https://bugs.openldap.org/show_bug.cgi?id=10333

Issue ID: 10333 Summary: Recurring crash in lmdb:mdb_page_alloc() Product: LMDB Version: 0.9.29 Hardware: All OS: All Status: UNCONFIRMED Keywords: needs_review Severity: normal Priority: --- Component: liblmdb Assignee: bugs@openldap.org Reporter: captgol2000@yahoo.com Target Milestone: ---

Full_Name: Version: 0.9.29 OS: Vendor built Linux based on Ubuntu kernel 4.14.173

mdb_page_alloc (SIGBUS) Number of Events: 610

We have seen a recurring crash in lmdb:mdb_page_alloc() but only in the field with our product. One instance yielded a core in which we confirmed the same stack back trace when examined. It matches the automated backtrace reports we received from field deployments. The lmdb is built from source at version 0.9.29, unchanged from upstream. The operating environment is an embedded Linux appliance with Linux kernel at version "linux-4.14.173-aum-01" and our own application that uses lmdb to store certain data while transactions are in progress. We build everything with gcc 7.3.0 and link with glibc version 2.27. Unfortunately no in-house reproduction has been possible, only a number of matching stack traces from field reports of application crashes. We are working on some system tests to see if can can cause a reproduction. This report is in hopes some others may have seen a similar backtrace to help locate a bug.

(gdb) where #0 mdb_page_alloc (num=num@entry=1, mp=mp@entry=0x7ffec216d120, mc=<optimized out>) at mdb.c:2310 #1 0x00007fa2508c764e in mdb_page_new (mc=mc@entry=0x7ffec216d670, flags=flags@entry=1, num=num@entry=1, mp=mp@entry=0x7ffec216d1d0) at mdb.c:7193 #2 0x00007fa2508c7a9b in mdb_page_split (mc=mc@entry=0x7ffec216d670, newkey=newkey@entry=0x7ffec216da70, newdata=newdata@entry=0x7ffec216da50, newpgno=newpgno@entry=18446744073709551615, nflags=0) at mdb.c:8649 #3 0x00007fa2508ca4ee in mdb_cursor_put (mc=0x7ffec216d670, key=0x7ffec216da70, data=0x7ffec216da50, flags=<optimized out>) at mdb.c:6957 #4 0x00007fa2508cc26a in mdb_put (txn=0xb767020, dbi=3, key=0x7ffec216da70, data=0x7ffec216da50, flags=0) at mdb.c:9045 #5 0x00007fa2520a69f4 in XXX_XXX_store::add (this=0x7fa25230ada0 <XXX_x509_validation_s::kv_store_h>, key=0x7ffec216dbc0 "\351\066\030\201\261\223\025\q\006\204g:\274\241\330\377Ã ", keylen=20, value=0x7ffec216db20 "\002", valuelen=19, flag=<optimized out>, table=XXX_KV_STORE_TABLE_MAIN) at lib/XXX_kv_store.cpp:479 #6 0x00007fa2520a03fa in ?? () #7 0x00007ffec216dc60 in ?? () #8 0x00007ffec216dca8 in ?? () #9 0x00007ffec216dc60 in ?? () #10 0x00007fa2520a2990 in ?? () Backtrace stopped: previous frame inner to this frame (corrupt stack?)