netdev - Re: [PATCH net-next] libbpf: use map

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: <CAEfhGiwai3fZsWPx--Axyc1n4+WRwxeFO+uLgQKbQ-8wiJqtAw@mail.gmail.com>
Date:   Thu, 28 Sep 2017 13:33:00 -0400
From:   Craig Gallek <kraigatgoog@...il.com>
To:     Daniel Borkmann <daniel@...earbox.net>
Cc:     Alexei Starovoitov <ast@...com>,
        "David S . Miller" <davem@...emloft.net>,
        Chonggang Li <chonggangli@...gle.com>,
        netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH net-next] libbpf: use map_flags when creating maps

On Wed, Sep 27, 2017 at 6:03 PM, Daniel Borkmann <daniel@...earbox.net> wrote:
> On 09/27/2017 06:29 PM, Alexei Starovoitov wrote:
>>
>> On 9/27/17 7:04 AM, Craig Gallek wrote:
>>>
>>> From: Craig Gallek <kraig@...gle.com>
>>>
>>> This extends struct bpf_map_def to include a flags field.  Note that
>>> this has the potential to break the validation logic in
>>> bpf_object__validate_maps and bpf_object__init_maps as they use
>>> sizeof(struct bpf_map_def) as a minimal allowable size of a map section.
>>> Any bpf program compiled with a smaller struct bpf_map_def will fail this
>>> check.
>>>
>>> I don't believe this will be an issue in practice as both compile-time
>>> definitions of struct bpf_map_def (in samples/bpf/bpf_load.h and
>>> tools/testing/selftests/bpf/bpf_helpers.h) have always been larger
>>> than this newly updated version in libbpf.h.
>>>
>>> Signed-off-by: Craig Gallek <kraig@...gle.com>
>>> ---
>>>  tools/lib/bpf/libbpf.c | 2 +-
>>>  tools/lib/bpf/libbpf.h | 1 +
>>>  2 files changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
>>> index 35f6dfcdc565..6bea85f260a3 100644
>>> --- a/tools/lib/bpf/libbpf.c
>>> +++ b/tools/lib/bpf/libbpf.c
>>> @@ -874,7 +874,7 @@ bpf_object__create_maps(struct bpf_object *obj)
>>>                        def->key_size,
>>>                        def->value_size,
>>>                        def->max_entries,
>>> -                      0);
>>> +                      def->map_flags);
>>>          if (*pfd < 0) {
>>>              size_t j;
>>>              int err = *pfd;
>>> diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
>>> index 7959086eb9c9..6e20003109e0 100644
>>> --- a/tools/lib/bpf/libbpf.h
>>> +++ b/tools/lib/bpf/libbpf.h
>>> @@ -207,6 +207,7 @@ struct bpf_map_def {
>>>      unsigned int key_size;
>>>      unsigned int value_size;
>>>      unsigned int max_entries;
>>> +    unsigned int map_flags;
>>>  };
>>
>>
>> yes it will break loading of pre-compiled .o
>> Instead of breaking, let's fix the loader to do it the way
>> samples/bpf/bpf_load.c does.
>> See commit 156450d9d964 ("samples/bpf: make bpf_load.c code compatible
>> with ELF maps section changes")
>
>
> +1, iproute2 loader also does map spec fixup
>
> For libbpf it would be good also such that it reduces the diff
> further between the libbpf and bpf_load so that it allows move
> to libbpf for samples in future.

Fair enough, I'll try to get this to work more dynamically.  I did
noticed that the fields of struct bpf_map_def in
selftests/.../bpf_helpers.h and iproute2's struct bpf_elf_map have
diverged. The flags field is the only thing missing from libbpf right
now (and they are at the same offset for both), so it won't be an
issue for this change, but it is going to make unifying all of these
things under libbpf not trivial at some point...