On Fri, 2021-10-22 at 01:46 +0200, Ilya Leoshkevich wrote:
> This is the same as commit c9e982b87946 ("libbpf: Fix dumping
> big-endian bitfields"), but for CO-RE. Make the code structure as
> similar as possible to that of btf_dump_get_bitfield_value().
>
> Signed-off-by: Ilya Leoshkevich <iii@xxxxxxxxxxxxx>
> ---
> tools/lib/bpf/relo_core.c | 11 ++++++++---
> 1 file changed, 8 insertions(+), 3 deletions(-)
>
> diff --git a/tools/lib/bpf/relo_core.c b/tools/lib/bpf/relo_core.c
> index b5b8956a1be8..fd814b985e1e 100644
> --- a/tools/lib/bpf/relo_core.c
> +++ b/tools/lib/bpf/relo_core.c
> @@ -661,13 +661,18 @@ static int bpf_core_calc_field_relo(const char
> *prog_name,
> if (validate)
> *validate = true; /* signedness is never
> ambiguous */
> break;
> - case BPF_FIELD_LSHIFT_U64:
> + case BPF_FIELD_LSHIFT_U64: {
> + __u32 bits_offset = bit_off - byte_off * 8;
> + __u8 nr_copy_bits;
> +
> #if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
> - *val = 64 - (bit_off + bit_sz - byte_off * 8);
> + nr_copy_bits = bit_sz + bits_offset;
> #else
> - *val = (8 - byte_sz) * 8 + (bit_off - byte_off * 8);
> + nr_copy_bits = byte_sz * 8 - bits_offset;
> #endif
> + *val = 64 - nr_copy_bits;
> break;
> + }
> case BPF_FIELD_RSHIFT_U64:
> *val = 64 - bit_sz;
> if (validate)
At a closer look this patch is not necessary: the new and the old
expressions yield the same result. Please disregard it.
Best regards,
Ilya
