17.1

[PATCH v2 bpf-next 3/5] libbpf: Expose bpf_find_kernel_btf as a LIBBPF_API

[Martin KaFai Lau]

This patch exposes bpf_find_kernel_btf() as a LIBBPF_API.
It will be used in 'bpftool map dump' in a following patch
to dump a map with btf_vmlinux_value_type_id set.

bpf_find_kernel_btf() is renamed to libbpf_find_kernel_btf()
and moved to btf.c.  As <linux/kernel.h> is included,
some of the max/min type casting needs to be fixed.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 tools/lib/bpf/btf.c      | 102 ++++++++++++++++++++++++++++++++++++---
 tools/lib/bpf/btf.h      |   2 +
 tools/lib/bpf/libbpf.c   |  93 ++---------------------------------
 tools/lib/bpf/libbpf.map |   1 +
 4 files changed, 102 insertions(+), 96 deletions(-)

diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c
index cfeb6a44480b..3d1c25fc97ae 100644
--- a/tools/lib/bpf/btf.c
+++ b/tools/lib/bpf/btf.c
@@ -8,6 +8,10 @@
 #include <fcntl.h>
 #include <unistd.h>
 #include <errno.h>
+#include <sys/utsname.h>
+#include <sys/param.h>
+#include <sys/stat.h>
+#include <linux/kernel.h>
 #include <linux/err.h>
 #include <linux/btf.h>
 #include <gelf.h>
@@ -20,8 +24,8 @@
 /* make sure libbpf doesn't use kernel-only integer typedefs */
 #pragma GCC poison u8 u16 u32 u64 s8 s16 s32 s64
 
-#define BTF_MAX_NR_TYPES 0x7fffffff
-#define BTF_MAX_STR_OFFSET 0x7fffffff
+#define BTF_MAX_NR_TYPES 0x7fffffffU
+#define BTF_MAX_STR_OFFSET 0x7fffffffU
 
 static struct btf_type btf_void;
 
@@ -53,7 +57,7 @@ static int btf_add_type(struct btf *btf, struct btf_type *t)
 		if (btf->types_size == BTF_MAX_NR_TYPES)
 			return -E2BIG;
 
-		expand_by = max(btf->types_size >> 2, 16);
+		expand_by = max(btf->types_size >> 2, 16U);
 		new_size = min(BTF_MAX_NR_TYPES, btf->types_size + expand_by);
 
 		new_types = realloc(btf->types, sizeof(*new_types) * new_size);
@@ -289,7 +293,7 @@ int btf__align_of(const struct btf *btf, __u32 id)
 	switch (kind) {
 	case BTF_KIND_INT:
 	case BTF_KIND_ENUM:
-		return min(sizeof(void *), t->size);
+		return min(sizeof(void *), (size_t)t->size);
 	case BTF_KIND_PTR:
 		return sizeof(void *);
 	case BTF_KIND_TYPEDEF:
@@ -1401,7 +1405,7 @@ static int btf_dedup_hypot_map_add(struct btf_dedup *d,
 	if (d->hypot_cnt == d->hypot_cap) {
 		__u32 *new_list;
 
-		d->hypot_cap += max(16, d->hypot_cap / 2);
+		d->hypot_cap += max((size_t)16, d->hypot_cap / 2);
 		new_list = realloc(d->hypot_list, sizeof(__u32) * d->hypot_cap);
 		if (!new_list)
 			return -ENOMEM;
@@ -1697,7 +1701,7 @@ static int btf_dedup_strings(struct btf_dedup *d)
 		if (strs.cnt + 1 > strs.cap) {
 			struct btf_str_ptr *new_ptrs;
 
