* [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool
@ 2018-08-29 21:43 Yonghong Song
2018-08-29 21:43 ` [PATCH bpf-next 1/3] bpf: add bpffs pretty print for percpu arraymap/hash/lru_hash Yonghong Song
` (3 more replies)
0 siblings, 4 replies; 5+ messages in thread
From: Yonghong Song @ 2018-08-29 21:43 UTC (permalink / raw)
To: ast, kafai, daniel, netdev; +Cc: kernel-team
Commit a26ca7c982cb ("bpf: btf: Add pretty print support to the
basic arraymap") and Commit 699c86d6ec21 ("bpf: btf: add pretty print
for hash/lru_hash maps") added bpffs pretty print for array, hash and
lru hash maps. The pretty print gives users a structurally formatted
dump for keys/values which much easy to understand than raw bytes.
This patch set implemented bpffs pretty print support for
percpu arraymap, percpu hashmap and percpu lru hashmap.
For complex key/value types, the pretty print here is even more useful
due to
. large volumne of data making it even harder to correlate bytes
to a particular field in a particular cpu.
. kernel rounds the value size for each cpu to multiple of 8.
User has to be aware of this otherwise wrong value may be
derived from cpu 1/2/...
For example, we may have a bpffs pretty print like below:
43602: {
cpu0: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
cpu1: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
cpu2: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
cpu3: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
}
for a percpu map.
This patch also added percpu formatted print on bpftool. For example,
bpftool may print like below:
{
"key": 0,
"values": [{
"cpu": 0,
"value": {
"ui32": 0,
"ui16": 0,
}
},{
"cpu": 1,
"value": {
"ui32": 1,
"ui16": 0,
}
},{
"cpu": 2,
"value": {
"ui32": 2,
"ui16": 0,
}
},{
"cpu": 3,
"value": {
"ui32": 3,
"ui16": 0,
}
}
]
}
Patch #1 implemented bpffs pretty print for percpu arraymap/hash/lru_hash
in kernel. Patch #2 added the test case in tools bpf selftest test_btf.
Patch #3 added percpu map btf based dump.
Yonghong Song (3):
bpf: add bpffs pretty print for percpu arraymap/hash/lru_hash
tools/bpf: add bpffs percpu map pretty print tests in test_btf
tools/bpf: bpftool: add btf percpu map formated dump
kernel/bpf/arraymap.c | 24 +++++
kernel/bpf/hashtab.c | 31 ++++++
tools/bpf/bpftool/map.c | 33 +++++-
tools/testing/selftests/bpf/test_btf.c | 179 ++++++++++++++++++++++++++-------
4 files changed, 230 insertions(+), 37 deletions(-)
--
2.14.3
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH bpf-next 1/3] bpf: add bpffs pretty print for percpu arraymap/hash/lru_hash
2018-08-29 21:43 [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool Yonghong Song
@ 2018-08-29 21:43 ` Yonghong Song
2018-08-29 21:43 ` [PATCH bpf-next 2/3] tools/bpf: add bpffs percpu map pretty print tests in test_btf Yonghong Song
` (2 subsequent siblings)
3 siblings, 0 replies; 5+ messages in thread
From: Yonghong Song @ 2018-08-29 21:43 UTC (permalink / raw)
To: ast, kafai, daniel, netdev; +Cc: kernel-team
Added bpffs pretty print for percpu arraymap, percpu hashmap
and percpu lru hashmap.
For each map <key, value> pair, the format is:
<key_value>: {
cpu0: <value_on_cpu0>
cpu1: <value_on_cpu1>
...
cpun: <value_on_cpun>
}
For example, on my VM, there are 4 cpus, and
for test_btf test in the next patch:
cat /sys/fs/bpf/pprint_test_percpu_hash
You may get:
...
