Torsten Bronger schrieb am 31.07.2016 um 21:39: > Hallöchen! > > Hartmut Knaack writes: > >> Appearance >> ========== >> >> In my test case, ufraw crashes with the message "Segmentation >> fault (core dumped)", when I open a Raw image of a camera which is >> unsupported by lensfun, and select in the lens correction tab the >> model "PanoTools Lens Model". >> >> Commit of introduction >> ====================== >> >> 41fa7a (Patch #3532 Fix vignetting in case of non-matching crop >> factors) >> >> Details >> ======= >> >> By debugging lfModifier::lfModifier I figured that crop is set to >> zero by ufraw in case that no lensfun profile exists for the >> specific camera. This causes a division by zero when calculating >> coordinate_correction. This is a dump of the affected variables >> at the end of lfModifier::lfModifier: crop: 0.000000, >> coordinate_correction: -nan, NormScale: -nan NormUnScale: -nan, >> CenterX: -nan, CenterY: -nan MaxX: -nan, MaxY: -nan >> >> As a result, there is probably an image buffer accessed with out >> of range coordinates. > > Also a thank you for your efforts so far from me. The given patch > does not affect that way image data is being accessed. (Well, not > directly at least; I don't claim to see all of its implications > already.) However, it does change the ABI of Lensfun. And this has > made a lot of trouble in the past. Have you re-compiles also ufraw > in every step of the bisecting? > > Tschö, > Torsten. >
Hi, I can't guarantee that I built ufraw every single time I've tested a commit of lensfun. But I did when I started with the distortion issue and kept doing it countless times. However, I did the whole build process for commit 41fa7a and its predecessor f8f0eba, with the result that this problem appears with commit 41fa7a. Thanks, Hartmut ------------------------------------------------------------------------------ _______________________________________________ Lensfun-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/lensfun-users
