samba - Feb 2002 - VM_GROWSDOWN problem on S/390

We're running Samba 2.2.2 on an S/390 mainframe and have been experiencing
problems with Samba seeming to access an invalid memory location. Below is
part of the message log:

Feb 15 10:48:38 linux1 kernel: VM_GROWSDOWN not set, but address 0        
     
Feb 15 10:48:38 linux1 kernel: not in vma 0106a960 (start 400000 end
53A000)    
Feb 15 10:48:38 linux1 kernel: User process fault: interruption code 0x10 
     
Feb 15 10:48:38 linux1 kernel: failing address: 0                         
     
Feb 15 10:48:38 linux1 kernel: CPU:    0                                  
     
Feb 15 10:48:38 linux1 kernel: Process smbd (pid: 3725,
stackpage=00339000)     
Feb 15 10:48:38 linux1 kernel:                                            
     
Feb 15 10:48:38 linux1 kernel: User PSW:    0709c000 8043a2d8             
     
Feb 15 10:48:38 linux1 kernel: task: 00338000 tss: 003382d8 ksp: 00339cf8
pt_reg
s: 00339f68                                                               
     
Feb 15 10:48:38 linux1 kernel: User GPRS:                                 
     
Feb 15 10:48:38 linux1 kernel: 00000000  00000000  00000000  00000000     
     
Feb 15 10:48:38 linux1 kernel: 005cd940  005cd940  0000ffff  00000028     
     
Feb 15 10:48:38 linux1 kernel: 00000028  7ffffa68  0000004c  005cc908     
     
Feb 15 10:48:38 linux1 kernel: c013cc30  8043a02c  8043a1c0  7ffff4e0     
     
Feb 15 10:48:38 linux1 kernel: User ACRS:                                 
     
Feb 15 10:48:38 linux1 kernel: 00000000  00000000  00000000  00000000     
     
Feb 15 10:48:38 linux1 kernel: 00000001  00000000  00000000  00000000     
     
Feb 15 10:48:38 linux1 kernel: 00000000  00000000  00000000  00000000     
     
Feb 15 10:48:38 linux1 kernel: 00000000  00000000  00000000  00000000     
     
Feb 15 10:48:38 linux1 kernel: Kernel BackChain  CallChain    BackChain 
CallCha
in                                                                        
     
Feb 15 10:48:38 linux1 kernel:        00339cf8   0000004c                 
     

After a few of these messages, the system eventually hangs - No network
connection and no ability to enter commands directly from the management
console.  Previous versions worked just fine for us.

BTW, the 2.2.2 version of SWAT seems to behave differently. I had v2.2.1a
setup so that user accessed the SWAT pages with only their Samba
userid/passwd, and the could manage them that way.  Now, it seems to only
grant access when you put in your Linux id/passwd, and it doesn't change
the Samba password at all.

Thanks in advance for your help.

John Nealy
Programmer/Analyst
Information Services
School District of Hillsborough County
Tampa, FL.

John Nealy wrote:
> 
> We're running Samba 2.2.2 on an S/390 mainframe and have been
experiencing
> problems with Samba seeming to access an invalid memory location. Below is
> part of the message log:
> 
> Feb 15 10:48:38 linux1 kernel: VM_GROWSDOWN not set, but address 0
> 
> Feb 15 10:48:38 linux1 kernel: not in vma 0106a960 (start 400000 end
> 53A000)
> Feb 15 10:48:38 linux1 kernel: User process fault: interruption code 0x10
> 
> Feb 15 10:48:38 linux1 kernel: failing address: 0
> 
> Feb 15 10:48:38 linux1 kernel: CPU:    0
> 
> Feb 15 10:48:38 linux1 kernel: Process smbd (pid: 3725,
> stackpage=00339000)
> Feb 15 10:48:38 linux1 kernel:
> 
> Feb 15 10:48:38 linux1 kernel: User PSW:    0709c000 8043a2d8
> 
> Feb 15 10:48:38 linux1 kernel: task: 00338000 tss: 003382d8 ksp: 00339cf8
> pt_reg
> s: 00339f68
> 
> Feb 15 10:48:38 linux1 kernel: User GPRS:
> 
> Feb 15 10:48:38 linux1 kernel: 00000000  00000000  00000000  00000000
> 
> Feb 15 10:48:38 linux1 kernel: 005cd940  005cd940  0000ffff  00000028
> 
> Feb 15 10:48:38 linux1 kernel: 00000028  7ffffa68  0000004c  005cc908
> 
> Feb 15 10:48:38 linux1 kernel: c013cc30  8043a02c  8043a1c0  7ffff4e0
> 
> Feb 15 10:48:38 linux1 kernel: User ACRS:
> 
> Feb 15 10:48:38 linux1 kernel: 00000000  00000000  00000000  00000000
> 
> Feb 15 10:48:38 linux1 kernel: 00000001  00000000  00000000  00000000
> 
> Feb 15 10:48:38 linux1 kernel: 00000000  00000000  00000000  00000000
> 
> Feb 15 10:48:38 linux1 kernel: 00000000  00000000  00000000  00000000
> 
> Feb 15 10:48:38 linux1 kernel: Kernel BackChain  CallChain    BackChain
> CallCha
> in
> 
> Feb 15 10:48:38 linux1 kernel:        00339cf8   0000004c
> 
> 
> After a few of these messages, the system eventually hangs - No network
> connection and no ability to enter commands directly from the management
> console.  Previous versions worked just fine for us.
Well, the first thing to note is that if Samba brings down the whole
server, then the server is broken.   But you probably have a bug in
Samba that triggers what looks to be a simple SegV, just s390 style.
> BTW, the 2.2.2 version of SWAT seems to behave differently. I had v2.2.1a
> setup so that user accessed the SWAT pages with only their Samba
> userid/passwd, and the could manage them that way.  Now, it seems to only
> grant access when you put in your Linux id/passwd, and it doesn't
change
> the Samba password at all.
SWAT always uses the Linux password to grant access, but my allow
'password changes' for the samba password, given the old and new samba
password.

Andrew Bartlett

-- 
Andrew Bartlett                                 abartlet@pcug.org.au
Manager, Authentication Subsystems, Samba Team  abartlet@samba.org
Student Network Administrator, Hawker College   abartlet@hawkerc.net
http://samba.org     http://build.samba.org     http://hawkerc.net