43602: {
cpu0: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
cpu1: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
cpu2: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
cpu3: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
}
72847: {
cpu0: {72847,0,-72847,0x3,0x11c8f,0x3,{72847|[143,28,1,0,0,0,0,0]},ENUM_THREE}
cpu1: {72847,0,-72847,0x3,0x11c8f,0x3,{72847|[143,28,1,0,0,0,0,0]},ENUM_THREE}
cpu2: {72847,0,-72847,0x3,0x11c8f,0x3,{72847|[143,28,1,0,0,0,0,0]},ENUM_THREE}
cpu3: {72847,0,-72847,0x3,0x11c8f,0x3,{72847|[143,28,1,0,0,0,0,0]},ENUM_THREE}
}
...
Signed-off-by: Yonghong Song <yhs@fb.com>
---
kernel/bpf/arraymap.c | 24 ++++++++++++++++++++++++
kernel/bpf/hashtab.c | 31 +++++++++++++++++++++++++++++++
2 files changed, 55 insertions(+)
diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c
index 0c17aab3ce5f..f9d24121be99 100644
--- a/kernel/bpf/arraymap.c
+++ b/kernel/bpf/arraymap.c
@@ -358,6 +358,29 @@ static void array_map_seq_show_elem(struct bpf_map *map, void *key,
rcu_read_unlock();
}
+static void percpu_array_map_seq_show_elem(struct bpf_map *map, void *key,
+ struct seq_file *m)
+{
+ struct bpf_array *array = container_of(map, struct bpf_array, map);
+ u32 index = *(u32 *)key;
+ void __percpu *pptr;
+ int cpu;
+
+ rcu_read_lock();
+
+ seq_printf(m, "%u: {\n", *(u32 *)key);
+ pptr = array->pptrs[index & array->index_mask];
+ 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(pptr, cpu), m);
+ seq_puts(m, "\n");
+ }
+ seq_puts(m, "}\n");
+
+ rcu_read_unlock();
+}
+
static int array_map_check_btf(const struct bpf_map *map,
const struct btf_type *key_type,
const struct btf_type *value_type)
@@ -398,6 +421,7 @@ const struct bpf_map_ops percpu_array_map_ops = {
.map_lookup_elem = percpu_array_map_lookup_elem,
.map_update_elem = array_map_update_elem,
.map_delete_elem = array_map_delete_elem,
+ .map_seq_show_elem = percpu_array_map_seq_show_elem,
.map_check_btf = array_map_check_btf,
};
diff --git a/kernel/bpf/hashtab.c b/kernel/bpf/hashtab.c
index 03cc59ee9c95..2c1790288138 100644
--- a/kernel/bpf/hashtab.c
+++ b/kernel/bpf/hashtab.c
@@ -1285,6 +1285,35 @@ int bpf_percpu_hash_update(struct bpf_map *map, void *key, void *value,
return ret;
}
+static void htab_percpu_map_seq_show_elem(struct bpf_map *map, void *key,
+ struct seq_file *m)
+{
+ struct htab_elem *l;
+ void __percpu *pptr;
+ int cpu;
+
+ rcu_read_lock();
+
+ l = __htab_map_lookup_elem(map, key);
+ if (!l) {
+ rcu_read_unlock();
+ return;
+ }
+
+ btf_type_seq_show(map->btf, map->btf_key_type_id, key, m);
+ seq_puts(m, ": {\n");
+ pptr = htab_elem_get_ptr(l, map->key_size);
+ 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(pptr, cpu), m);
+ seq_puts(m, "\n");
+ }
+ seq_puts(m, "}\n");
+
+ rcu_read_unlock();
+}
+
const struct bpf_map_ops htab_percpu_map_ops = {
.map_alloc_check = htab_map_alloc_check,
.map_alloc = htab_map_alloc,
@@ -1293,6 +1322,7 @@ const struct bpf_map_ops htab_percpu_map_ops = {
.map_lookup_elem = htab_percpu_map_lookup_elem,
