This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: Any rescue other than recompiling the kernel when the kernel function is inlined?
- From: Arkady <arkady dot miasnikov at gmail dot com>
- To: Shiyao Ma <i at introo dot me>
- Cc: systemtap at sourceware dot org
- Date: Wed, 5 Apr 2017 18:55:02 +0300
- Subject: Re: Any rescue other than recompiling the kernel when the kernel function is inlined?
- Authentication-results: sourceware.org; auth=none
- References: <CAJQX3Dwc9RTpw37b-RYTjWYV_9J7g6hQg1hwaV=+vwQQOHhGzw@mail.gmail.com>
On Wed, Apr 5, 2017 at 6:47 PM, Shiyao Ma <i@introo.me> wrote:
> Hi,
>
> Since the to-be monitored kernel is on production machines, so it's
> impossible to recompile the kernel with more debug info and install on
> the machines.
>
This is not because of "inline" - DWARF contains debug information for
all occurrences of the inline calls and theoretically STAP could place
breakpoints in all such places. The problem is that due to a
compilation flag there is no debug information for the specific inline
function in the specific kernel build (if my understanding is
correct).
> Any other way to monitor on the ".return" of an inlined kernel function?
>
My best advice is to try to look for alternative places in the code to
place a breakpoint.
>
>
> Best,
>
> shiyao.