-			strs.cap += max(strs.cnt / 2, 16);
+			strs.cap += max(strs.cnt / 2, 16U);
 			new_ptrs = realloc(strs.ptrs,
 					   sizeof(strs.ptrs[0]) * strs.cap);
 			if (!new_ptrs) {
@@ -2931,3 +2935,89 @@ static int btf_dedup_remap_types(struct btf_dedup *d)
 	}
 	return 0;
 }
+
+static struct btf *btf_load_raw(const char *path)
+{
+	struct btf *btf;
+	size_t read_cnt;
+	struct stat st;
+	void *data;
+	FILE *f;
+
+	if (stat(path, &st))
+		return ERR_PTR(-errno);
+
+	data = malloc(st.st_size);
+	if (!data)
+		return ERR_PTR(-ENOMEM);
+
+	f = fopen(path, "rb");
+	if (!f) {
+		btf = ERR_PTR(-errno);
+		goto cleanup;
+	}
+
+	read_cnt = fread(data, 1, st.st_size, f);
+	fclose(f);
+	if (read_cnt < st.st_size) {
+		btf = ERR_PTR(-EBADF);
+		goto cleanup;
+	}
+
+	btf = btf__new(data, read_cnt);
+
+cleanup:
+	free(data);
+	return btf;
+}
+
+/*
+ * Probe few well-known locations for vmlinux kernel image and try to load BTF
+ * data out of it to use for target BTF.
+ */
+struct btf *libbpf_find_kernel_btf(void)
+{
+	struct {
+		const char *path_fmt;
+		bool raw_btf;
+	} locations[] = {
+		/* try canonical vmlinux BTF through sysfs first */
+		{ "/sys/kernel/btf/vmlinux", true /* raw BTF */ },
+		/* fall back to trying to find vmlinux ELF on disk otherwise */
+		{ "/boot/vmlinux-%1$s" },
+		{ "/lib/modules/%1$s/vmlinux-%1$s" },
+		{ "/lib/modules/%1$s/build/vmlinux" },
+		{ "/usr/lib/modules/%1$s/kernel/vmlinux" },
+		{ "/usr/lib/debug/boot/vmlinux-%1$s" },
+		{ "/usr/lib/debug/boot/vmlinux-%1$s.debug" },
+		{ "/usr/lib/debug/lib/modules/%1$s/vmlinux" },
+	};
+	char path[PATH_MAX + 1];
+	struct utsname buf;
+	struct btf *btf;
+	int i;
+
+	uname(&buf);
+
+	for (i = 0; i < ARRAY_SIZE(locations); i++) {
+		snprintf(path, PATH_MAX, locations[i].path_fmt, buf.release);
+
+		if (access(path, R_OK))
+			continue;
+
+		if (locations[i].raw_btf)
+			btf = btf_load_raw(path);
+		else
+			btf = btf__parse_elf(path, NULL);
+
+		pr_debug("loading kernel BTF '%s': %ld\n",
+			 path, IS_ERR(btf) ? PTR_ERR(btf) : 0);
+		if (IS_ERR(btf))
+			continue;
+
+		return btf;
+	}
+
+	pr_warn("failed to find valid kernel BTF\n");
+	return ERR_PTR(-ESRCH);
+}
diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h
index 8d73f7f5551f..70c1b7ec2bd0 100644
--- a/tools/lib/bpf/btf.h
+++ b/tools/lib/bpf/btf.h
@@ -102,6 +102,8 @@ LIBBPF_API int btf_ext__reloc_line_info(const struct btf *btf,
 LIBBPF_API __u32 btf_ext__func_info_rec_size(const struct btf_ext *btf_ext);
 LIBBPF_API __u32 btf_ext__line_info_rec_size(const struct btf_ext *btf_ext);
 
+LIBBPF_API struct btf *libbpf_find_kernel_btf(void);
+
 struct btf_dedup_opts {
 	unsigned int dedup_table_size;
 	bool dont_resolve_fwds;
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 23868883477f..3afaca9bce1d 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -73,7 +73,6 @@
 
 #define __printf(a, b)	__attribute__((format(printf, a, b)))
 
-static struct btf *bpf_find_kernel_btf(void);
 static struct bpf_map *bpf_object__add_map(struct bpf_object *obj);
 static struct bpf_program *bpf_object__find_prog_by_idx(struct bpf_object *obj,
 							int idx);
@@ -848,7 +847,7 @@ static int bpf_object__init_kern_struct_ops_maps(struct bpf_object *obj)
 			continue;
 
 		if (!kern_btf) {
-			kern_btf = bpf_find_kernel_btf();
+			kern_btf = libbpf_find_kernel_btf();
 			if (IS_ERR(kern_btf))
 				return PTR_ERR(kern_btf);
 		}
@@ -4300,92 +4299,6 @@ static int bpf_core_reloc_insn(struct bpf_program *prog,
 	return 0;
 }
 
