Cliff Woolley wrote: > Greg, can you try this patch on daedalus for me?
:-( We have another similar looking dump, with this patch on. It's /usr/local/apache2.0.35c/corefiles/httpd.core.2 #0 mmap_cleanup (themmap=0x8153ac0) at mmap.c:98 #1 0x280cc8f0 in apr_mmap_delete (mm=0x8153ac0) at mmap.c:202 #2 0x280ad926 in mmap_destroy (data=0x8131088) at apr_buckets_mmap.c:82 #3 0x280adf08 in apr_brigade_cleanup (data=0x814c4f8) at apr_brigade.c:86 #4 0x280adebe in brigade_cleanup (data=0x814c4f8) at apr_brigade.c:72 #5 0x280cdd3b in run_cleanups (c=0x81393f8) at apr_pools.c:1713 #6 0x280cd51c in apr_pool_destroy (pool=0x813f010) at apr_pools.c:638 #7 0x805ea1c in ap_process_http_connection (c=0x812c120) at http_core.c:300 #8 0x806ecf3 in ap_run_process_connection (c=0x812c120) at connection.c:85 #9 0x806f008 in ap_process_connection (c=0x812c120, csd=0x812c048) at connection.c:207 #10 0x80648f0 in child_main (child_num_arg=547) at prefork.c:671 Didn't I see a refcount inside the object that's being deleted? If so, how can we trust it after the thing is deleted? Greg