[mythtv] .18 and the new configure script
J. Donavan Stanley
jdonavan at jdonavan.net
Sat Apr 9 15:07:38 UTC 2005
Daniel Kristjansson wrote:
>On Sat, 2005-04-09 at 10:34 -0400, J. Donavan Stanley wrote:
>>Myself and others have had major stability issues if we allow the new
>>configure script to select our optimizations. On my P4 backend I'd get
>>segfaults when watching live SDTV live TV, on my celeron frontend I'd
>>get segfaults when watching 1080i content. In both of my cases, the
>>areas that were segfaulting were completely unrelated. It appears as
>>though some of the P4 optimizations lead to code that has buffer
>>overruns in it.
>I've been using -march=pentium4 for a year or so with various versions
>of gcc without any of these optimization related problems. Can you
>produce some backtraces?
Well that's kind of the point, the backtraces are useless. If the
backtraces actually pointed to the real location of the problems I'd
have fixed them and checked in the code.
More information about the mythtv-dev