netdev - Re: [PATCH bpf-next v4 2/3] bpftool: conditionally append / to the progtype

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: <2f2fd146-222a-ecdb-7fe1-d9f67f5ac1de@isovalent.com>
Date:   Thu, 21 Oct 2021 20:55:52 +0100
From:   Quentin Monnet <quentin@...valent.com>
To:     Stanislav Fomichev <sdf@...gle.com>, netdev@...r.kernel.org,
        bpf@...r.kernel.org
Cc:     ast@...nel.org, daniel@...earbox.net, andrii@...nel.org
Subject: Re: [PATCH bpf-next v4 2/3] bpftool: conditionally append / to the
 progtype

2021-10-21 09:56 UTC-0700 ~ Stanislav Fomichev <sdf@...gle.com>
> Otherwise, attaching with bpftool doesn't work with strict section names.
> 
> Also, switch to libbpf strict mode to use the latest conventions
> (note, I don't think we have any cli api guarantees?).
> 
> Cc: Quentin Monnet <quentin@...valent.com>
> Signed-off-by: Stanislav Fomichev <sdf@...gle.com>
> ---
>  tools/bpf/bpftool/main.c | 4 ++++
>  tools/bpf/bpftool/prog.c | 9 +++++++--
>  2 files changed, 11 insertions(+), 2 deletions(-)
> 
> diff --git a/tools/bpf/bpftool/main.c b/tools/bpf/bpftool/main.c
> index 02eaaf065f65..8223bac1e401 100644
> --- a/tools/bpf/bpftool/main.c
> +++ b/tools/bpf/bpftool/main.c
> @@ -409,6 +409,10 @@ int main(int argc, char **argv)
>  	block_mount = false;
>  	bin_name = argv[0];
>  
> +	ret = libbpf_set_strict_mode(LIBBPF_STRICT_ALL);
> +	if (ret)
> +		p_err("failed to enable libbpf strict mode: %d", ret);
> +
>  	hash_init(prog_table.table);
>  	hash_init(map_table.table);
>  	hash_init(link_table.table);
> diff --git a/tools/bpf/bpftool/prog.c b/tools/bpf/bpftool/prog.c
> index 277d51c4c5d9..b04990588ccf 100644
> --- a/tools/bpf/bpftool/prog.c
> +++ b/tools/bpf/bpftool/prog.c
> @@ -1420,8 +1420,13 @@ static int load_with_options(int argc, char **argv, bool first_prog_only)
>  			err = get_prog_type_by_name(type, &common_prog_type,
>  						    &expected_attach_type);
>  			free(type);
> -			if (err < 0)
> -				goto err_free_reuse_maps;

Thanks a lot for the change! Can you please test it for e.g. an XDP
program? You should see that "bpftool prog load prog.o <path> type xdp"
prints a debug message from libbpf about the first attempt (above)
failing, before the second attempt (below) succeeds.

We need to get rid of this message. I think it should be easy, because
we explicitly "ask" for that message in get_prog_type_by_name(), in the
same file, if it fails to load in the first place.

Could you please update get_prog_type_by_name() to take an additional
switch as an argument, to tell if the debug-info should be retrieved
(then first attempt here would skip it, second would keep it)?
An alternative could be to move all the '/' and retries handling to that
function, and I think it would end up in bpftool keeping support for the
legacy object files with the former convention - but that would somewhat
defeat the objectives of the strict mode, so maybe not the best option.

> +			if (err < 0) {

We could run the second attempt only on libbpf returning -ESRCH, maybe?

> +				err = get_prog_type_by_name(*argv, &common_prog_type,
> +							    &expected_attach_type);
> +				if (err < 0)
> +
> +					goto err_free_reuse_maps;
> +			}
>  
>  			NEXT_ARG();
>  		} else if (is_prefix(*argv, "map")) {
>