Hi Dale! It seems like you are right - Well spotted!
I'll update at once. Where are you experiencing the crashes - do you by any chance have a backtrace? Regards, Klaus Post http://www.klauspost.com On Thu, Sep 24, 2009 at 6:58 PM, Dale Olds <[email protected]> wrote: > A few weeks ago, I had some trouble with segfaults using rawstudio from > trunk. When I disabled the asm i386 optimizations, things got much better, > but I eventually reverted to an older packaged build from OpenSUSE. Things > are looking really good though, so I am looking forward to the next release. > > While researching the segfault, I noticed something I believe would cause an > intermittent use of some of the asm optimizations. The modules contains the > following code: > > FFTWindow::FFTWindow( int _w, int _h ) : > analysis(FloatImagePlane(_w, _h)), > synthesis(FloatImagePlane(_w,_h)) > { > analysisIsFlat = true; > synthesisIsFlat = true; > analysis.allocateImage(); > synthesis.allocateImage(); > #if defined (__i386__) || defined (__x86_64__) > SSEAvailable == !!(rs_detect_cpu_features() & RS_CPU_FLAG_SSE); > #endif > } > > The problem is this line: > SSEAvailable == !!(rs_detect_cpu_features() & RS_CPU_FLAG_SSE); > > As I read it, this means that SSEAvailable is not actually assigned any value > here, and is left uninitialized (at least by this code). Shouldn't this line > be as follows? > SSEAvailable = !!(rs_detect_cpu_features() & RS_CPU_FLAG_SSE); > > --Dale > > > > _______________________________________________ > Rawstudio-dev mailing list > [email protected] > http://rawstudio.org/cgi-bin/mailman/listinfo/rawstudio-dev > _______________________________________________ Rawstudio-dev mailing list [email protected] http://rawstudio.org/cgi-bin/mailman/listinfo/rawstudio-dev