-static struct btf *btf_load_raw(const char *path)
-{
-	struct btf *btf;
-	size_t read_cnt;
-	struct stat st;
-	void *data;
-	FILE *f;
-
-	if (stat(path, &st))
-		return ERR_PTR(-errno);
-
-	data = malloc(st.st_size);
-	if (!data)
-		return ERR_PTR(-ENOMEM);
-
-	f = fopen(path, "rb");
-	if (!f) {
-		btf = ERR_PTR(-errno);
-		goto cleanup;
-	}
-
-	read_cnt = fread(data, 1, st.st_size, f);
-	fclose(f);
-	if (read_cnt < st.st_size) {
-		btf = ERR_PTR(-EBADF);
-		goto cleanup;
-	}
-
-	btf = btf__new(data, read_cnt);
-
-cleanup:
-	free(data);
-	return btf;
-}
-
-/*
- * Probe few well-known locations for vmlinux kernel image and try to load BTF
- * data out of it to use for target BTF.
- */
-static struct btf *bpf_find_kernel_btf(void)
-{
-	struct {
-		const char *path_fmt;
-		bool raw_btf;
-	} locations[] = {
-		/* try canonical vmlinux BTF through sysfs first */
-		{ "/sys/kernel/btf/vmlinux", true /* raw BTF */ },
-		/* fall back to trying to find vmlinux ELF on disk otherwise */
-		{ "/boot/vmlinux-%1$s" },
-		{ "/lib/modules/%1$s/vmlinux-%1$s" },
-		{ "/lib/modules/%1$s/build/vmlinux" },
-		{ "/usr/lib/modules/%1$s/kernel/vmlinux" },
-		{ "/usr/lib/debug/boot/vmlinux-%1$s" },
-		{ "/usr/lib/debug/boot/vmlinux-%1$s.debug" },
-		{ "/usr/lib/debug/lib/modules/%1$s/vmlinux" },
-	};
-	char path[PATH_MAX + 1];
-	struct utsname buf;
-	struct btf *btf;
-	int i;
-
-	uname(&buf);
-
-	for (i = 0; i < ARRAY_SIZE(locations); i++) {
-		snprintf(path, PATH_MAX, locations[i].path_fmt, buf.release);
-
-		if (access(path, R_OK))
-			continue;
-
-		if (locations[i].raw_btf)
-			btf = btf_load_raw(path);
-		else
-			btf = btf__parse_elf(path, NULL);
-
-		pr_debug("loading kernel BTF '%s': %ld\n",
-			 path, IS_ERR(btf) ? PTR_ERR(btf) : 0);
-		if (IS_ERR(btf))
-			continue;
-
-		return btf;
-	}
-
-	pr_warn("failed to find valid kernel BTF\n");
-	return ERR_PTR(-ESRCH);
-}
-
 /* Output spec definition in the format:
  * [<type-id>] (<type-name>) + <raw-spec> => <offset>@<spec>,
  * where <spec> is a C-syntax view of recorded field access, e.g.: x.a[3].b
@@ -4620,7 +4533,7 @@ bpf_core_reloc_fields(struct bpf_object *obj, const char *targ_btf_path)
 	if (targ_btf_path)
 		targ_btf = btf__parse_elf(targ_btf_path, NULL);
 	else
-		targ_btf = bpf_find_kernel_btf();
+		targ_btf = libbpf_find_kernel_btf();
 	if (IS_ERR(targ_btf)) {
 		pr_warn("failed to get target BTF: %ld\n", PTR_ERR(targ_btf));
 		return PTR_ERR(targ_btf);
@@ -6595,7 +6508,7 @@ static int bpf_object__collect_struct_ops_map_reloc(struct bpf_object *obj,
 int libbpf_find_vmlinux_btf_id(const char *name,
 			       enum bpf_attach_type attach_type)
 {
-	struct btf *btf = bpf_find_kernel_btf();
+	struct btf *btf = libbpf_find_kernel_btf();
 	char raw_tp_btf[128] = BTF_PREFIX;
 	char *dst = raw_tp_btf + sizeof(BTF_PREFIX) - 1;
 	const char *btf_name;
diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
index a19f04e6e3d9..fad8fefacfdb 100644
--- a/tools/lib/bpf/libbpf.map
+++ b/tools/lib/bpf/libbpf.map
@@ -227,4 +227,5 @@ LIBBPF_0.0.7 {
 		bpf_program__is_struct_ops;
 		bpf_program__set_struct_ops;
 		btf__align_of;
+		libbpf_find_kernel_btf;
 } LIBBPF_0.0.6;
-- 
2.17.1

[PATCH v2 bpf-next 4/5] bpftool: Add struct_ops map name

[Martin KaFai Lau]

This patch adds BPF_MAP_TYPE_STRUCT_OPS to "struct_ops" name mapping
so that "bpftool map show" can print the "struct_ops" map type
properly.

