[linuxtv-softmpeg] Re: problems with libsoftmpeg

[Michael Hunold <hunold@convergence.de>]

Hello Henrik,

On 05/17/04 09:32, Henrik Johansson wrote:
> The problem with the sound described in my previous message was caused 
> by me applying the  fusionsound_realtimepriority_reset_on_flush.diff 
> patch to the CVS version of FusionSound. I got a suceeded with fuzz 
> during processing of ifusionsoundstream.c but instead of calling the 
> fs_core_reset function from the flush function, the call turned up in 
> the IFusionSoundStream_GetPresentationDelay function. Don't asked me 
> how, the context in the patch didn't match a single line of this function.
Ah, thank you for you debugging. I updated the patches in the 
libsoftmpeg CVS repository to match the latest developments within 
FusionSound.

> When debugging libsoftmpeg I found a potential bug in the decode thread 
> in video.c where a uint64 is used for the variable "diff". Shouldn't a 
> signed data type by used instead?
The logic is somewhat screwed up. If "diff" becomes negative (ie. 
current->pts < av.stc) then this is caught a few lines below with an 
explicit check, "diff" isn't used in that case. So all is fine. 8-)

> Is anyone working on adding support for 2.6 kernels in the DirectFB 
> version of  viafb?  It would be really great to have a version that has 
> all that is needed by the DirectFB driver. If no one else is working on 
> this I'm volunteering to do it myself along with the mode generation patch.
Sorry, I cannot comment on this. If this is still urgent, please ask 
again on the DirectFB mailing list please.

> /Henrik
CU
Michael.


--
Info:  To unsubscribe send a mail to ecartis@linuxtv.org with 
"unsubscribe linuxtv-softmpeg" as subject.