On 05/18/2011 01:01 AM, Pablo Neira Ayuso wrote: > thanks, I'll keep investigating this tomorrow. This tastes like a memory > corruption, but not sure that this occurs in the _output() function. > > I'll try to get a fix for this by tomorrow. >
The easiest guess is that we either need the va_start/va_end inside the loop of _output, or need to copy the arguments again. to reproduce we should probably use the same arch and GCC/(E)GLIBC version that harald is using. holger