.map_update_elem = htab_percpu_map_update_elem,
.map_delete_elem = htab_map_delete_elem,
+ .map_seq_show_elem = htab_percpu_map_seq_show_elem,
};
const struct bpf_map_ops htab_lru_percpu_map_ops = {
@@ -1303,6 +1333,7 @@ const struct bpf_map_ops htab_lru_percpu_map_ops = {
.map_lookup_elem = htab_lru_percpu_map_lookup_elem,
.map_update_elem = htab_lru_percpu_map_update_elem,
.map_delete_elem = htab_lru_map_delete_elem,
+ .map_seq_show_elem = htab_percpu_map_seq_show_elem,
};
static int fd_htab_map_alloc_check(union bpf_attr *attr)
--
2.14.3
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH bpf-next 2/3] tools/bpf: add bpffs percpu map pretty print tests in test_btf
2018-08-29 21:43 [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool Yonghong Song
2018-08-29 21:43 ` [PATCH bpf-next 1/3] bpf: add bpffs pretty print for percpu arraymap/hash/lru_hash Yonghong Song
@ 2018-08-29 21:43 ` Yonghong Song
2018-08-29 21:43 ` [PATCH bpf-next 3/3] tools/bpf: bpftool: add btf percpu map formated dump Yonghong Song
2018-08-30 12:25 ` [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool Daniel Borkmann
3 siblings, 0 replies; 5+ messages in thread
From: Yonghong Song @ 2018-08-29 21:43 UTC (permalink / raw)
To: ast, kafai, daniel, netdev; +Cc: kernel-team
The bpf selftest test_btf is extended to test bpffs
percpu map pretty print for percpu array, percpu hash and
percpu lru hash.
Signed-off-by: Yonghong Song <yhs@fb.com>
---
tools/testing/selftests/bpf/test_btf.c | 179 ++++++++++++++++++++++++++-------
1 file changed, 144 insertions(+), 35 deletions(-)
diff --git a/tools/testing/selftests/bpf/test_btf.c b/tools/testing/selftests/bpf/test_btf.c
index 6b5cfeb7a9cc..f42b3396d622 100644
--- a/tools/testing/selftests/bpf/test_btf.c
+++ b/tools/testing/selftests/bpf/test_btf.c
@@ -4,6 +4,7 @@
#include <linux/bpf.h>
#include <linux/btf.h>
#include <linux/err.h>
+#include <linux/kernel.h>
#include <bpf/bpf.h>
#include <sys/resource.h>
#include <libelf.h>
@@ -45,7 +46,6 @@ static int count_result(int err)
return err;
}
-#define min(a, b) ((a) < (b) ? (a) : (b))
#define __printf(a, b) __attribute__((format(printf, a, b)))
__printf(1, 2)
@@ -130,6 +130,7 @@ struct btf_raw_test {
bool map_create_err;
bool ordered_map;
bool lossless_map;
+ bool percpu_map;
int hdr_len_delta;
int type_off_delta;
int str_off_delta;
@@ -2157,6 +2158,7 @@ static struct btf_pprint_test_meta {
const char *map_name;
bool ordered_map;
bool lossless_map;
+ bool percpu_map;
} pprint_tests_meta[] = {
{
.descr = "BTF pretty print array",
@@ -2164,6 +2166,7 @@ static struct btf_pprint_test_meta {
.map_name = "pprint_test_array",
.ordered_map = true,
.lossless_map = true,
+ .percpu_map = false,
},
{
@@ -2172,6 +2175,7 @@ static struct btf_pprint_test_meta {
.map_name = "pprint_test_hash",
.ordered_map = false,
.lossless_map = true,
+ .percpu_map = false,
},
{
@@ -2180,30 +2184,83 @@ static struct btf_pprint_test_meta {
.map_name = "pprint_test_lru_hash",
.ordered_map = false,
