Re: [RFC PATCH 2/6] perf bpf: Add pt_regs convert table for x86

[Alexei Starovoitov <ast@plumgrid.com>]

On 5/5/15 3:10 AM, He Kuang wrote:
> Convert register number in debuginfo to its index in pt_regs.
>
> Signed-off-by: He Kuang <hekuang@huawei.com>
> ---
>   tools/perf/arch/x86/util/dwarf-regs.c | 31 +++++++++++++++++++++++++++++++
>   1 file changed, 31 insertions(+)
>
> diff --git a/tools/perf/arch/x86/util/dwarf-regs.c b/tools/perf/arch/x86/util/dwarf-regs.c
> index be22dd4..e586a47 100644
> --- a/tools/perf/arch/x86/util/dwarf-regs.c
> +++ b/tools/perf/arch/x86/util/dwarf-regs.c
> @@ -59,10 +59,31 @@ const char *x86_64_regs_table[X86_64_MAX_REGS] = {
>   	"%r15",
>   };
>
> +const char x86_64_pt_regs_table[X86_64_MAX_REGS] = {
> +	10,
> +	12,
> +	11,
> +	5,
> +	13,
> +	14,
> +	4,
> +	19,
> +	9,
> +	8,
> +	7,
> +	6,
> +	3,
> +	2,
> +	1,
> +	0,
> +};

magic numbers need a comment in .c (not only in commit log)
Also I would combine them into
struct { const char *regname, int index } and
merge with x86_64_regs_table array
and would use something like offsetof(struct pt_regs, di) / 8
instead of hard coding.

In general patch 2 and 3 are in wrong order.
Generic 3 should come first. Otherwise patch 2 introduces things
that no one is using.