tracing on demand

[Eugene Velesevich]

On Fri, Apr 12, 2013 at 2:17 AM, José Fonseca <jose.r.fonseca at gmail.com> wrote:
>
> This should be fixed now. See https://github.com/apitrace/apitrace/issues/117
>
> Jose

Hello,

The bug still exists.
Let a trace contain two swapBuffers in succession. The Enter() part of
second swapBuffers is before first swapBuffers, the Leave() part is
after it:
Enter SwapBuffers 2
Enter SwapBuffers 1
Leave SwapBuffers 1
Leave SwapBuffers 2

So, scanTrace marks Leave part of second swapBuffers as a frame with
one call. But when the user opens the frame, fetchFrameContent obtains
a call after second swapBuffers and reads next frame with greater
numOfChildren. It results in a memory corruption and a crash.

I have a trace with this bug, but... you need the backtrace patch v3
to read this trace :)

Frame 322
-------------- next part --------------
A non-text attachment was scrubbed...
Name: app_process.4.trace
Type: application/octet-stream
Size: 3141252 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/apitrace/attachments/20130423/0e46b485/attachment-0001.obj>