Re: [mpls-linux-general] ldp aborts consistently

[James R. L. <jl...@mi...>]

I found the problem.

I'm not handling the case where multiple LDP PDU come in the same TCP buffer.
I will work on this.

Jim

> Hi,
> The latest CVS ldp_zebra seems to dump core quite cosistently after an
> assertion failure once the 2 neighbours start a session.
> Attached is a log of the (end of the) console output, and a backtrace.
> It seems to depend on the timing of events - if I reduce the trace level
> on the failing neighbour the other one would abort instead (althugh in a
> little earlier - after a keep-alive message).
> -- 
> Eran Mann                 Direct  : 972-4-9936230
> Senior Software Engineer  Fax     : 972-4-9890430
> Optical Access            Email   : em...@op...
> ENTER: ldp_event
> TCP_DATA
> ->>>>>>>>>>>>>>>>>>>>   SOCK: ldp_socket_tcp_read       fd 8 
> TCP_DATA 62
> Before header decode: size = 62
> OUT: Found type 300
> OUT: 03 00 00 0e 00 00 00 04 01 01 00 06 00 01 0a 0b 
> OUT: 0c b5 00 01 00 1e 0a 0b 0c b5 00 00 04 00 00 14 
> OUT: 00 00 00 05 01 00 00 04 02 00 01 00 02 00 00 04 
> OUT: 00 00 01 b2 
> OUT: decodedSize for Adr msg = 18
> OUT: LPD Header : protocolVersion = 1
> OUT:    pduLength = 24
> OUT:    lsrAddress = a0b0cb5
> OUT:    labelSpace = 0
> OUT: ADDRESS MSG ***START***:
> OUT:    baseMsg : uBit = 0
> OUT:              msgType = 300
> OUT:              msgLength = 14
> OUT:              msgId = 4
> OUT:     adrListTlv:
> OUT:     Tlv:
> OUT:     BaseTlv: uBit = 0
> OUT:              fBit = 0
> OUT:              type = 101
> OUT:              length = 6
> OUT:     adrListTlv data: addrFamily = 1
> OUT:     adrListTlv addresses (with 1 addresses) :
> OUT: 
>                         a0b0cb5  
> OUT: ADDRESS MSG ***END***:
> OUT: Mesg size: 18 (52)
> ENTER: ldp_state_machine
> OUT: FSM: state 5, event 4
> ENTER: ldp_state_process
> ENTER: ldp_addr_process
> OUT: Addr Recv: session(1)
> EXIT: ldp_addr_process
> EXIT: ldp_state_process
> EXIT: ldp_state_machine
> OUT: Found type 1
> OUT: 00 01 00 1e 0a 0b 0c b5 00 00 04 00 00 14 00 00 
> OUT: 00 05 01 00 00 04 02 00 01 00 02 00 00 04 00 00 
> OUT: 01 b2 
> OUT: decodedSize for Notif msg = -63
> ENTER: ldp_state_machine
> OUT: FSM: state 5, event 6
> Entered notif state
> ENTER: ldp_state_notif
> ENTER: ldp_notif_process
> ldp_zebra: ldp_notif.c:160: ldp_notif_process: Assertion `0' failed.
> Aborted (core dumped)
> 

> (gdb) bt
> #0  0x400624e1 in __kill () from /lib/libc.so.6
> #1  0x40062156 in raise (sig=6) at ../sysdeps/posix/raise.c:27
> #2  0x40063868 in abort () at ../sysdeps/generic/abort.c:88
> #3  0x4005c3ce in __assert_fail (assertion=0x8083be3 "0", 
>     file=0x8083bd7 "ldp_notif.c", line=160, 
>     function=0x8083bc5 "ldp_notif_process") at assert.c:59
> #4  0x8064a7a in ldp_notif_process (g=0x80ace70, s=0x80bba70, a=0x0, e=0x0, 
>     r_attr=0x80bfe88) at ldp_notif.c:160
> #5  0x805f1c0 in ldp_state_notif (g=0x80ace70, s=0x80bba70, adj=0x0, 
>     entity=0x0, event=6, msg=0xbfffbc64, from=0xbffffc54)
>     at ldp_state_funcs.c:437
> #6  0x805e40e in ldp_state_machine (g=0x80ace70, session=0x80bba70, adj=0x0, 
>     entity=0x0, event=6, msg=0xbfffbc64, from=0xbffffc54)
>     at ldp_state_machine.c:358
> #7  0x805e230 in ldp_event (g=0x80ace70, socket=0x80ad8c8, user_from=0x0, 
>     if_handle=0x0, buf=0x8091840, extra=0x80bba70, event=LDP_EVENT_NOTIF)
>     at ldp_state_machine.c:288
> #8  0x806999c in zldp_socket_read_handler (t=0xbffffce4)
>     at zldp_socket_impl.c:73
> #9  0x80759a6 in thread_call (thread=0xbffffce4) at thread.c:728
> #10 0x804a266 in main (argc=1, argv=0xbffffd54, envp=0xbffffd5c)
>     at ldp_zebra.c:613
> (gdb) 


-- 
James R. Leu