FPS should be ignored if there is only a single frame

[Pablo Rodríguez]

For some reason unknown to me, lightspark is not able to display the attached file, which I took from Wikipedia (http://en.wikipedia.org/wiki/File:Seal_of_the_United_States_Supreme_Court.svg), converted to PDF format with inkscape and converted to SWF with pdf2swf (http://www.swftools.org).

The weird thing in this is that according what I can understand from the log, the file seems to be rendered fine, but it is not drawn.

Thanks for your help,

Pablo

[Pablo Rodríguez (ousia)]

I have just realized that rendering is aborted.

Just in case it helps, I attach the most verbose log info from lightspark opening the above mentioned file.

[Alessandro Pignotti (a-pignotti)]

Actually on my box the file renders, although not completely. Try and wait for ten seconds or so, the framerate declared by this SWF seems to be very low (e.g. 0.25 FPS) so the first frame gets rendered after 4 seconds.

[Pablo Rodríguez (ousia)]

lightspark aborts the file rendering after less than a second, so I cannot wait ten seconds.

[Alessandro Pignotti (a-pignotti)]

Please, post a gdb backtrace of the abort

[Pablo Rodríguez (ousia)]

Here you have it.

I hope it helps.

(gdb) run Seal_of_the_USSCt.swf
Starting program: /usr/bin/lightspark Seal_of_the_USSCt.swf
[Thread debugging using libthread_db enabled]
[New Thread 0xb7fd4b70 (LWP 3275)]
[New Thread 0xb77d3b70 (LWP 3278)]
[New Thread 0xb6fd2b70 (LWP 3279)]
[New Thread 0xb67d1b70 (LWP 3280)]
[New Thread 0xb5fd0b70 (LWP 3281)]
[New Thread 0xb57cfb70 (LWP 3282)]
[New Thread 0xb4d92b70 (LWP 3283)]
INFO: RenderThread this=0xbffff7f0
[New Thread 0xb4591b70 (LWP 3284)]
INFO: Creating input thread
[New Thread 0xb3d90b70 (LWP 3285)]
INFO: Compressed SWF file: Version 6 Length 150154
INFO: FrameRate 0.64
INFO: End of parsing @ 150154
INFO: Fragment shader compilation
INFO: Vertex shader compilation
ERROR: Incomplete FBO status 36054... Aborting

Program received signal SIGABRT, Aborted.
[Switching to Thread 0xb4591b70 (LWP 3284)]
0x0012d422 in __kernel_vsyscall ()
(gdb) bt
#0 0x0012d422 in __kernel_vsyscall ()
#1 0x00ded651 in raise () from /lib/tls/i686/cmov/libc.so.6
#2 0x00df0a82 in abort () from /lib/tls/i686/cmov/libc.so.6
#3 0x0815e943 in lightspark::RenderThread::commonGLInit(int, int, unsigned int*) ()
#4 0x0815f0dd in lightspark::RenderThread::sdl_worker(lightspark::RenderThread*) ()
#5 0x0013396e in start_thread () from /lib/tls/i686/cmov/libpthread.so.0
#6 0x00e90a0e in clone () from /lib/tls/i686/cmov/libc.so.6

[Alessandro Pignotti (a-pignotti)]

What graphics card/video driver do you have?

[Pablo Rodríguez (ousia)]

The same I reported in another bug: ATI Radeon Mobility X1400 using the radeon drivers (that come with Ubuntu 10.04).

My glewinfo output is available at http://launchpadlibrarian.net/48824515/glewinfo.txt.

[Alessandro Pignotti (a-pignotti)]

I think this is the same FBO issue reported here in bug 582825, probably ATI related. I'll investigate.

[Pablo Rodríguez (ousia)]

After the main bug seems to have been fixed, I get nothing or a garbage screen (it is also random).

BTW, when it shows garbage it takes 4 seconds to do it. According to swfdump the file only contains a single frame, so it makes no sense to wait 4 seconds to render it. FPS should be only taken into account if there is more than one frame.

At least, both Adobe and gnash doesn't wait 4 seconds to render the file.

[Alessandro Pignotti (a-pignotti)]

Please add a screen-shot of the garbage you're getting, and a log for both the 'nothing rendered' and 'garbage' cases.

[Pablo Rodríguez (ousia)]

After this morning upgrade, lightspark renders the file fine. It is far from being perfect, but I guess it won't be displayed differently on your computer.

When I click on the lightspark window I get the following message, which I don't know whether it is ok:

    lightspark: radeon_texture.c:86: radeonFreeTexImageData: Assertion `!image->base.Data' failed.
    Aborted

BTW, I'm changing the bug title, since it reflects the issue with Flash documents that only contain a single frame and FPS shouldn't be honored in those cases.

[Alessandro Pignotti (a-pignotti)]

When clicking on the windows a texture readback is done to identify the clicked object. It seems that there is an issue with radeon drivers then, please open a different bug for this.

[Alessandro Pignotti (a-pignotti)]

I've found an out of bound array access when doing texture readback. Should be fixed in next release (0.4.0 rc3)