valgrind errors

[Stefan (metze) Metzmacher]

Stefan (metze) Metzmacher schrieb:

> Hi tridge,
> 
> I got this:
> when I run two ncacn_np: RPC-SAMR tests against a samba4 server
> (without my generic server infastructure patch!)
> 
> valgrind bin/smbd -i -M standard -d1
> ==15177== Memcheck, a.k.a. Valgrind, a memory error detector for x86-linux.
> ==15177== Copyright (C) 2002-2003, and GNU GPL'd, by Julian Seward.
> ==15177== Using valgrind-2.0.0, a program supervision framework for 
> x86-linux.
> ==15177== Copyright (C) 2000-2003, and GNU GPL'd, by Julian Seward.
> ==15177== Estimated CPU clock rate is 451 MHz
> ==15177== For more details, rerun with: -v
> ==15177==
> smbd version 3.9.0-SVN-build-1475 started.
> Copyright Andrew Tridgell and the Samba Team 1992-2004
> Using standard process model
> Unknown parameter encountered: "server services"
> Ignoring unknown parameter "server services"
> No path in service vfstest - using /tmp
> No path in service vfstest1 - using /tmp
> No path in service vfstest2 - using /tmp
> bind failed on port 1024 - Address already in use
> bind failed on port 1025 - Address already in use
> rewrite: Not doing session claim
> rewrite: Not doing session claim
> ==15329== Invalid free() / delete / delete[]
> ==15329==    at 0x4002A692: free (vg_replace_malloc.c:231)
> ==15329==    by 0x80F5269: ltdb_modify_internal 
> (lib/ldb/ldb_tdb/ldb_tdb.c:473)
> ==15329==    by 0x80F55DD: ltdb_modify (lib/ldb/ldb_tdb/ldb_tdb.c:574)
> ==15329==    by 0x80F1891: ldb_modify (lib/ldb/common/ldb.c:124)
> ==15329==    Address 0x41B59EAC is 24 bytes inside a block of size 93 
> alloc'd
> ==15329==    at 0x4002A39F: malloc (vg_replace_malloc.c:153)
> ==15329==    by 0x808646B: talloc (lib/talloc.c:164)
> ==15329==    by 0x8086576: talloc_realloc (lib/talloc.c:198)
> ==15329==    by 0x8087224: talloc_ldb_alloc (lib/talloc.c:597)
> ==15307== Invalid free() / delete / delete[]
> ==15307==    at 0x4002A692: free (vg_replace_malloc.c:231)
> ==15307==    by 0x80F5269: ltdb_modify_internal 
> (lib/ldb/ldb_tdb/ldb_tdb.c:473)
> ==15307==    by 0x80F55DD: ltdb_modify (lib/ldb/ldb_tdb/ldb_tdb.c:574)
> ==15307==    by 0x80F1891: ldb_modify (lib/ldb/common/ldb.c:124)
> ==15307==    Address 0x41B1120C is 24 bytes inside a block of size 93 
> alloc'd
> ==15307==    at 0x4002A39F: malloc (vg_replace_malloc.c:153)
> ==15307==    by 0x808646B: talloc (lib/talloc.c:164)
> ==15307==    by 0x8086576: talloc_realloc (lib/talloc.c:198)
> ==15307==    by 0x8087224: talloc_ldb_alloc (lib/talloc.c:597)

It seems that this is caused by the global ldb_context and our 
ldb_set_alloc(sam_ctx->ldb, talloc_ldb_alloc, mem_ctx);

btw: process_model single segfaults too:-(

-- 
metze

Stefan Metzmacher <metze at samba.org> www.samba.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 194 bytes
Desc: OpenPGP digital signature
Url : http://lists.samba.org/archive/samba-technical/attachments/20040714/d1f5eda6/signature.bin