netdev - Re: [PATCH bpf-next] libbpf: making bpf_prog_load() ignore name if kernel doesn't support

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <407e67b7-b4f2-40db-6e13-409784fe32aa@isovalent.com>
Date:   Fri, 12 Aug 2022 09:38:17 +0100
From:   Quentin Monnet <quentin@...valent.com>
To:     Hangbin Liu <liuhangbin@...il.com>, netdev@...r.kernel.org
Cc:     Andrii Nakryiko <andrii@...nel.org>,
        Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Martin KaFai Lau <martin.lau@...ux.dev>,
        Song Liu <song@...nel.org>, Yonghong Song <yhs@...com>,
        John Fastabend <john.fastabend@...il.com>,
        KP Singh <kpsingh@...nel.org>,
        Stanislav Fomichev <sdf@...gle.com>,
        Hao Luo <haoluo@...gle.com>, Jiri Olsa <jolsa@...nel.org>,
        bpf@...r.kernel.org
Subject: Re: [PATCH bpf-next] libbpf: making bpf_prog_load() ignore name if
 kernel doesn't support

On 12/08/2022 03:40, Hangbin Liu wrote:
> Similar with commit 10b62d6a38f7 ("libbpf: Add names for auxiliary maps"),
> let's make bpf_prog_load() also ignore name if kernel doesn't support
> program name.
> 
> To achieve this, we need to call sys_bpf_prog_load() directly in
> probe_kern_prog_name() to avoid circular dependency. sys_bpf_prog_load()
> also need to be exported in the bpf.h file.
> 
> Signed-off-by: Hangbin Liu <liuhangbin@...il.com>
> ---
>  tools/lib/bpf/bpf.c    |  6 ++----
>  tools/lib/bpf/bpf.h    |  3 +++
>  tools/lib/bpf/libbpf.c | 11 +++++++++--
>  3 files changed, 14 insertions(+), 6 deletions(-)
> 
> diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
> index 6a96e665dc5d..575867d69496 100644
> --- a/tools/lib/bpf/bpf.c
> +++ b/tools/lib/bpf/bpf.c
> @@ -84,9 +84,7 @@ static inline int sys_bpf_fd(enum bpf_cmd cmd, union bpf_attr *attr,
>  	return ensure_good_fd(fd);
>  }
>  
> -#define PROG_LOAD_ATTEMPTS 5
> -
> -static inline int sys_bpf_prog_load(union bpf_attr *attr, unsigned int size, int attempts)
> +int sys_bpf_prog_load(union bpf_attr *attr, unsigned int size, int attempts)
>  {
>  	int fd;
>  
> @@ -263,7 +261,7 @@ int bpf_prog_load(enum bpf_prog_type prog_type,
>  	attr.prog_ifindex = OPTS_GET(opts, prog_ifindex, 0);
>  	attr.kern_version = OPTS_GET(opts, kern_version, 0);
>  
> -	if (prog_name)
> +	if (prog_name && kernel_supports(NULL, FEAT_PROG_NAME))
>  		libbpf_strlcpy(attr.prog_name, prog_name, sizeof(attr.prog_name));
>  	attr.license = ptr_to_u64(license);
>  
> diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h
> index 9c50beabdd14..125c580e45f8 100644
> --- a/tools/lib/bpf/bpf.h
> +++ b/tools/lib/bpf/bpf.h
> @@ -35,6 +35,9 @@
>  extern "C" {
>  #endif
>  
> +#define PROG_LOAD_ATTEMPTS 5
> +int sys_bpf_prog_load(union bpf_attr *attr, unsigned int size, int attempts);
> +

bpf.h is the user-facing header, should these go into libbpf_internal.h
instead?

By the way, I observe that libbpf_set_memlock_rlim() in bpf.h (below) is
not prefixed with LIBBPF_API, although it is exposed in the libbpf.map,
Andrii is this expected?

>  int libbpf_set_memlock_rlim(size_t memlock_bytes);
>  
>  struct bpf_map_create_opts {
> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> index 3f01f5cd8a4c..1bcb2735d3f1 100644
> --- a/tools/lib/bpf/libbpf.c
> +++ b/tools/lib/bpf/libbpf.c
> @@ -4419,10 +4419,17 @@ static int probe_kern_prog_name(void)
>  		BPF_MOV64_IMM(BPF_REG_0, 0),
>  		BPF_EXIT_INSN(),
>  	};
> -	int ret, insn_cnt = ARRAY_SIZE(insns);
> +	union bpf_attr attr = {
> +		.prog_type = BPF_PROG_TYPE_SOCKET_FILTER,
> +		.prog_name = "test",
> +		.license = ptr_to_u64("GPL"),
> +		.insns = ptr_to_u64(insns),
> +		.insn_cnt = (__u32)ARRAY_SIZE(insns),
> +	};

I think you cannot initialise "attr" directly, you need a "memset(&attr,
0, sizeof(attr));" first, in case the struct contains padding between
the fields.

> +	int ret;
>  
>  	/* make sure loading with name works */
> -	ret = bpf_prog_load(BPF_PROG_TYPE_SOCKET_FILTER, "test", "GPL", insns, insn_cnt, NULL);
> +	ret = sys_bpf_prog_load(&attr, sizeof(attr), PROG_LOAD_ATTEMPTS);
>  	return probe_fd(ret);
>  }
>