Thread (21 messages) flat view 21 messages, 3 authors, 2019-10-23

Re: [PATCH bpf-next 1/3] libbpf: Store map pin path in struct bpf_map

From: Andrii Nakryiko <hidden>
Date: 2019-10-22 17:37:30
Also in: bpf

On Tue, Oct 22, 2019 at 9:08 AM Toke Høiland-Jørgensen [off-list ref] wrote:
quoted hunk ↗ jump to hunk
From: Toke Høiland-Jørgensen <redacted>

When pinning a map, store the pin path in struct bpf_map so it can be
re-used later for un-pinning. This simplifies the later addition of per-map
pin paths.

Signed-off-by: Toke Høiland-Jørgensen <redacted>
---
 tools/lib/bpf/libbpf.c |   19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index cccfd9355134..b4fdd8ee3bbd 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -226,6 +226,7 @@ struct bpf_map {
        void *priv;
        bpf_map_clear_priv_t clear_priv;
        enum libbpf_map_type libbpf_type;
+       char *pin_path;
 };

 struct bpf_secdata {
@@ -1929,6 +1930,7 @@ int bpf_map__reuse_fd(struct bpf_map *map, int fd)
        if (err)
                goto err_close_new_fd;
        free(map->name);
+       zfree(&map->pin_path);
While you are touching this function, can you please also fix error
handling in it? We should store -errno locally on error, before we
call close() which might change errno.
quoted hunk ↗ jump to hunk
        map->fd = new_fd;
        map->name = new_name;
@@ -4022,6 +4024,7 @@ int bpf_map__pin(struct bpf_map *map, const char *path)
                return -errno;
        }

+       map->pin_path = strdup(path);
if (!map->pin_path) {
    err = -errno;
    goto err_close_new_fd;
}

quoted hunk ↗ jump to hunk
        pr_debug("pinned map '%s'\n", path);

        return 0;
@@ -4031,6 +4034,9 @@ int bpf_map__unpin(struct bpf_map *map, const char *path)
 {
        int err;

+       if (!path)
+               path = map->pin_path;
This semantics is kind of weird. Given we now remember pin_path,
should we instead check that user-provided path is actually correct
and matches what we stored? Alternatively, bpf_map__unpin() w/o path
argument looks like a cleaner API.
quoted hunk ↗ jump to hunk
+
        err = check_path(path);
        if (err)
                return err;
@@ -4044,6 +4050,7 @@ int bpf_map__unpin(struct bpf_map *map, const char *path)
        if (err != 0)
                return -errno;
        pr_debug("unpinned map '%s'\n", path);
+       zfree(&map->pin_path);

        return 0;
 }
@@ -4088,17 +4095,10 @@ int bpf_object__pin_maps(struct bpf_object *obj, const char *path)

 err_unpin_maps:
        while ((map = bpf_map__prev(map, obj))) {
-               char buf[PATH_MAX];
-               int len;
-
-               len = snprintf(buf, PATH_MAX, "%s/%s", path,
-                              bpf_map__name(map));
-               if (len < 0)
-                       continue;
-               else if (len >= PATH_MAX)
+               if (!map->pin_path)
                        continue;

-               bpf_map__unpin(map, buf);
+               bpf_map__unpin(map, NULL);
        }

        return err;
@@ -4248,6 +4248,7 @@ void bpf_object__close(struct bpf_object *obj)

        for (i = 0; i < obj->nr_maps; i++) {
                zfree(&obj->maps[i].name);
+               zfree(&obj->maps[i].pin_path);
                if (obj->maps[i].clear_priv)
                        obj->maps[i].clear_priv(&obj->maps[i],
                                                obj->maps[i].priv);
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help