Could reproduce it again, now with corefile ;-) nearly the same as before:
This thread asserted, and it's image pointer doesn't look good Thread 1 (process 6033): #0 0xffffe425 in __kernel_vsyscall () #1 0xf7a11db1 in raise () from /lib/libc.so.6 #2 0xf7a13668 in abort () from /lib/libc.so.6 #3 ...
I cannot reproduce the race condition with helgrund again, but wenn running Ernst's program again and again in an endlessloop I was able to generate an assertion:
Thread 1 (process 2537): #0 0xffffe425 in __kernel_vsyscall () #1 0xf79ffdb1 in raise () from /lib/libc.so.6 #2 0xf7a01668 in abort ...
Works fine! Neither my test application nor Ernst's crashes anymore. Thank you!
But I could see multiple times following output with helgrind
==1780== Possible data race during read of size 4 at 0x6d60660 by thread #25 ==1780== at 0x5EBB2F8: GetVirtualPixelsFromNexus (cache.c:1736) ==1780== by ...
Hi, I'm running a server daemon written in c++ that is using MagickWand to rescale JPEG images or render watermarks on it. Some days ago I've updated imagemagick from version 6.3.3 to 6.3.5.10. Since this time my daemon segfaulted occasionally. After about 100 processed images the daemon crashes. It ...
hi, i got the same effect when working with DNG files. All I can get is a thumbnail size image out of my DNG files. (256px) Even identify tells me that it's only such a small picture. I'm using a Canon EOS 350D and converted the images with adobe's DNG converter version 3.3.