On Tue, Jul 12, 2022 at 10:59 PM Anquan Wu <leiqi96@xxxxxxxxxxx> wrote:
>
> On Tue, 2022-07-12 at 10:31 +0200, Jiri Olsa wrote:
> > On Tue, Jul 12, 2022 at 11:15:40AM +0800, Anquan Wu wrote:
> > > BPF map name is limited to BPF_OBJ_NAME_LEN.
> > > A map name is defined as being longer than BPF_OBJ_NAME_LEN,
> > > it will be truncated to BPF_OBJ_NAME_LEN when a userspace program
> > > calls libbpf to create the map. A pinned map also generates a path
> > > in the /sys. If the previous program wanted to reuse the map，
> > > it can not get bpf_map by name, because the name of the map is only
> > > partially the same as the name which get from pinned path.
> > >
> > > The syscall information below show that map name
> > > "process_pinned_map"
> > > is truncated to "process_pinned_".
> > >
> > > bpf(BPF_OBJ_GET, {pathname="/sys/fs/bpf/process_pinned_map",
> > > bpf_fd=0, file_flags=0}, 144) = -1 ENOENT (No such file or
> > > directory)
> > >
> > > bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_HASH, key_size=4,
> > > value_size=4,max_entries=1024, map_flags=0, inner_map_fd=0,
> > > map_name="process_pinned_",map_ifindex=0, btf_fd=3,
> > > btf_key_type_id=6,
> > > btf_value_type_id=10,btf_vmlinux_value_type_id=0}, 72) = 4
> > >
> > > This patch check that if the name of pinned map are the same as the
> > > actual name for the first (BPF_OBJ_NAME_LEN - 1),
> > > bpf map still uses the name which is included in bpf object.
> > >
> > > Signed-off-by: Anquan Wu <leiqi96@xxxxxxxxxxx>
> > > ---
> > >
> > > v2: compare against zero explicitly
> > >
> > > v1:
> > > https://lore.kernel.org/linux-kernel/OSZP286MB1725A2361FA2EE8432C4D5F4B8879@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/
> > > ---
> > > tools/lib/bpf/libbpf.c | 8 +++++++-
> > > 1 file changed, 7 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> > > index e89cc9c885b3..7b4d3604dfb4 100644
> > > --- a/tools/lib/bpf/libbpf.c
> > > +++ b/tools/lib/bpf/libbpf.c
> > > @@ -4328,6 +4328,7 @@ int bpf_map__reuse_fd(struct bpf_map *map,
> > > int
> > > fd)
> > > {
> > > struct bpf_map_info info = {};
> > > __u32 len = sizeof(info);
> > > + __u32 name_len;
> > > int new_fd, err;
> > > char *new_name;
> > >
> > > @@ -4337,7 +4338,12 @@ int bpf_map__reuse_fd(struct bpf_map *map,
> > > int
> > > fd)
> > > if (err)
> > > return libbpf_err(err);
> > >
> > > - new_name = strdup(info.name);
> > > + name_len = strlen(info.name);
> > > + if (name_len == BPF_OBJ_NAME_LEN - 1 && strncmp(map->name,
> > > info.name, name_len) == 0)
> >
> > so what if the map->name is different after 'name_len' ?
> >
> > jirka
> >
>
> If A map name is defined as being longer than name_len (name_len is
> "BPF_OBJ_NAME_LEN - 1" in this context), a program will fail to get a
> reused bpf_map by bpf_object__find_map_by_name().
>
> fromhttps://github.com/libbpf/libbpf/blob/master/src/libbpf.c#L9295,
> pos->name in bpf_object__find_map_by_name() is from new_name
> in
> bpf_map_reuse_fd(). It can not find map by the name which is defined
> in bpf object.
>
> I wrote some code to verify this problem and test the solution
> mentioned above.
> Link: https://github.com/leiqi96/libbpf-fix
>
It would be great to have something like this as a selftest, please
send a follow up patch adding a test under selftests/bpf for map
reuse. See prog_tests/pinning.c, this might belong there.
To also answer Jiri's question. This is not an ideal solution, but it
improves the current situation. And while potentially it's not 100%
correct (because only checks first 15 characters), user normally would
use bpf_map__reuse_fd() on well-known and presumably correct map, so
chance of misuse here are pretty minimal.
So I added
Fixes: 26736eb9a483 ("tools: libbpf: allow map reuse")
and applied to bpf-next, thanks.
> Anquan
>
>
> > > + new_name = strdup(map->name);
> > > + else
> > > + new_name = strdup(info.name);
> > > +
> > > if (!new_name)
> > > return libbpf_err(-errno);
> > >
> > > --
> > > 2.32.0
> > >
>
>
>
