smbsh cores on Solaris 8

David Collier-Brown davecb at canada.sun.com
Tue Feb 13 17:31:00 GMT 2001


David Highley wrote:
> > > I have attached the 2.0.7 version of smbsh build on a Solaris 8 system.

	He had a problem where smbsh running ksh caused a core
	dump on his Solarios 8 system.  This does not change if
	he recompiles on Solaris 8, nor if he uses the 2.2.0 code
	base.

	Using his smbsh.so, I can get a coredump when I exit:
	Truss says:
----
25463:  llseek(63, 0, SEEK_END)                         = 25768
25463:  write(63, " e x i t\n\0", 6)                    = 6
25463:  fdsync(63, O_RDONLY|O_SYNC)                     = 0
25463:  getcontext(0xFFBEEBF0)
25463:      Incurred fault #6, FLTBOUNDS  %pc = 0x00031A90
25463:        siginfo: SIGSEGV SEGV_MAPERR addr=0x00000008
25463:      Received signal #11, SIGSEGV [default]
25463:        siginfo: SIGSEGV SEGV_MAPERR addr=0x00000008
25463:          *** process killed ***
----
	Apptrace says:
----
  25478:smbsh    -> libc.so.1:strcpy(dst = "t", src = "t") = "t"
  25478:smbsh    -> libc.so.1:strcmp(s1 = "exit", s2 = "elif") = 0xc
  25478:smbsh    -> libc.so.1:strcmp(s1 = "exit", s2 = "else") = 0xc
  25478:smbsh    -> libc.so.1:strcmp(s1 = "exit", s2 = "esac") = 0x5
  25478:smbsh    -> libc.so.1:wctype(charclass = "blank") = 0x40
  25478:smbsh    -> libc.so.1:iswctype(wc = 0xa, charclass = 0x40) =
0x0 errno = 0 (Error 0)
  25478:smbsh    -> libc.so.1:lseek64(fildes = 0x3f, offset = 0x0,
whence = 0x0) = 0x2
  25478:smbsh    -> libc.so.1:iswspace(wc = 0x65) = 0x0 errno = 0
(Error 0)
  25478:smbsh    -> libc.so.1:iswspace(wc = 0x78) = 0x0 errno = 0
(Error 0)
  25478:smbsh    -> libc.so.1:iswspace(wc = 0x69) = 0x0 errno = 0
(Error 0)
  25478:smbsh    -> libc.so.1:iswspace(wc = 0x74) = 0x0 errno = 0
(Error 0)
  25478:smbsh    -> libc.so.1:iswspace(wc = 0xa) = 0x8
  25478:smbsh    -> (0x3f, 0x49268, 0x6)
  25478:smbsh    -> libc.so.1:fsync(fildes = 0x3f) = 0x0
apptrace: smbsh: Segmentation Fault
----

	dbx (and pmap) said:
(/opcom/forte/SUNWspro/bin/../WS6/bin/sparcv9/dbx) cont
Reading en_US.ISO8859-1.so.2
signal SEGV (no mapping at the fault address) in sh_exec at 0x31a90
0x00031a90: sh_exec+0x0634:     ld      [%o1 + 0x8], %o0
(/opcom/forte/SUNWspro/bin/../WS6/bin/sparcv9/dbx) where
=>[1] sh_exec(0x0, 0x0, 0x20, 0x0, 0x4954c, 0x41ec8), at 0x31a90
  [2] 0x295a4(0x4954c, 0x407e0, 0x407e0, 0x41f50, 0x41f54, 0x41f50),
at 0x295a3
  [3] main(0x41ee8, 0xffbef344, 0x41d94, 0x41fdc, 0x41d80, 0x41f54),
at 0x28dd8

	As we just returned from smbwrapper.so,  I'm suspicious
	that one of the mappings (like _pwrite) mentioned in the last
	message may have contributed to the problem... 

	I'll send him my smbwrappers.so and see if it core dumps
	for him, and if so, try remapping them to syscalls.

--dave
-- 
David Collier-Brown,           | Always do right. This will gratify 
Performance & Engineering Team | some people and astonish the rest.
Americas Customer Engineering  |                      -- Mark Twain
(905) 415-2849                 | davecb at canada.sun.com




More information about the samba-technical mailing list