.lossless_map = false,
+ .percpu_map = false,
+},
+
+{
+ .descr = "BTF pretty print percpu array",
+ .map_type = BPF_MAP_TYPE_PERCPU_ARRAY,
+ .map_name = "pprint_test_percpu_array",
+ .ordered_map = true,
+ .lossless_map = true,
+ .percpu_map = true,
+},
+
+{
+ .descr = "BTF pretty print percpu hash",
+ .map_type = BPF_MAP_TYPE_PERCPU_HASH,
+ .map_name = "pprint_test_percpu_hash",
+ .ordered_map = false,
+ .lossless_map = true,
+ .percpu_map = true,
+},
+
+{
+ .descr = "BTF pretty print lru percpu hash",
+ .map_type = BPF_MAP_TYPE_LRU_PERCPU_HASH,
+ .map_name = "pprint_test_lru_percpu_hash",
+ .ordered_map = false,
+ .lossless_map = false,
+ .percpu_map = true,
},
};
-static void set_pprint_mapv(struct pprint_mapv *v, uint32_t i)
+static void set_pprint_mapv(struct pprint_mapv *v, uint32_t i,
+ int num_cpus, int rounded_value_size)
{
- v->ui32 = i;
- v->si32 = -i;
- v->unused_bits2a = 3;
- v->bits28 = i;
- v->unused_bits2b = 3;
- v->ui64 = i;
- v->aenum = i & 0x03;
+ int cpu;
+
+ for (cpu = 0; cpu < num_cpus; cpu++) {
+ v->ui32 = i + cpu;
+ v->si32 = -i;
+ v->unused_bits2a = 3;
+ v->bits28 = i;
+ v->unused_bits2b = 3;
+ v->ui64 = i;
+ v->aenum = i & 0x03;
+ v = (void *)v + rounded_value_size;
+ }
}
+static int check_line(const char *expected_line, int nexpected_line,
+ int expected_line_len, const char *line)
+{
+ if (CHECK(nexpected_line == expected_line_len,
+ "expected_line is too long"))
+ return -1;
+
+ if (strcmp(expected_line, line)) {
+ fprintf(stderr, "unexpected pprint output\n");
+ fprintf(stderr, "expected: %s", expected_line);
+ fprintf(stderr, " read: %s", line);
+ return -1;
+ }
+
+ return 0;
+}
+
+
static int do_test_pprint(void)
{
const struct btf_raw_test *test = &pprint_test_template;
struct bpf_create_map_attr create_attr = {};
+ bool ordered_map, lossless_map, percpu_map;
+ int err, ret, num_cpus, rounded_value_size;
+ struct pprint_mapv *mapv = NULL;
unsigned int key, nr_read_elems;
- bool ordered_map, lossless_map;
int map_fd = -1, btf_fd = -1;
- struct pprint_mapv mapv = {};
unsigned int raw_btf_size;
char expected_line[255];
FILE *pin_file = NULL;
@@ -2212,7 +2269,6 @@ static int do_test_pprint(void)
char *line = NULL;
uint8_t *raw_btf;
ssize_t nread;
- int err, ret;
fprintf(stderr, "%s......", test->descr);
raw_btf = btf_raw_create(&hdr_tmpl, test->raw_types,
@@ -2261,9 +2317,18 @@ static int do_test_pprint(void)
if (CHECK(err, "bpf_obj_pin(%s): errno:%d.", pin_path, errno))
goto done;
+ percpu_map = test->percpu_map;
+ num_cpus = percpu_map ? bpf_num_possible_cpus() : 1;
+ rounded_value_size = round_up(sizeof(struct pprint_mapv), 8);
+ mapv = calloc(num_cpus, rounded_value_size);
+ if (CHECK(!mapv, "mapv allocation failure")) {
+ err = -1;
+ goto done;
+ }
+
for (key = 0; key < test->max_entries; key++) {
- set_pprint_mapv(&mapv, key);
- bpf_map_update_elem(map_fd, &key, &mapv, 0);
+ set_pprint_mapv(mapv, key, num_cpus, rounded_value_size);
+ bpf_map_update_elem(map_fd, &key, mapv, 0);
}
pin_file = fopen(pin_path, "r");
