Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-20 Thread Benjamin Wolsey
The first thing I notice is that the container we're iterating on is somewhat non-trivial: typedef boost::multi_index_container Property, boost::multi_index::indexed_by boost::multi_index::ordered_unique boost::multi_index::composite_key Property,

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-17 Thread strk
On Fri, Oct 16, 2009 at 07:40:21PM -0400, Bernie Innocenti wrote: Secondly, the Property::setReachable() method is out of line and it also seems non-trivial as it accesses a boost::variant (the mBound member): A possible way to speed this up is using the apply_visitor function of boost:

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-16 Thread Rob Savoye
On 10/14/09 14:40, Bernie Innocenti wrote: In oprof_start, enable call graphs and set the path to your elf vmlinux image (/usr/lib/debug/lib/modules/2.6.31.1-56.fc12.x86_64/vmlinux in my case). Ok after a few frustrating days, I've gotten oprofile to work. Bizarrely, it crashed another

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-16 Thread Bernie Innocenti
El Fri, 16-10-2009 a las 10:39 -0600, Rob Savoye escribió: Ok after a few frustrating days, I've gotten oprofile to work. Bizarrely, it crashed another system I was working on. Here's a log I made from a YouTube video with callgraphs and symbol info. Now anyone motivated should be able

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-16 Thread Rob Savoye
On 10/16/09 14:33, Bernie Innocenti wrote: motivated should be able to get started on tracking down performance issues. http://www.gnashdev.org/testcases/oprof_gnash_all.log Very interesting. This looks by far the hottest spot: gnash::PropertyList::setReachable() That adds it

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-16 Thread Bastiaan Jacques
On Fri, 16 Oct 2009, Bernie Innocenti wrote: El Fri, 16-10-2009 a las 10:39 -0600, Rob Savoye escribió: Ok after a few frustrating days, I've gotten oprofile to work. Bizarrely, it crashed another system I was working on. Here's a log I made from a YouTube video with callgraphs and symbol

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-16 Thread Bernie Innocenti
El Fri, 16-10-2009 a las 15:29 -0600, Rob Savoye escribió: gnash::PropertyList::setReachable() That adds it to the garbage collector. http://gnashdev.org/doc/html/classgnash_1_1PropertyList.html#d7bae96d62a404a9656db68cc7aa408e. I just updated the doxygen pages on Gnash from

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-14 Thread Rob Savoye
On 10/13/09 23:39, Kevin Kofler wrote: On Wednesday 14 October 2009, Rob Savoye wrote: Is the Gstreamer packages standard now, We build Gnash against GStreamer in Fedora, for the usual patent reasons. (GStreamer's modular architecture allows adding the patent-encumbered codecs from

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-14 Thread Rob Savoye
On 10/12/09 18:30, Michael Stone wrote: Lastly, if there's a simple solution out there or reason why our approach was doomed from the start, then I'd also appreciate it if you could link to it a little bit more prominently from your wiki since Google doesn't seem to know anything about it.

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-14 Thread Bernie Innocenti
El Tue, 13-10-2009 a las 22:57 -0600, Rob Savoye escribió: Called from where ? We couldn't find out, because symbols availability is a pre-requisite for dumping the call graphs. I don't have oprofile setup on my other Fedora 11 machine. It's amazingly simple these days: just install the

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-14 Thread Rob Savoye
On 10/14/09 14:40, Bernie Innocenti wrote: It's amazingly simple these days: just install the oprofile-gui rpm and then run sudo oprof_start. Of course you also need to install the kernel-debuginfo and gnash-debuginfo. You probably also want glib-debuginfo and gtk-debuginfo. Can I do this my

[Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-13 Thread Michael Stone
Dear gnash folks, I just wanted to mention to you that Bernie Innocenti and I spent a few minutes together last week trying to learn something about why gnash stutters on assorted content on our Rawhide and Debian Sid machines. We approached the problem via oprofile and learned two things:

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-13 Thread John Gilmore
One easy thing to do is to pull down the gnash source code and just run ./configure and make. That should get you one with symbols, which you can do performance analysis on. You can leave Fedora's or Debian's symbols problems for later. I just wanted to mention to you that Bernie Innocenti and

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-13 Thread Rob Savoye
On 10/12/09 18:30, Michael Stone wrote: I just wanted to mention to you that Bernie Innocenti and I spent a few minutes together last week trying to learn something about why gnash stutters on assorted content on our Rawhide and Debian Sid machines. The stuttering (if you mean for video) is

Re: [Gnash-dev] oprofile difficulties, easy access to debuginfo, and other thoughts

2009-10-13 Thread Kevin Kofler
On Wednesday 14 October 2009, Rob Savoye wrote: Is the Gstreamer packages standard now, We build Gnash against GStreamer in Fedora, for the usual patent reasons. (GStreamer's modular architecture allows adding the patent-encumbered codecs from third-party repositories. FFmpeg's monolithic