[root@arch-fb-vm1 bpf]# ~/devshare/fb-kernel/linux/tools/bpf/bpftool/bpftool map show id 8
8: struct_ops  name dctcp  flags 0x0
	key 4B  value 256B  max_entries 1  memlock 4096B
	btf_id 7

Acked-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 tools/bpf/bpftool/map.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tools/bpf/bpftool/map.c b/tools/bpf/bpftool/map.c
index 45c1eda6512c..4c5b15d736b6 100644
--- a/tools/bpf/bpftool/map.c
+++ b/tools/bpf/bpftool/map.c
@@ -48,6 +48,7 @@ const char * const map_type_name[] = {
 	[BPF_MAP_TYPE_QUEUE]			= "queue",
 	[BPF_MAP_TYPE_STACK]			= "stack",
 	[BPF_MAP_TYPE_SK_STORAGE]		= "sk_storage",
+	[BPF_MAP_TYPE_STRUCT_OPS]		= "struct_ops",
 };
 
 const size_t map_type_name_size = ARRAY_SIZE(map_type_name);
-- 
2.17.1

[PATCH v2 bpf-next 5/5] bpftool: Support dumping a map with btf_vmlinux_value_type_id

[Martin KaFai Lau]

This patch makes bpftool support dumping a map's value properly
when the map's value type is a type of the running kernel's btf.
(i.e. map_info.btf_vmlinux_value_type_id is set instead of
map_info.btf_value_type_id).  The first usecase is for the
BPF_MAP_TYPE_STRUCT_OPS.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 tools/bpf/bpftool/map.c | 62 +++++++++++++++++++++++++++++++++--------
 1 file changed, 51 insertions(+), 11 deletions(-)

diff --git a/tools/bpf/bpftool/map.c b/tools/bpf/bpftool/map.c
index 4c5b15d736b6..33636f86bcbb 100644
--- a/tools/bpf/bpftool/map.c
+++ b/tools/bpf/bpftool/map.c
@@ -20,6 +20,7 @@
 #include "btf.h"
 #include "json_writer.h"
 #include "main.h"
+#include "libbpf_internal.h"
 
 const char * const map_type_name[] = {
 	[BPF_MAP_TYPE_UNSPEC]			= "unspec",
@@ -252,6 +253,7 @@ static int do_dump_btf(const struct btf_dumper *d,
 		       struct bpf_map_info *map_info, void *key,
 		       void *value)
 {
+	__u32 value_id;
 	int ret;
 
 	/* start of key-value pair */
@@ -265,9 +267,12 @@ static int do_dump_btf(const struct btf_dumper *d,
 			goto err_end_obj;
 	}
 
+	value_id = map_info->btf_vmlinux_value_type_id ?
+		: map_info->btf_value_type_id;
+
 	if (!map_is_per_cpu(map_info->type)) {
 		jsonw_name(d->jw, "value");
-		ret = btf_dumper_type(d, map_info->btf_value_type_id, value);
+		ret = btf_dumper_type(d, value_id, value);
 	} else {
 		unsigned int i, n, step;
 
@@ -279,8 +284,7 @@ static int do_dump_btf(const struct btf_dumper *d,
 			jsonw_start_object(d->jw);
 			jsonw_int_field(d->jw, "cpu", i);
 			jsonw_name(d->jw, "value");
-			ret = btf_dumper_type(d, map_info->btf_value_type_id,
-					      value + i * step);
+			ret = btf_dumper_type(d, value_id, value + i * step);
 			jsonw_end_object(d->jw);
 			if (ret)
 				break;
@@ -932,6 +936,44 @@ static int maps_have_btf(int *fds, int nb_fds)
 	return 1;
 }
 
+static struct btf *btf_vmlinux;
+
+static struct btf *get_map_kv_btf(const struct bpf_map_info *info)
+{
+	struct btf *btf = NULL;
+
+	if (info->btf_vmlinux_value_type_id) {
+		if (!btf_vmlinux) {
+			btf_vmlinux = libbpf_find_kernel_btf();
+			if (IS_ERR(btf_vmlinux))
+				p_err("failed to get kernel btf");
+		}
+		return btf_vmlinux;
+	} else if (info->btf_value_type_id) {
+		int err;
+
+		err = btf__get_from_id(info->btf_id, &btf);
+		if (err || !btf) {
+			p_err("failed to get btf");
+			btf = err ? ERR_PTR(err) : ERR_PTR(-ESRCH);
+		}
+	}
+
+	return btf;
+}
+
+static void free_map_kv_btf(struct btf *btf)
+{
+	if (!IS_ERR(btf) && btf != btf_vmlinux)
+		btf__free(btf);
+}
+
+static void free_btf_vmlinux(void)
+{
+	if (!IS_ERR(btf_vmlinux))
+		btf__free(btf_vmlinux);
+}
+
 static int
 map_dump(int fd, struct bpf_map_info *info, json_writer_t *wtr,
 	 bool show_header)
@@ -952,13 +994,10 @@ map_dump(int fd, struct bpf_map_info *info, json_writer_t *wtr,
 	prev_key = NULL;
 