@@ -2286,33 +2351,74 @@ static int do_test_pprint(void)
ordered_map = test->ordered_map;
lossless_map = test->lossless_map;
do {
+ struct pprint_mapv *cmapv;
ssize_t nexpected_line;
unsigned int next_key;
+ int cpu;
next_key = ordered_map ? nr_read_elems : atoi(line);
- set_pprint_mapv(&mapv, next_key);
- nexpected_line = snprintf(expected_line, sizeof(expected_line),
- "%u: {%u,0,%d,0x%x,0x%x,0x%x,{%lu|[%u,%u,%u,%u,%u,%u,%u,%u]},%s}\n",
- next_key,
- mapv.ui32, mapv.si32,
- mapv.unused_bits2a, mapv.bits28, mapv.unused_bits2b,
- mapv.ui64,
- mapv.ui8a[0], mapv.ui8a[1], mapv.ui8a[2], mapv.ui8a[3],
- mapv.ui8a[4], mapv.ui8a[5], mapv.ui8a[6], mapv.ui8a[7],
- pprint_enum_str[mapv.aenum]);
-
- if (CHECK(nexpected_line == sizeof(expected_line),
- "expected_line is too long")) {
- err = -1;
- goto done;
+ set_pprint_mapv(mapv, next_key, num_cpus, rounded_value_size);
+ cmapv = mapv;
+
+ for (cpu = 0; cpu < num_cpus; cpu++) {
+ if (percpu_map) {
+ /* for percpu map, the format looks like:
+ * <key>: {
+ * cpu0: <value_on_cpu0>
+ * cpu1: <value_on_cpu1>
+ * ...
+ * cpun: <value_on_cpun>
+ * }
+ *
+ * let us verify the line containing the key here.
+ */
+ if (cpu == 0) {
+ nexpected_line = snprintf(expected_line,
+ sizeof(expected_line),
+ "%u: {\n",
+ next_key);
+
+ err = check_line(expected_line, nexpected_line,
+ sizeof(expected_line), line);
+ if (err == -1)
+ goto done;
+ }
+
+ /* read value@cpu */
+ nread = getline(&line, &line_len, pin_file);
+ if (nread < 0)
+ break;
+ }
+
+ nexpected_line = snprintf(expected_line, sizeof(expected_line),
+ "%s%u: {%u,0,%d,0x%x,0x%x,0x%x,"
+ "{%lu|[%u,%u,%u,%u,%u,%u,%u,%u]},%s}\n",
+ percpu_map ? "\tcpu" : "",
+ percpu_map ? cpu : next_key,
+ cmapv->ui32, cmapv->si32,
+ cmapv->unused_bits2a,
+ cmapv->bits28,
+ cmapv->unused_bits2b,
+ cmapv->ui64,
+ cmapv->ui8a[0], cmapv->ui8a[1],
+ cmapv->ui8a[2], cmapv->ui8a[3],
+ cmapv->ui8a[4], cmapv->ui8a[5],
+ cmapv->ui8a[6], cmapv->ui8a[7],
+ pprint_enum_str[cmapv->aenum]);
+
+ err = check_line(expected_line, nexpected_line,
+ sizeof(expected_line), line);
+ if (err == -1)
+ goto done;
+
+ cmapv = (void *)cmapv + rounded_value_size;
}
- if (strcmp(expected_line, line)) {
- err = -1;
- fprintf(stderr, "unexpected pprint output\n");
- fprintf(stderr, "expected: %s", expected_line);
- fprintf(stderr, " read: %s", line);
- goto done;
+ if (percpu_map) {
+ /* skip the last bracket for the percpu map */
+ nread = getline(&line, &line_len, pin_file);
+ if (nread < 0)
+ break;
}
nread = getline(&line, &line_len, pin_file);
@@ -2334,6 +2440,8 @@ static int do_test_pprint(void)
err = 0;
done:
+ if (mapv)
+ free(mapv);
if (!err)
fprintf(stderr, "OK");
if (*btf_log_buf && (err || args.always_log))
@@ -2361,6 +2469,7 @@ static int test_pprint(void)
pprint_test_template.map_name = pprint_tests_meta[i].map_name;
pprint_test_template.ordered_map = pprint_tests_meta[i].ordered_map;
pprint_test_template.lossless_map = pprint_tests_meta[i].lossless_map;
