[PATCHv3 bpf-next] libbpf: Add names for auxiliary maps

[PATCHv3 bpf-next] libbpf: Add names for auxiliary maps

[Hangbin Liu]

The bpftool self-created maps can appear in final map show output due to
deferred removal in kernel. These maps don't have a name, which would make
users confused about where it comes from.

With a libbpf_ prefix name, users could know who created these maps.
It also could make some tests (like test_offload.py, which skip base maps
without names as a workaround) filter them out.

Kernel adds bpf prog/map name support in the same merge
commit fadad670a8ab ("Merge branch 'bpf-extend-info'"). So we can also use
kernel_supports(NULL, FEAT_PROG_NAME) to check if kernel supports map name.

As disscussed[1], Let's make bpf_map_create accept non-null
name string, and silently ignore the name if kernel doesn't support.

[1] https://lore.kernel.org/bpf/CAEf4BzYL1TQwo1231s83pjTdFPk9XWWhfZC5=KzkU-VO0k=0Ug@mail.gmail.com/

Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
---
v3: let bpf_map_create ignore the name if kernel doesn't support
v2: rename the wrapper with proper name
---
 tools/lib/bpf/bpf.c    | 2 +-
 tools/lib/bpf/libbpf.c | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
index efcc06dafbd9..6a96e665dc5d 100644
--- a/tools/lib/bpf/bpf.c
+++ b/tools/lib/bpf/bpf.c
@@ -183,7 +183,7 @@ int bpf_map_create(enum bpf_map_type map_type,
 		return libbpf_err(-EINVAL);
 
 	attr.map_type = map_type;
-	if (map_name)
+	if (map_name && kernel_supports(NULL, FEAT_PROG_NAME))
 		libbpf_strlcpy(attr.map_name, map_name, sizeof(attr.map_name));
 	attr.key_size = key_size;
 	attr.value_size = value_size;
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index f7364ea82ac1..a075211b3730 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -4432,7 +4432,7 @@ static int probe_kern_global_data(void)
 	};
 	int ret, map, insn_cnt = ARRAY_SIZE(insns);
 
-	map = bpf_map_create(BPF_MAP_TYPE_ARRAY, NULL, sizeof(int), 32, 1, NULL);
+	map = bpf_map_create(BPF_MAP_TYPE_ARRAY, "libbpf_global", sizeof(int), 32, 1, NULL);
 	if (map < 0) {
 		ret = -errno;
 		cp = libbpf_strerror_r(ret, errmsg, sizeof(errmsg));
@@ -4565,7 +4565,7 @@ static int probe_kern_array_mmap(void)
 	LIBBPF_OPTS(bpf_map_create_opts, opts, .map_flags = BPF_F_MMAPABLE);
 	int fd;
 
-	fd = bpf_map_create(BPF_MAP_TYPE_ARRAY, NULL, sizeof(int), sizeof(int), 1, &opts);
+	fd = bpf_map_create(BPF_MAP_TYPE_ARRAY, "libbpf_mmap", sizeof(int), sizeof(int), 1, &opts);
 	return probe_fd(fd);
 }
 
@@ -4612,7 +4612,7 @@ static int probe_prog_bind_map(void)
 	};
 	int ret, map, prog, insn_cnt = ARRAY_SIZE(insns);
 
-	map = bpf_map_create(BPF_MAP_TYPE_ARRAY, NULL, sizeof(int), 32, 1, NULL);
+	map = bpf_map_create(BPF_MAP_TYPE_ARRAY, "libbpf_det_bind", sizeof(int), 32, 1, NULL);
 	if (map < 0) {
 		ret = -errno;
 		cp = libbpf_strerror_r(ret, errmsg, sizeof(errmsg));
-- 
2.35.3

Re: [PATCHv3 bpf-next] libbpf: Add names for auxiliary maps

[Andrii Nakryiko]

On Wed, Aug 10, 2022 at 8:40 PM Hangbin Liu <liuhangbin@gmail.com> wrote:
>
> The bpftool self-created maps can appear in final map show output due to
> deferred removal in kernel. These maps don't have a name, which would make
> users confused about where it comes from.
>
> With a libbpf_ prefix name, users could know who created these maps.
> It also could make some tests (like test_offload.py, which skip base maps
> without names as a workaround) filter them out.
>
> Kernel adds bpf prog/map name support in the same merge
> commit fadad670a8ab ("Merge branch 'bpf-extend-info'"). So we can also use
> kernel_supports(NULL, FEAT_PROG_NAME) to check if kernel supports map name.
>
> As disscussed[1], Let's make bpf_map_create accept non-null
> name string, and silently ignore the name if kernel doesn't support.
>
> [1] https://lore.kernel.org/bpf/CAEf4BzYL1TQwo1231s83pjTdFPk9XWWhfZC5=KzkU-VO0k=0Ug@mail.gmail.com/
>
> Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
> ---
> v3: let bpf_map_create ignore the name if kernel doesn't support
> v2: rename the wrapper with proper name
> ---

Applied to bpf-next. But let's also do the same for bpf_prog_load().
We can make probe_kern_prog_name() use sys_bpf_prog_load() directly
and avoid calling bpf_prog_load() and thus avoiding circular
dependency.

