Author Topic: Samba On Oz-3.5.3/gpe Suddenly Stops Working. ='(  (Read 2507 times)

vorondil28

  • Newbie
  • *
  • Posts: 36
    • View Profile
    • http://
Samba On Oz-3.5.3/gpe Suddenly Stops Working. ='(
« on: March 16, 2006, 11:41:26 pm »
I got samba v3.0.10 installed the other day and after fiddling around in smb.conf, I finally got it working.  I could browse to and from my 6000L over WiFi.  File xfers were slow, but they worked.  I've spent part of today continuing to play with it and suddenly, many of the samba programs stopped working.  For instance,

Code: [Select]
root@Navi:~# smbpasswd -h
(output omitted, but list of options is correct)
root@Navi:~# smbpasswd -a user
Bus error
root@Navi:~# smbpasswd
Bus error
root@Navi:~# _

Where "user" is a valid username on my Z.

Another example,

Code: [Select]
root@Navi:~# smbclient -L \\\\COMPUTER -U=user
Password:
Domain=[COMPUTER] OS=[Windows 5.1] Server=[Windows 2000 LAN Manager]
Bus error
root@Navi:~# _

Where "user" is a valid username on the WindowsXP machine: "COMPUTER," and the password was typed correctly.

Smbd seems to be having problems too:

Code: [Select]
root@Navi:~# smbd -i
smbd version 3.0.10 started.
Copyright Andrew Tridgell and the Samba Team 1992-2004
===============================================================
INTERNAL ERROR: Signal 7 in pid 4113 (3.0.10)
Please read the appendix Bugs of the Samba HOWTO collection
===============================================================
PANIC: internal error
BACKTRACE: 2 stack frames:
 #0 smbd(smb_panic2+0x6c) [0x1649e8]
 #1 [(nil)]
Aborted
root@Navi:~# _

This happens when I start smbd in interactive mode (-i) and attempt to connect to it from a WindowsXP machine by going Start > Run... > "\\IP_ADDRESS_OF_ZAURUS".  The "INTERNAL ERROR" occurs immediately after hitting OK in the Run dialog.  (Note that "smbd -i" continues to appear in a ps listing although I still cant view shares on the Z from another machine.)  And the Windows machine then says: "\\IP_ADDRESS_OF_ZAURUS  No network provider accepted the given network path."  I've been all over Google and this forum but I have yet to find anyone who's seen this error.    

The Bugs portion of the Samba HOWTO doesn't yield anything useful other than how to submit a bug report and where to go to join a mailing list to ask about it.  (Which I  plan on doing if I can't figure this out.  No need in bothering the Samba devs if a fellow OZ user happens to have seen this before.     )  

I've tried reinstalling Samba ("ipkg remove samba;  ipkg -dest home install samba; ipkg-link add samba;"  and even  "ipkg -dest home -force-reinstall install samba")  Reboots don't change anything.  Haven't tried reflashing, which would probably work, but if I reinstalled Windows every time I had a problem...  


I'm at a loss.  I've no clue what to try.  Does a fellow OZ'er have any ideas?

Thanks for your help!  
« Last Edit: March 16, 2006, 11:47:43 pm by vorondil28 »

lardman

  • Hero Member
  • *****
  • Posts: 4512
    • View Profile
    • http://people.bath.ac.uk/enpsgp/Zaurus/
Samba On Oz-3.5.3/gpe Suddenly Stops Working. ='(
« Reply #1 on: March 17, 2006, 09:11:49 am »
Did you change something between the working and non-working invocations - install a new lib for example?

Might be worth using strace to see what the error is generated by.


Si
C750 OZ3.5.4 (GPE, 2.6.x kernel)
SL5500 OZ3.5.4 (Opie)
Nokia 770
Serial GPS, WCF-12, Socket Ethernet & BT, Ratoc USB
WinXP, Mandriva

vorondil28

  • Newbie
  • *
  • Posts: 36
    • View Profile
    • http://
Samba On Oz-3.5.3/gpe Suddenly Stops Working. ='(
« Reply #2 on: March 17, 2006, 11:44:39 am »
Quote
Did you change something between the working and non-working invocations - install a new lib for example?

Might be worth using strace to see what the error is generated by.


