samba - Jun 2000 - smbstatus fails with "Segmentation fault"

Arnold Troeger wrote:
> I'm running Samba 2.0.7 on a heavily patched Solaris 2.6 machine.  For
> about 3 minutes after I restart smb and nmb I can run smbstatus (as
> root) with no trouble.  After that though smbstatus starts segfaulting.
> It shows the shares properly, but the locks and memory usage messages do
> not appear.  Have I found a bug in smbstatus?
>
> Thanks in advance for any suggestions.
>
> Regards,
> Arnold
>
> --
>
Yes, I did find a bug.  The problem is Samba 2.0.x and the 64 bit aspect of
Solaris 2.6 (Sparc, not Intel).  Specifically, it's the configure script
that
compiles the program.  The configure script sets CPPFLAGS to
-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64.  This evidently screws up the
offset calculations for identifying locks in the locking_shm.c routines.  I
have no idea why you would want to compile large file support into Samba as
neither Windows, nor Linux, can currently handle large files.  Samba compiles
with warning messages about shifting counts being too big (ntrans.c, reply.c,
trans2.c and util.c) when the large file support flags are removed.  What's
more, smbstatus now works.

There should probably be an option in the configure script for removing the
large file support, at least with Solaris 2.6.  Has anyone tried compiling
Samba 2.0.7 in Solaris 8 for Sparc architecture?  Does the 64 bit support
work properly?

--
Arnold Troeger                          Unocal Thailand
------------------------------------------------------------------------
"Microsoft Windows:  for when your machine is just too fast"

Arnold Troeger wrote:
| Yes, I did find a bug.  The problem is Samba 2.0.x and the 64 bit
aspect
| of Solaris 2.6 (Sparc, not Intel).  Specifically, it's the configure 
| script that compiles the program.  The configure script sets
CPPFLAGS to
| -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64.  This evidently screws
up 
| the offset calculations for identifying locks in the locking_shm.c
| routines.  

	In that case, I'd look at the offset calculation that's
	being screwed up!

| I have no idea why you would want to compile large file support into
Samba
| as neither Windows, nor Linux, can currently handle large files. 

	Large files are standard Unix, and are being phased in
	by all vendors as we speak: it's a really bad idea to
	try to compile them out, as you would be choosing to depend
	on something that the vendors promise to remove!

| Samba compiles with warning messages about shifting counts being too
big
| (ntrans.c, reply.c, trans2.c and util.c) when the large file support
flags
| are removed.  What's more, smbstatus now works.

	Those warnings are the tip-off: we need to hunt down the
	erroneous shifts and repair them.

| Has anyone tried compiling Samba 2.0.7 in Solaris 8 for Sparc 
| architecture?  Does the 64 bit support work properly?

	I can reproduce the segfault on a 64-bit Solaris 8, but
	never saw it on my (former) 2.6 or 2.7 machines, or
	Solaris 8 compiled in 32-bit mode.

	Please note:
		Solaris 2.6 is a 32-bit os with 64-bit filesizes.
		Solaris 7 can be run as either a 32 or 64-bit os,
			and has 64-bit filesizes in either case.
		Solaris 8 ditto.
	Sun compilers default to building 32-bit binaries for
	any of those Solarii: these include 64-bit filesizes
	defined as long longs.


	Am I correct in thinking you're compiling with gcc in 32-bit
	mode, or are you building a 64-bit samba application?

	[Followups redirected to samba-technical@samba.org]

--dave
-- 
David Collier-Brown,  | Always do right. This will gratify some people
185 Ellerslie Ave.,   | and astonish the rest.        -- Mark Twain
Willowdale, Ontario   | //www.oreilly.com/catalog/samba/author.html
Work: (905) 415-2849 Home: (416) 223-8968 Email: davecb@canada.sun.com