lkml.org 
[lkml]   [2008]   [Dec]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [PATCH 4/5] ftrace: function graph return for function entry
    2008/12/3 Steven Rostedt <rostedt@goodmis.org>:
    >
    > On Wed, 3 Dec 2008, Ingo Molnar wrote:
    >
    >>
    >> * Steven Rostedt <rostedt@goodmis.org> wrote:
    >>
    >> > From: Steven Rostedt <srostedt@redhat.com>
    >> >
    >> > Impact: feature, let entry function decide to trace or not
    >> >
    >> > This patch lets the graph tracer entry function decide if the tracing
    >> > should be done at the end as well. This requires all function graph
    >> > entry functions return 1 if it should trace, or 0 if the return should
    >> > not be traced.
    >>
    >> > diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
    >> > index 47aa5f0..cef05a2 100644
    >> > --- a/arch/x86/kernel/entry_64.S
    >> > +++ b/arch/x86/kernel/entry_64.S
    >> > @@ -119,6 +119,9 @@ ENTRY(mcount)
    >> > #ifdef CONFIG_FUNCTION_GRAPH_TRACER
    >> > cmpq $ftrace_stub, ftrace_graph_return
    >> > jnz ftrace_graph_caller
    >> > +
    >> > + cmpq $ftrace_graph_entry_stub, ftrace_graph_entry
    >> > + jnz ftrace_graph_caller
    >>
    >> hm, that's in the hotpath. What's the point of this? Do we want some sort
    >> of configuration vector that allows per function graphing versus
    >> entry-only traces?
    >
    > Actually, I'm fine with taking it out. But when we register a function
    > graph, we register both a entry and exit. I would think it should still
    > work if one was still set as the default. Unless we document it someplace
    > that either the entry or exit must be set.


    I like it that way too. One should choose wether setting a callback on
    entry or return or both.
    But it makes sense for me that should work with only one...

    ...unless. Actually setting only a callback on entry shoud better been
    done from the ftrace-function-tracer.

    > This change has the function graph work if one or the other is set to
    > something other than the default.
    >
    > Also, this is only in the static ftrace path, not dynamic, but then again,
    > maybe that means it is even hotter.
    >
    > I'm fine with pulling this out. It was not really needed, I was just
    > trying to make it intuitive.
    >
    > -- Steve
    >


    \
     
     \ /
      Last update: 2008-12-03 14:21    [W:0.047 / U:91.264 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site