+ pprint_test_template.percpu_map = pprint_tests_meta[i].percpu_map;
err |= count_result(do_test_pprint());
}
--
2.14.3
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH bpf-next 3/3] tools/bpf: bpftool: add btf percpu map formated dump
2018-08-29 21:43 [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool Yonghong Song
2018-08-29 21:43 ` [PATCH bpf-next 1/3] bpf: add bpffs pretty print for percpu arraymap/hash/lru_hash Yonghong Song
2018-08-29 21:43 ` [PATCH bpf-next 2/3] tools/bpf: add bpffs percpu map pretty print tests in test_btf Yonghong Song
@ 2018-08-29 21:43 ` Yonghong Song
2018-08-30 12:25 ` [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool Daniel Borkmann
3 siblings, 0 replies; 5+ messages in thread
From: Yonghong Song @ 2018-08-29 21:43 UTC (permalink / raw)
To: ast, kafai, daniel, netdev; +Cc: kernel-team
The btf pretty print is added to percpu arraymap,
percpu hashmap and percpu lru hashmap.
For each <key, value> pair, the following will be
added to plain/json output:
{
"key": <pretty_print_key>,
"values": [{
"cpu": 0,
"value": <pretty_print_value_on_cpu0>
},{
"cpu": 1,
"value": <pretty_print_value_on_cpu1>
},{
....
},{
"cpu": n,
"value": <pretty_print_value_on_cpun>
}
]
}
For example, the following could be part of plain or json formatted
output:
{
"key": 0,
"values": [{
"cpu": 0,
"value": {
"ui32": 0,
"ui16": 0,
}
},{
"cpu": 1,
"value": {
"ui32": 1,
"ui16": 0,
}
},{
"cpu": 2,
"value": {
"ui32": 2,
"ui16": 0,
}
},{
"cpu": 3,
"value": {
"ui32": 3,
"ui16": 0,
}
}
]
}
Signed-off-by: Yonghong Song <yhs@fb.com>
---
tools/bpf/bpftool/map.c | 33 +++++++++++++++++++++++++++++++--
1 file changed, 31 insertions(+), 2 deletions(-)
diff --git a/tools/bpf/bpftool/map.c b/tools/bpf/bpftool/map.c
index b2ec20e562bd..df175bc33c5d 100644
--- a/tools/bpf/bpftool/map.c
+++ b/tools/bpf/bpftool/map.c
@@ -169,9 +169,28 @@ static int do_dump_btf(const struct btf_dumper *d,
if (ret)
goto err_end_obj;
- jsonw_name(d->jw, "value");
+ 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);
+ } else {
+ unsigned int i, n, step;
- ret = btf_dumper_type(d, map_info->btf_value_type_id, value);
+ jsonw_name(d->jw, "values");
+ jsonw_start_array(d->jw);
+ n = get_possible_cpus();
+ step = round_up(map_info->value_size, 8);
+ for (i = 0; i < n; i++) {
+ 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);
+ jsonw_end_object(d->jw);
+ if (ret)
+ break;
+ }
+ jsonw_end_array(d->jw);
+ }
err_end_obj:
/* end of key-value pair */
@@ -298,6 +317,16 @@ static void print_entry_json(struct bpf_map_info *info, unsigned char *key,
jsonw_end_object(json_wtr);
}
jsonw_end_array(json_wtr);
+ if (btf) {
+ struct btf_dumper d = {
+ .btf = btf,
+ .jw = json_wtr,
+ .is_plain_text = false,
+ };
+
+ jsonw_name(json_wtr, "formatted");
+ do_dump_btf(&d, info, key, value);
+ }
}
jsonw_end_object(json_wtr);
--
2.14.3
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool
2018-08-29 21:43 [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool Yonghong Song
` (2 preceding siblings ...)
