netdev - Re: [PATCH bpf-next] libbpf: clarify/fix unaligned data issues for btf typed dump

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] [day] [month] [year] [list]

Message-ID: <CAEf4BzZ_2arevAp_qwetCvdMk-gigvPo7tKsb7d0xF-xnezL_w@mail.gmail.com>
Date:   Fri, 16 Jul 2021 15:05:43 -0700
From:   Andrii Nakryiko <andrii.nakryiko@...il.com>
To:     Alan Maguire <alan.maguire@...cle.com>
Cc:     Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Andrii Nakryiko <andrii@...nel.org>, Martin Lau <kafai@...com>,
        Song Liu <songliubraving@...com>, Yonghong Song <yhs@...com>,
        john fastabend <john.fastabend@...il.com>,
        KP Singh <kpsingh@...nel.org>,
        Bill Wendling <morbo@...gle.com>,
        Shuah Khan <shuah@...nel.org>, bpf <bpf@...r.kernel.org>,
        Networking <netdev@...r.kernel.org>,
        "open list:KERNEL SELFTEST FRAMEWORK" 
        <linux-kselftest@...r.kernel.org>,
        open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH bpf-next] libbpf: clarify/fix unaligned data issues for
 btf typed dump

On Fri, Jul 16, 2021 at 2:44 PM Alan Maguire <alan.maguire@...cle.com> wrote:
>
> If data is packed, data structures can store it outside of usual
> boundaries.  For example a 4-byte int can be stored on a unaligned
> boundary in a case like this:
>
> struct s {
>         char f1;
>         int f2;
> } __attribute((packed));
>
> ...the int is stored at an offset of one byte.  Some platforms have
> problems dereferencing data that is not aligned with its size, and
> code exists to handle most cases of this for BTF typed data display.
> However pointer display was missed, and a simple macro to test if
> "data_is_unaligned(data, data_sz)" would help clarify this code.
>
> Suggested-by: Andrii Nakryiko <andrii@...nel.org>
> Signed-off-by: Alan Maguire <alan.maguire@...cle.com>
> ---
>  tools/lib/bpf/btf_dump.c | 13 +++++++++----
>  1 file changed, 9 insertions(+), 4 deletions(-)
>
> diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c
> index 929cf93..9dfe9c1 100644
> --- a/tools/lib/bpf/btf_dump.c
> +++ b/tools/lib/bpf/btf_dump.c
> @@ -1654,6 +1654,8 @@ static int btf_dump_base_type_check_zero(struct btf_dump *d,
>         return 0;
>  }
>
> +#define data_is_unaligned(data, data_sz)       (((uintptr_t)data) % data_sz)
> +

there is no need for macro, please use static function. And
ptr_is_aligned() is probably a better form:

if (!ptr_is_aligned(data, sz)) {
    /* handle uncommon case */
}

ptr_is_aligned() can be probably reused more readily in some other places later.

>  static int btf_dump_int_data(struct btf_dump *d,
>                              const struct btf_type *t,
>                              __u32 type_id,
> @@ -1672,7 +1674,7 @@ static int btf_dump_int_data(struct btf_dump *d,
>         /* handle packed int data - accesses of integers not aligned on
>          * int boundaries can cause problems on some platforms.
>          */
> -       if (((uintptr_t)data) % sz)
> +       if (data_is_unaligned(data, sz))
>                 return btf_dump_bitfield_data(d, t, data, 0, 0);
>
>         switch (sz) {
> @@ -1739,7 +1741,7 @@ static int btf_dump_float_data(struct btf_dump *d,
>         int sz = t->size;
>
>         /* handle unaligned data; copy to local union */
> -       if (((uintptr_t)data) % sz) {
> +       if (data_is_unaligned(data, sz)) {
>                 memcpy(&fl, data, sz);
>                 flp = &fl;
>         }
> @@ -1897,7 +1899,10 @@ static int btf_dump_ptr_data(struct btf_dump *d,
>                               __u32 id,
>                               const void *data)
>  {
> -       btf_dump_type_values(d, "%p", *(void **)data);
> +       void *ptrval;

sizeof(void *) could be 4 on the host system and 8 in BTF. If you want
to preserve the speed, I'd do something like:

if (ptr_is_aligned(data, sizeof(void *)) && sizeof(void *) == d->ptr_sz) {
    btf_dump_type_values(d, "%p", *(void **)data);
} else {
    /* fetch pointer value as unaligned integer */
    if (d->ptr_sz == 4)
        printf("0x%x")
    else
        printf("0x%llx")
}

Maybe there is some cleaner way. But that should work, no?

> +
> +       memcpy(&ptrval, data, d->ptr_sz);
> +       btf_dump_type_values(d, "%p", ptrval);
>         return 0;
>  }
>
> @@ -1910,7 +1915,7 @@ static int btf_dump_get_enum_value(struct btf_dump *d,
>         int sz = t->size;
>
>         /* handle unaligned enum value */
> -       if (((uintptr_t)data) % sz) {
> +       if (data_is_unaligned(data, sz)) {
>                 *value = (__s64)btf_dump_bitfield_get_data(d, t, data, 0, 0);
>                 return 0;
>         }
> --
> 1.8.3.1
>