On Aug 3, 2015, at 4:02 PM, [email protected] wrote: > We are using jemalloc in our internal project and we have been experiencing > very strange memory corruptions lately, so we decided to build jemalloc with > --enable-debug and --enable-fill and give it a try. > > Also, we used this env variable: > MALLOC_CONF=tcache:false,redzone:true,stats_print:true > > jemalloc version is 3.6.0 > > This is what we are seeing in logs: > > <jemalloc>: include/jemalloc/internal/arena.h:508: Failed assertion: "pageind > >= map_bias" > > Can you tell us anything about this assertion? > Is there a way this could happen? Is it a possible bug in jemalloc or is it > more likely that this is some random corruption in internal jemalloc > structures?
I can't guess much about the nature of the memory corruption a backtrace through the jemalloc functions. If you're still experiencing this issue, please post a backtrace and I'll take a look. Thanks, Jason _______________________________________________ jemalloc-discuss mailing list [email protected] http://www.canonware.com/mailman/listinfo/jemalloc-discuss