2018-08-29 21:43 ` [PATCH bpf-next 3/3] tools/bpf: bpftool: add btf percpu map formated dump Yonghong Song
@ 2018-08-30 12:25 ` Daniel Borkmann
3 siblings, 0 replies; 5+ messages in thread
From: Daniel Borkmann @ 2018-08-30 12:25 UTC (permalink / raw)
To: Yonghong Song, ast, kafai, netdev; +Cc: kernel-team
On 08/29/2018 11:43 PM, Yonghong Song wrote:
> Commit a26ca7c982cb ("bpf: btf: Add pretty print support to the
> basic arraymap") and Commit 699c86d6ec21 ("bpf: btf: add pretty print
> for hash/lru_hash maps") added bpffs pretty print for array, hash and
> lru hash maps. The pretty print gives users a structurally formatted
> dump for keys/values which much easy to understand than raw bytes.
>
> This patch set implemented bpffs pretty print support for
> percpu arraymap, percpu hashmap and percpu lru hashmap.
> For complex key/value types, the pretty print here is even more useful
> due to
> . large volumne of data making it even harder to correlate bytes
> to a particular field in a particular cpu.
> . kernel rounds the value size for each cpu to multiple of 8.
> User has to be aware of this otherwise wrong value may be
> derived from cpu 1/2/...
>
> For example, we may have a bpffs pretty print like below:
> 43602: {
> cpu0: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
> cpu1: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
> cpu2: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
> cpu3: {43602,0,-43602,0x3,0xaa52,0x3,{43602|[82,170,0,0,0,0,0,0]},ENUM_TWO}
> }
> for a percpu map.
>
> This patch also added percpu formatted print on bpftool. For example,
> bpftool may print like below:
> {
> "key": 0,
> "values": [{
> "cpu": 0,
> "value": {
> "ui32": 0,
> "ui16": 0,
> }
> },{
> "cpu": 1,
> "value": {
> "ui32": 1,
> "ui16": 0,
> }
> },{
> "cpu": 2,
> "value": {
> "ui32": 2,
> "ui16": 0,
> }
> },{
> "cpu": 3,
> "value": {
> "ui32": 3,
> "ui16": 0,
> }
> }
> ]
> }
>
> Patch #1 implemented bpffs pretty print for percpu arraymap/hash/lru_hash
> in kernel. Patch #2 added the test case in tools bpf selftest test_btf.
> Patch #3 added percpu map btf based dump.
>
> Yonghong Song (3):
> bpf: add bpffs pretty print for percpu arraymap/hash/lru_hash
> tools/bpf: add bpffs percpu map pretty print tests in test_btf
> tools/bpf: bpftool: add btf percpu map formated dump
>
> kernel/bpf/arraymap.c | 24 +++++
> kernel/bpf/hashtab.c | 31 ++++++
> tools/bpf/bpftool/map.c | 33 +++++-
> tools/testing/selftests/bpf/test_btf.c | 179 ++++++++++++++++++++++++++-------
> 4 files changed, 230 insertions(+), 37 deletions(-)
Applied to bpf-next, thanks Yonghong!
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2018-08-30 16:27 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-08-29 21:43 [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool Yonghong Song
2018-08-29 21:43 ` [PATCH bpf-next 1/3] bpf: add bpffs pretty print for percpu arraymap/hash/lru_hash Yonghong Song
2018-08-29 21:43 ` [PATCH bpf-next 2/3] tools/bpf: add bpffs percpu map pretty print tests in test_btf Yonghong Song
2018-08-29 21:43 ` [PATCH bpf-next 3/3] tools/bpf: bpftool: add btf percpu map formated dump Yonghong Song
2018-08-30 12:25 ` [PATCH bpf-next 0/3] bpf: implement percpu map pretty print for bpffs and bpftool Daniel Borkmann
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.