 	if (wtr) {
-		if (info->btf_id) {
-			err = btf__get_from_id(info->btf_id, &btf);
-			if (err || !btf) {
-				err = err ? : -ESRCH;
-				p_err("failed to get btf");
-				goto exit_free;
-			}
+		btf = get_map_kv_btf(info);
+		if (IS_ERR(btf)) {
+			err = PTR_ERR(btf);
+			goto exit_free;
 		}
 
 		if (show_header) {
@@ -999,7 +1038,7 @@ map_dump(int fd, struct bpf_map_info *info, json_writer_t *wtr,
 	free(key);
 	free(value);
 	close(fd);
-	btf__free(btf);
+	free_map_kv_btf(btf);
 
 	return err;
 }
@@ -1067,6 +1106,7 @@ static int do_dump(int argc, char **argv)
 		close(fds[i]);
 exit_free:
 	free(fds);
+	free_btf_vmlinux();
 	return err;
 }
 
-- 
2.17.1

Re: [PATCH v2 bpf-next 3/5] libbpf: Expose bpf_find_kernel_btf as a LIBBPF_API

[Andrii Nakryiko]

On Wed, Jan 15, 2020 at 2:26 PM Martin KaFai Lau <kafai@fb.com> wrote:
>
> This patch exposes bpf_find_kernel_btf() as a LIBBPF_API.
> It will be used in 'bpftool map dump' in a following patch
> to dump a map with btf_vmlinux_value_type_id set.
>
> bpf_find_kernel_btf() is renamed to libbpf_find_kernel_btf()
> and moved to btf.c.  As <linux/kernel.h> is included,
> some of the max/min type casting needs to be fixed.
>
> Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> ---

Thanks!

Acked-by: Andrii Nakryiko <andriin@fb.com>

>  tools/lib/bpf/btf.c      | 102 ++++++++++++++++++++++++++++++++++++---
>  tools/lib/bpf/btf.h      |   2 +
>  tools/lib/bpf/libbpf.c   |  93 ++---------------------------------
>  tools/lib/bpf/libbpf.map |   1 +
>  4 files changed, 102 insertions(+), 96 deletions(-)
>

[...]

Re: [PATCH v2 bpf-next 5/5] bpftool: Support dumping a map with btf_vmlinux_value_type_id

[Andrii Nakryiko]

On Wed, Jan 15, 2020 at 2:28 PM Martin KaFai Lau <kafai@fb.com> wrote:
>
> This patch makes bpftool support dumping a map's value properly
> when the map's value type is a type of the running kernel's btf.
> (i.e. map_info.btf_vmlinux_value_type_id is set instead of
> map_info.btf_value_type_id).  The first usecase is for the
> BPF_MAP_TYPE_STRUCT_OPS.
>
> Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> ---

LGTM.

Acked-by: Andrii Nakryiko <andriin@fb.com>

>  tools/bpf/bpftool/map.c | 62 +++++++++++++++++++++++++++++++++--------
>  1 file changed, 51 insertions(+), 11 deletions(-)
>

[...]

Re: [PATCH v2 bpf-next 5/5] bpftool: Support dumping a map with btf_vmlinux_value_type_id

[Martin Lau]

On Wed, Jan 15, 2020 at 02:46:10PM -0800, Andrii Nakryiko wrote:
> On Wed, Jan 15, 2020 at 2:28 PM Martin KaFai Lau <kafai@fb.com> wrote:
> >
> > This patch makes bpftool support dumping a map's value properly
> > when the map's value type is a type of the running kernel's btf.
> > (i.e. map_info.btf_vmlinux_value_type_id is set instead of
> > map_info.btf_value_type_id).  The first usecase is for the
> > BPF_MAP_TYPE_STRUCT_OPS.
> >
> > Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> > ---
> 
> LGTM.
> 
> Acked-by: Andrii Nakryiko <andriin@fb.com>
Thanks for the review!

