INN 2.4.0 dumps core on Solaris 9

[Russ Allbery]

Nathan Coraor <nate at cse.psu.edu> writes:
> "Jeffrey M. Vinocur" said:

>> Can you (in the same source tree you've been using) do `make clean`,
>> `make warnings`, and then `make update` (which will install the new
>> binaries but not touch your config files).  That'll get us more useful
>> backtraces.

>   Okay, it attempts to write to an overview file, and:

>   14194/1:        llseek(39, 0, SEEK_CUR)                         = 454
>   14194/1:        pwrite64(38, "\0\0\0\0\0\0\0\0\0\001C6".., 40, 5120) = 40
>   14194/1:        memcntl(0xFEDB2000, 8192, MC_SYNC, MS_ASYNC, 0, 0) = 0
>   14194/1:        fcntl(10, F_SETLKW64, 0xFFBFF180)               = 0
>   14194/1:            Incurred fault #6, FLTBOUNDS  %pc = 0xFEFB3464
>   14194/1:              siginfo: SIGSEGV SEGV_MAPERR addr=0x00000000
>   14194/1:            Received signal #11, SIGSEGV [default]
>   14194/1:              siginfo: SIGSEGV SEGV_MAPERR addr=0x00000000
>   14213:  read(0, 0x0025A25C, 8192)                       = 0
>   14213:  llseek(4, 0, SEEK_CUR)                          Err#29 ESPIPE
>   14213:  close(4)                                        = 0

>   (gdb) bt
>   #0  0xfefb3464 in strlen () from /usr/lib/libc.so.1
>   #1  0x0008af6c in HashMessageID (MessageID=0x0) at hash.c:68
>   #2  0x00086acc in hisv6_write (history=0x294bb8, key=0x0, arrived=1066960140,
>       posted=1066841312, expires=0, token=0xffbff490) at hisv6/hisv6.c:852
>   #3  0x000851f8 in HISwrite (h=0x2915f0, key=0x0, arrived=1066960140,
>       posted=1066841312, expires=0, token=0xffbff490) at his.c:265
>   #4  0x0006b888 in InndHisWrite (key=0x0, arrived=1066960140,
>       posted=1066841312, expires=0, token=0xffbff490) at util.c:356
>   #5  0x00051f0c in ARTpost (cp=0x2a5f20) at art.c:2331

Someone's trying to feed you a message that doesn't have a message ID.
Sorry, I should have thought of that; I thought you were segfaulting
before the server came all the way up.  This is a bug in INN 2.4.0 that's
fixed in current STABLE snapshots and will be fixed in INN 2.4.1; the
easiest thing to do is probably to switch to a STABLE snapshot from
<ftp://ftp.isc.org/isc/inn/snapshots/>.  A STABLE snapshot will correctly
reject such articles as malformed.

I really need to get INN 2.4.1 out....

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>

    Please send questions to the list rather than mailing me directly.
     <http://www.eyrie.org/~eagle/faqs/questions.html> explains why.