Will you be able to do this change?


>  tools/lib/bpf/bpf.c    | 2 +-
>  tools/lib/bpf/libbpf.c | 6 +++---
>  2 files changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
> index efcc06dafbd9..6a96e665dc5d 100644
> --- a/tools/lib/bpf/bpf.c
> +++ b/tools/lib/bpf/bpf.c
> @@ -183,7 +183,7 @@ int bpf_map_create(enum bpf_map_type map_type,
>                 return libbpf_err(-EINVAL);
>
>         attr.map_type = map_type;
> -       if (map_name)
> +       if (map_name && kernel_supports(NULL, FEAT_PROG_NAME))
>                 libbpf_strlcpy(attr.map_name, map_name, sizeof(attr.map_name));
>         attr.key_size = key_size;
>         attr.value_size = value_size;
> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> index f7364ea82ac1..a075211b3730 100644
> --- a/tools/lib/bpf/libbpf.c
> +++ b/tools/lib/bpf/libbpf.c
> @@ -4432,7 +4432,7 @@ static int probe_kern_global_data(void)
>         };
>         int ret, map, insn_cnt = ARRAY_SIZE(insns);
>
> -       map = bpf_map_create(BPF_MAP_TYPE_ARRAY, NULL, sizeof(int), 32, 1, NULL);
> +       map = bpf_map_create(BPF_MAP_TYPE_ARRAY, "libbpf_global", sizeof(int), 32, 1, NULL);
>         if (map < 0) {
>                 ret = -errno;
>                 cp = libbpf_strerror_r(ret, errmsg, sizeof(errmsg));
> @@ -4565,7 +4565,7 @@ static int probe_kern_array_mmap(void)
>         LIBBPF_OPTS(bpf_map_create_opts, opts, .map_flags = BPF_F_MMAPABLE);
>         int fd;
>
> -       fd = bpf_map_create(BPF_MAP_TYPE_ARRAY, NULL, sizeof(int), sizeof(int), 1, &opts);
> +       fd = bpf_map_create(BPF_MAP_TYPE_ARRAY, "libbpf_mmap", sizeof(int), sizeof(int), 1, &opts);
>         return probe_fd(fd);
>  }
>
> @@ -4612,7 +4612,7 @@ static int probe_prog_bind_map(void)
>         };
>         int ret, map, prog, insn_cnt = ARRAY_SIZE(insns);
>
> -       map = bpf_map_create(BPF_MAP_TYPE_ARRAY, NULL, sizeof(int), 32, 1, NULL);
> +       map = bpf_map_create(BPF_MAP_TYPE_ARRAY, "libbpf_det_bind", sizeof(int), 32, 1, NULL);
>         if (map < 0) {
>                 ret = -errno;
>                 cp = libbpf_strerror_r(ret, errmsg, sizeof(errmsg));
> --
> 2.35.3
>

Re: [PATCHv3 bpf-next] libbpf: Add names for auxiliary maps

[Hangbin Liu]

On Fri, Aug 12, 2022 at 6:15 AM Andrii Nakryiko
<andrii.nakryiko@gmail.com> wrote:
>
> Applied to bpf-next. But let's also do the same for bpf_prog_load().
> We can make probe_kern_prog_name() use sys_bpf_prog_load() directly
> and avoid calling bpf_prog_load() and thus avoiding circular
> dependency.

Ah, yes, we can do it this way. I will post the patch today.

Re: [PATCHv3 bpf-next] libbpf: Add names for auxiliary maps

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to bpf/bpf-next.git (master)
by Andrii Nakryiko <andrii@kernel.org>:

On Thu, 11 Aug 2022 11:40:20 +0800 you wrote:
> The bpftool self-created maps can appear in final map show output due to
> deferred removal in kernel. These maps don't have a name, which would make
> users confused about where it comes from.
> 
> With a libbpf_ prefix name, users could know who created these maps.
> It also could make some tests (like test_offload.py, which skip base maps
> without names as a workaround) filter them out.
> 
> [...]

Here is the summary with links:
  - [PATCHv3,bpf-next] libbpf: Add names for auxiliary maps
    https://git.kernel.org/bpf/bpf-next/c/10b62d6a38f7

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html