I just noticied I forgot to remove the #include "libbpf_internal.h".
I will respin one more.

Re: [PATCH v2 bpf-next 5/5] bpftool: Support dumping a map with btf_vmlinux_value_type_id

[Andrii Nakryiko]

On Wed, Jan 15, 2020 at 2:50 PM Martin Lau <kafai@fb.com> wrote:
>
> On Wed, Jan 15, 2020 at 02:46:10PM -0800, Andrii Nakryiko wrote:
> > On Wed, Jan 15, 2020 at 2:28 PM Martin KaFai Lau <kafai@fb.com> wrote:
> > >
> > > This patch makes bpftool support dumping a map's value properly
> > > when the map's value type is a type of the running kernel's btf.
> > > (i.e. map_info.btf_vmlinux_value_type_id is set instead of
> > > map_info.btf_value_type_id).  The first usecase is for the
> > > BPF_MAP_TYPE_STRUCT_OPS.
> > >
> > > Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> > > ---
> >
> > LGTM.
> >
> > Acked-by: Andrii Nakryiko <andriin@fb.com>
> Thanks for the review!
>
> I just noticied I forgot to remove the #include "libbpf_internal.h".
> I will respin one more.

didn't notice that. Please also update a subject on patch exposing
libbpf_find_kernel_btf (it still mentions old name).

Re: [PATCH v2 bpf-next 5/5] bpftool: Support dumping a map with btf_vmlinux_value_type_id

[Martin Lau]

On Wed, Jan 15, 2020 at 02:59:59PM -0800, Andrii Nakryiko wrote:
> On Wed, Jan 15, 2020 at 2:50 PM Martin Lau <kafai@fb.com> wrote:
> >
> > On Wed, Jan 15, 2020 at 02:46:10PM -0800, Andrii Nakryiko wrote:
> > > On Wed, Jan 15, 2020 at 2:28 PM Martin KaFai Lau <kafai@fb.com> wrote:
> > > >
> > > > This patch makes bpftool support dumping a map's value properly
> > > > when the map's value type is a type of the running kernel's btf.
> > > > (i.e. map_info.btf_vmlinux_value_type_id is set instead of
> > > > map_info.btf_value_type_id).  The first usecase is for the
> > > > BPF_MAP_TYPE_STRUCT_OPS.
> > > >
> > > > Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> > > > ---
> > >
> > > LGTM.
> > >
> > > Acked-by: Andrii Nakryiko <andriin@fb.com>
> > Thanks for the review!
> >
> > I just noticied I forgot to remove the #include "libbpf_internal.h".
> > I will respin one more.
> 
> didn't notice that. Please also update a subject on patch exposing
> libbpf_find_kernel_btf (it still mentions old name).
oops. already went out. :p
The commit message mentioned the renaming from old to new name.

Re: [PATCH v2 bpf-next 5/5] bpftool: Support dumping a map with btf_vmlinux_value_type_id

[Andrii Nakryiko]

On Wed, Jan 15, 2020 at 3:12 PM Martin Lau <kafai@fb.com> wrote:
>
> On Wed, Jan 15, 2020 at 02:59:59PM -0800, Andrii Nakryiko wrote:
> > On Wed, Jan 15, 2020 at 2:50 PM Martin Lau <kafai@fb.com> wrote:
> > >
> > > On Wed, Jan 15, 2020 at 02:46:10PM -0800, Andrii Nakryiko wrote:
> > > > On Wed, Jan 15, 2020 at 2:28 PM Martin KaFai Lau <kafai@fb.com> wrote:
> > > > >
> > > > > This patch makes bpftool support dumping a map's value properly
> > > > > when the map's value type is a type of the running kernel's btf.
> > > > > (i.e. map_info.btf_vmlinux_value_type_id is set instead of
> > > > > map_info.btf_value_type_id).  The first usecase is for the
> > > > > BPF_MAP_TYPE_STRUCT_OPS.
> > > > >
> > > > > Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> > > > > ---
> > > >
> > > > LGTM.
> > > >
> > > > Acked-by: Andrii Nakryiko <andriin@fb.com>
> > > Thanks for the review!
> > >
> > > I just noticied I forgot to remove the #include "libbpf_internal.h".
> > > I will respin one more.
> >
> > didn't notice that. Please also update a subject on patch exposing
> > libbpf_find_kernel_btf (it still mentions old name).
> oops. already went out. :p
> The commit message mentioned the renaming from old to new name.

no worries