[Samba] Samba 3.0.8 breaks all printing from Windows 98

[Jonathan Knight]

We have small panic here as we've just put samba 3.0.8 up to fix the DoS 
attack but all printing from Windows 98 has now broken.  Printing from 
Windows XP is still working fine.

We're running on Fedora 2 using LPRng.  Everything was working fine under 
3.0.7 and no changes to the OS or the config files was made.  Samba was 
upgraded to 3.0.8 using the configure line from the 3.0.7 build.

The log.smbd file contains:

[2004/11/10 15:08:11, 1] smbd/service.c:make_connection_snum(648)
  rm4pc68-klab1 (160.5.110.156) signed connect to service lilab1 initially as 
user XXXX (uid=XXXX, gid=901) (pid 8669)
[2004/11/10 15:08:11, 0] printing/printing_db.c:get_print_db_byname(40)
  PANIC: assert failed at printing/printing_db.c(40)
[2004/11/10 15:08:11, 0] lib/util.c:smb_panic2(1403)
  PANIC: assert failed
[2004/11/10 15:08:11, 0] lib/util.c:smb_panic2(1411)
  BACKTRACE: 16 stack frames:
   #0 /usr/local/samba/sbin/smbd(smb_panic2+0x1f5) [0x82076be]
   #1 /usr/local/samba/sbin/smbd(smb_panic+0x19) [0x82074c7]
   #2 /usr/local/samba/sbin/smbd(get_print_db_byname+0x9f) [0x823ac1f]
   #3 /usr/local/samba/sbin/smbd [0x8226292]
   #4 /usr/local/samba/sbin/smbd(print_job_set_name+0x27) [0x8228f64]
   #5 /usr/local/samba/sbin/smbd [0x8095554]
   #6 /usr/local/samba/sbin/smbd(api_reply+0x391) [0x80992bb]
   #7 /usr/local/samba/sbin/smbd [0x808d8f6]
   #8 /usr/local/samba/sbin/smbd(reply_trans+0xde2) [0x808e8e2]
   #9 /usr/local/samba/sbin/smbd [0x80e372d]
   #10 /usr/local/samba/sbin/smbd [0x80e3803]
   #11 /usr/local/samba/sbin/smbd(process_smb+0x215) [0x80e3b9a]
   #12 /usr/local/samba/sbin/smbd(smbd_process+0x195) [0x80e49b9]
   #13 /usr/local/samba/sbin/smbd(main+0x881) [0x827395b]
   #14 /lib/tls/libc.so.6(__libc_start_main+0xe4) [0x42015704]
   #15 /usr/local/samba/sbin/smbd [0x80750d1]
   
   A quick check reveals that printing.c has undergone a fairly major change 
to the API with many functions changing from using a sharename to being sent 
a service number which then has to be converted into a name for 
get_print_db_byname.  The assert that fails is a check to make sure that the 
printer name is not NULL.

Obviously we're dead in the water here so any help would be appreciated.   


  ______    jonathan at cs.keele.ac.uk    Jonathan Knight,
    /                                  Department of Computer Science
   / _   __ Telephone: +44 1782 583437 University of Keele, Keele,
(_/ (_) / / Fax      : +44 1782 713082 Staffordshire.  ST5 5BG.  U.K.