Si
[div align=\"right\"][a href=\"index.php?act=findpost&pid=119013\"][{POST_SNAPBACK}][/a][/div]

Not that I can think of.  I was mostly creating/editing files to get samba to behave like I wanted it.  For instance, samba kept saying "could not find blah, creating lame lowcase table, creating lame upcase table"  so I went and touched blah, and the message went away.  I played around with smb.conf, etc'.  But I've tried it w/ all those files moved or renamed and still get theses odd errors.  Now, I was superuser the entire time, so this could just be a slip of the finger that (1) I didn't catch at the time, and (2) clobbered a lib or something.      That's what really bugs me: I have no clue what I could have done.  In all likelihood, I'll slap my forehead when we figure it out.    

Apparently I don't have strace installed.  I'll do that when I get back home later today.  Google tells me it's a "system call tracing tool."  I've not used it before, but I'll see what I can do with it.

Many Thanks.  

vorondil28

  • Newbie
  • *
  • Posts: 36
    • View Profile
    • http://
Samba On Oz-3.5.3/gpe Suddenly Stops Working. ='(
« Reply #3 on: March 18, 2006, 05:37:59 pm »
Alrighty, I installed strace and after playing with it some, this is what I've got.

Code: [Select]
root@Navi:/media/card# strace -o smbd-strace smbd -i > smbd-stdout &
root@Navi:/media/card# (This is where I connect to the Z from my windows machine.)
[1] + Done                       strace -o smb-strace smbd -i 1>smbd-stdout
root@Navi:/media/card# _

Okay, so here's the last bit of the strace output ("smbd-strace") starting (I think) at about where smbd accepts the connection from the windows box.  (Note the IP address of the windows machine is 172.16.0.12.):

Code: [Select]
listen(19, 50)                          = 0
pipe([20, 21])                          = 0
fcntl64(20, F_GETFL)                    = 0 (flags O_RDONLY)
fcntl64(20, F_SETFL, O_RDONLY|O_NONBLOCK) = 0
fcntl64(21, F_GETFL)                    = 0x1 (flags O_WRONLY)
fcntl64(21, F_SETFL, O_WRONLY|O_NONBLOCK) = 0
select(1024, [18 19 20], NULL, NULL, NULL) = 1 (in [18])
time(NULL)                              = 1142633523
accept(18, {sa_family=AF_INET, sin_port=htons(3830), sin_addr=inet_addr("172.16.0.12")}, [16]) = 22
fcntl64(22, F_GETFL)                    = 0x2 (flags O_RDWR)
fcntl64(22, F_SETFL, O_RDWR)            = 0
--- SIGBUS (Bus error) @ 0 (0) ---
write(1, "================================"..., 64) = 64
write(1, "INTERNAL ERROR: Signal 7 in pid "..., 47) = 47
write(1, "Please read the appendix Bugs of"..., 60) = 60
write(1, "================================"..., 64) = 64
write(1, "PANIC: internal error\n", 22) = 22
write(1, "BACKTRACE: 2 stack frames:\n", 27) = 27
write(1, " #0 smbd(smb_panic2+0x6c) [0x164"..., 37) = 37
write(1, " #1 [(nil)]\n", 12)           = 12
rt_sigaction(SIGABRT, {SIG_DFL}, {0x153e1c, [ABRT], SA_RESTART|0x4000000}, 8) = 0
rt_sigprocmask(SIG_UNBLOCK, [ABRT], NULL, 8) = 0
getpid()                                = 13409
kill(13409, SIGABRT)                    = 0
--- SIGABRT (Aborted) @ 0 (0) ---
+++ killed by SIGABRT +++
So, "fcntl64(22, F_SETFL, O_RDWR)" is the offending call?

And the contents of "smbd-stdout" is the same as  the third example in my original post.

If that's not enough to see what's wrong, I've attached the entire output of strace (renamed w/ ".txt" so the forum would let me upload it).


(As an offtopic sidenote:) Strace is kinda cool.  :-P  Am I correct that a line like:
Code: [Select]
setsockopt(18, SOL_SOCKET, SO_KEEPALIVE, [1], 4) = 0 is a system call to the function "setsockopt", where the stuff in the parethesis are the arguments, and the stuff after the equalsign is the value setsockopt returned?

Yay, I'm learning.  

Thanks,
v.