samba3 valgrind error

[Jeremy Allison]

On Tue, Sep 05, 2006 at 11:44:09AM +0200, Stefan (metze) Metzmacher wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi *,
> 
> I just found this while searching for some strange things happening with
> samba3's 'make test' on my laptop.
> 
> host=127.0.0.2 share=tmp user=metze myname=SERNOX
> Running TORTURE
> 1 clients started
> ==31704== Syscall param write(buf) points to uninitialised byte(s)
> ==31704==    at 0x423B7D3: __write_nocancel (in /lib/libc-2.4.so)
> ==31704==    by 0x80E96D2: write_data (lib/util_sock.c:556)
> ==31704==    by 0x8083C65: write_socket (libsmb/clientgen.c:133)
> ==31704==    by 0x8083E00: cli_send_smb (libsmb/clientgen.c:162)
> ==31704==    by 0x809770C: cli_issue_write (libsmb/clireadwrite.c:326)
> ==31704==    by 0x809782A: cli_write (libsmb/clireadwrite.c:360)
> ==31704==    by 0x80623FB: rw_torture (torture/torture.c:406)
> ==31704==    by 0x8062601: run_torture (torture/torture.c:459)
> ==31704==    by 0x806F9EB: create_procs (torture/torture.c:4864)
> ==31704==    by 0x806FC8C: run_test (torture/torture.c:4994)
> ==31704==    by 0x8070442: main (torture/torture.c:5196)
> ==31704==  Address 0x44F835F is 63 bytes inside a block of size 131,072
> alloc'd
> ==31704==    at 0x4021259: malloc (in
> /usr/lib/valgrind/x86-linux/vgpreload_memcheck.so)
> ==31704==    by 0x80E483B: malloc_ (lib/util.c:906)
> ==31704==    by 0x8088D4E: cli_negprot (libsmb/cliconnect.c:1205)
> ==31704==    by 0x8089A25: cli_start_connection (libsmb/cliconnect.c:1473)
> ==31704==    by 0x8089B8F: cli_full_connection (libsmb/cliconnect.c:1522)
> ==31704==    by 0x8061C1D: torture_open_connection_share
> (torture/torture.c:221)
> ==31704==    by 0x8061E21: torture_open_connection (torture/torture.c:277)
> ==31704==    by 0x806F957: create_procs (torture/torture.c:4852)
> ==31704==    by 0x806FC8C: run_test (torture/torture.c:4994)
> ==31704==    by 0x8070442: main (torture/torture.c:5196)

I don't think this is a real error. Looks like it's writing
uninitialized memory out of the standard request buffer. Not
something we do (I don't think) in actual code rather than
tests. I'll take a look though.

Jeremy.