lkml.org 
[lkml]   [2015]   [May]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [RFC PATCH 0/6] perf bpf: Probing with local variable
    On 2015/05/05 19:10, He Kuang wrote:
    > This patch set is based on https://lkml.org/lkml/2015/4/30/264
    >
    > By using bpf 'config' section like this:
    >
    > char _config2[] SEC("config") = "generic_perform_write=generic_perform_write+122 file->f_mapping->a_ops bytes offset";
    > SEC("generic_perform_write")
    > int NODE_generic_perform_write (struct pt_regs *ctx, void *a_ops, void *bytes, void* offset) {
    > char fmt[] = "NODE_generic_perform_write, a_ops=%p, bytes=%p, offset=%p\n";
    > bpf_trace_printk(fmt, sizeof(fmt), a_ops, bytes, offset);
    > return 1;
    > }
    >
    > In this example, 'bytes' and 'offset' are local variables, a_ops is in
    > the structure field of file parameter, and we probe in the body of the
    > generic_perform_write() function.
    >
    > Perf can fetch and convert all the arguments and then we translate them
    > into bpf bytecode as a prologue before calling bpf body functions. In
    > the prologue, we fetch arguments from bpf context register and place
    > them according to bpf calling conventions so the body function can
    > access them as formal parameters.
    >
    > The perf command is as following:
    >
    > $ perf bpf -v bpf_bytecode.o
    > ...
    > bpf_prologue: insn num=26
    > (bf) r6 = r1
    > (79) r3 = *(u64 *)(r6 +112)
    > (07) r3 += 248
    > (b7) r1 = 0
    > (7b) *(u64 *)(r10 -8) = r1
    > (bf) r1 = r10
    > (07) r1 += -8
    > (b7) r2 = 8
    > (85) call 4
    > (79) r3 = *(u64 *)(r10 -8)
    > (07) r3 += 104
    > (b7) r1 = 0
    > (7b) *(u64 *)(r10 -8) = r1
    > (bf) r1 = r10
    > (07) r1 += -8
    > (b7) r2 = 8
    > (85) call 4
    > (79) r3 = *(u64 *)(r10 -8)
    > (bf) r7 = r3
    > (79) r3 = *(u64 *)(r6 +24)
    > (bf) r8 = r3
    > (79) r3 = *(u64 *)(r6 +88)
    > (bf) r9 = r3
    > (bf) r2 = r7
    > (bf) r3 = r8
    > (bf) r4 = r9

    Hmm, this idea looks good to me, I need to learn what the bpf needs for review.
    BTW, if it is so easily to compose the byte code in perf, can we do it in the
    kernel too? I'd like to check the possibility of replacing the old fetch_arg
    functions with these byte code for optimizing performance and reduce redundant
    functionality. :)

    Thank you,



    --
    Masami HIRAMATSU
    Linux Technology Research Center, System Productivity Research Dept.
    Center for Technology Innovation - Systems Engineering
    Hitachi, Ltd., Research & Development Group
    E-mail: masami.hiramatsu.pt@hitachi.com


    \
     
     \ /
      Last update: 2015-05-06 03:41    [W:4.738 / U:1.612 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site