[PATCH v4] Backtrace via call detail

José Fonseca jose.r.fonseca at gmail.com
Mon May 13 09:14:36 PDT 2013


On Mon, May 13, 2013 at 1:28 PM, Alexander Monakov <amonakov at ispras.ru>wrote:

>
>
> On Mon, 13 May 2013, José Fonseca wrote:
>
> > BTW, another issue is that the "offset" values on linux are not really
> the
> > offset from start of shared object, but the absolute address. Anyway, I
> > think that using glibc backtrace for this is unsustainable long term, as
> it
> > requires executables to be built with -rdynamic, and shared objects not
> to
> > be built with -fvisibility=hidden, the exact opposite of common practice
> > these days.  I don't know if there is a better alternative for runtime
> > symbol resolution, otherwise I think we'll need to do offline symbol
> > resolution on linux by default.
>
> I'm working on replacing glibc backtrace with libbacktrace to address those
> issues.  I presume you're OK with importing it to APITrace?
>
> http://gcc.gnu.org/git/?p=gcc.git;a=tree;f=libbacktrace


Yes, it's a permissive license so no problem AFAICT.


>
> > One last request -- could you make it so that when one selects a call on
> > qapitrace, the default lower pane is the "Details View" (instead of the
> > "Backtrace").
>
> diff --git a/gui/mainwindow.cpp b/gui/mainwindow.cpp
> index e46fce6..2891863 100644
>

Thanks for updates.

The branch is merged now.

Jose
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/apitrace/attachments/20130513/5ec0b137/attachment.html>


More information about the apitrace mailing list