Legacy ImageMagick Discussions Archive

OK, that's cool. I'm actually fine with the 128KiB cutoff, and I'm a little afraid to mess with malloc in a JVM process, since we run so much on the same server. If it ain't broke. I am now sure that constant major GC runs are preventing the problem. My process RSS "only" grew 200MiB in the past 2 ...

^^^ adjustments == adjustments (in "C" code in the jmagick implementation)

It does make sense to just let malloc do the anonymous memory maps. I actually hesitate to make any adjustments, since I'm sure the JVM implementation uses malloc internally, and I have no idea what could go wrong. A couple suggestions to make this clearer: * Explain the malloc behavior in the ...

Amazing. I never knew that, and didn't even think to look into malloc. I certainly could do that, and 128 kB isn't much bigger than I was using in my example. It makes me wonder what is growing my malloc heap, though. We're only processing about 3,000-6,000 photos per day these days, and only the ...

We have a long-running daemon process that uses jmagick-6.4.0, which is a Java Native Interface (JNI) into the ImageMagick API. Therefore, we are using the ImageMagick-6.9.4-8 shared library libMagickCore-6.Q16.so. We encountered a memory leak in the "C" malloc heap because: (A) We had a code path ...