[Serusers] Problems with locking in timer routine - sr-users

28 May 2003

Hi folks,

I am having problems with locking in timer routine (ser 0.8.10), 
particularly with the my recent addition to the nathelper - udp pinger. 
This piece of code is being invoked periodically by the timer, retrieves 
list of all currently registered contacts and sends short udp message to 
each of them. Obviously, routine which retrieves all contacts locks each 
domain before accessing it, but apparently it doesn't work as expected.

Following is the dump of debugging session:

-bash-2.05b$ sudo gdb ~/PortaSIP/ser/work/ser-0.8.10/ser ser.core
GNU gdb 4.18 (FreeBSD)
Copyright 1998 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain 
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "i386-unknown-freebsd"...Deprecated bfd_read 
called at 
/usr/src/gnu/usr.bin/binutils/gdb/../../../../contrib/gdb/gdb/dbxread.c 
line 2627 in elfstab_build_psymtabs
Deprecated bfd_read called at 
/usr/src/gnu/usr.bin/binutils/gdb/../../../../contrib/gdb/gdb/dbxread.c 
line 933 in fill_symbuf

Core was generated by `ser'.
Program terminated with signal 11, Segmentation fault.
Reading symbols from /usr/lib/libc.so.4...done.
Reading symbols from /usr/local/lib/ser/modules/sl.so...done.
Reading symbols from /usr/local/lib/ser/modules/tm.so...done.
Reading symbols from /usr/local/lib/ser/modules/rr.so...done.
Reading symbols from /usr/local/lib/ser/modules/maxfwd.so...done.
Reading symbols from /usr/local/lib/ser/modules/usrloc.so...done.
Reading symbols from /usr/local/lib/ser/modules/registrar.so...done.
Reading symbols from /usr/local/lib/ser/modules/nathelper.so...done.
Reading symbols from /usr/local/lib/ser/modules/textops.so...done.
Reading symbols from /usr/local/lib/ser/modules/radius_auth.so...done.
Reading symbols from /usr/local/lib/libradiusclient.so.0...done.
Reading symbols from /usr/lib/libmd.so.2...done.
Reading symbols from /usr/lib/libcrypt.so.2...done.
Reading symbols from /usr/libexec/ld-elf.so.1...done.
#0  0x2a1a62df in get_all_ucontacts (buf=0x80d5248, len=1402) at dlist.c:110
110                                     if (c->c.len <= 0)
(gdb) bt
#0  0x2a1a62df in get_all_ucontacts (buf=0x80d5248, len=1402) at dlist.c:110
#1  0x2a1c4b3e in _init () from /usr/local/lib/ser/modules/nathelper.so
#2  0x8073679 in timer_ticker () at timer.c:118
#3  0x805e912 in main_loop () at main.c:654
#4  0x80611a1 in main (argc=1, argv=0xbfbffdb8) at main.c:1383
#5  0x804c5a6 in _start ()
(gdb) print c
$1 = (ucontact_t *) 0x460a0d30
(gdb) print *c
Cannot access memory at address 0x460a0d30.
(gdb) print *r
$2 = {domain = 0x282ec0d8, aor = {s = 0x282f4638 "011801", len = 6}, 
contacts = 0x282f5cb8, slot = 0x282ed418, d_ll = {
     prev = 0x282ef0d8, next = 0x0}, s_ll = {prev = 0x0, next = 0x0}}
(gdb) print *r->contacts
$3 = {domain = 0x7a3d6863, aor = 0x34476839, c = {s = 0x37364b62 
<Address 0x37364b62 out of bounds>, len = 959328819},
   expires = 825243494, q = 2.12359957e+20, callid = {s = 0x32663238 
<Address 0x32663238 out of bounds>, len = 1714774885},
   cseq = 1631019574, state = 1631020084, user_agent = {s = 0x62386438 
<Address 0x62386438 out of bounds>, len = 775107636},
   next = 0x460a0d30, prev = 0x3a6d6f72}
(gdb) print *r->contacts->next
Cannot access memory at address 0x460a0d30.
(gdb) l 100
95              void *cp;
96              int shortage;
97
98              cp = buf;
99              shortage = 0;
100             /* Reserve space for terminating 0000 */
101             len -= sizeof(c->c.len);
102             for (p = root; p != NULL; p = p->next) {
103                     lock_udomain(p->d);
104                     if (p->d->d_ll.n <= 0) {
(gdb) l
105                             unlock_udomain(p->d);
106                             continue;
107                     }
108                     for (r = p->d->d_ll.first; r != NULL; r = 
r->d_ll.next) {
109                             for (c = r->contacts; c != NULL; c = 
c->next) {
110                                     if (c->c.len <= 0)
111                                             continue;
112                                     if (len >= 
(int)(sizeof(c->c.len) + c->c.len)) {
113                                             memcpy(cp, &c->c.len, 
sizeof(c->c.len));
114                                             cp += sizeof(c->c.len);

As you can see, we are locked domain in question (line 103), but still, 
found one of records to be in the inconsistent state (contacts aren't 
initialized).

Does anyone have any ideas about what could be wrong with this? I an 
seeing similar problem in the code that periodically expires contacts.

-Maxim