HEAD CVS does not create libnss_winbind.so anymore...how do I do this (nsswitch/REAQDME doesnt help)

[Mike Papper]

Hello, I compiled the latest HEAD code from CVS, using the --enable-developer 
AND the -enable-debug for configure. 

Note: normally a "make" will create the file 
source/nsswitch/libnss_winbind.so and I normally copy this file to /lib and 
link it to /lib/libnss_winbind.so.2

However, in the version from CVS make did not create this library. I read the 
README file in the nsswitch directory and tried a "make nsswitch" which did 
output dome stuff but did not seem to create anything and there was no 
source/nsswitch/libnss_winbind.so either. It appears that I do need the 
libnss_winbind.so.2 file. So, I do not know how to build a 
libnss_winbind.so.2 (whats with the .2 anyway, the docs indicate it looks 
just for .so).

Heres the output when I do "sudo make nsswitch" (from the source directory):
----------------------
[mike at 207-214-146-180-dhcp source]$ make nsswitch
Using FLAGS =  -O  -g -g -Wall -Wshadow -Wstrict-prototypes -Wpointer-arith 
-Wcast-qual -Wcast-align -DDEBUG_PASSWORD -DDEVELOPER -I/usr/kerberos/include 
 -Iinclude -I./include -I./ubiqx -I./smbwrapper  -I. -D_LARGEFILE64_SOURCE 
-D_FILE_OFFSET_BITS=64 -D_GNU_SOURCE  -I/usr/kerberos/include -I.
      FLAGS32 =  -O  -g -g -Wall -Wshadow -Wstrict-prototypes -Wpointer-arith 
-Wcast-qual -Wcast-align -DDEBUG_PASSWORD -DDEVELOPER -I/usr/kerberos/include 
 -Iinclude -I./include -I./ubiqx -I./smbwrapper  -I. -D_LARGEFILE64_SOURCE 
-D_FILE_OFFSET_BITS=64 -D_GNU_SOURCE  -I/usr/kerberos/include -I.
      LIBS = -lresolv  -ldl -lcrypt -lnsl -lcom_err -lk5crypto -lkrb5 
-lgssapi_krb5 -llber -lldap -lpam -lpopt
      LDSHFLAGS = -shared  -L/usr/kerberos/lib -O  -g -g -Wall -Wshadow 
-Wstrict-prototypes -Wpointer-arith -Wcast-qual -Wcast-align -DDEBUG_PASSWORD 
-DDEVELOPER -I/usr/kerberos/include
      LDFLAGS = -L/usr/kerberos/lib
--------------------
It doesnt seem to actually do anything. I did run configure with 
"--with-winbind" and "--with-pam" as well as the "--enable-debug"
Was I supposed to do "configure.developer --enable-debug --with-winbind 
--with-pam" or something else? Is there docs for this?


Mike Papper

On Tuesday 02 April 2002 01:02 pm, Andrew Bartlett wrote:
> Mike Papper wrote:
> > It figures, I was running it in a debugger and sent that output to
> > samba at samba.org with my bug report. Since then my /home filesystem was
> > corrupted (so I cant find my sent-messages). However, I will try again.
> >
> > -But, can you explain what you mean by "set panic action=...". Do you
> > mean run a shell command named "set" or do you mean go into the code and
> > set it. I found the following code in source/param/loadparm.c:
> >
> > {"panic action", P_STRING, P_GLOBAL, &Globals.szPanicAction, NULL, NULL,
> > 0},
> >
> > would I then set this line to:
> >
> > {"panic action", P_STRING, P_GLOBAL, &Globals.szPanicAction, "/bin/sleep
> > 9000", NULL, 0}
> >
> > and recompile? Or I could edit source/lib/util.c in the smb_panic
> > function.
>
> Nothing nearly as complex.  Its an smb.conf paramater, just set it like
> any other.
>
> > -Related question: When I use the debugger (which I rarely do) Iuse ddd.
> > It seems to be similar to gdb or whatever (excpet it has a gui). What
> > does the winbindd -i option do?
>
> Stops winbindd from forking - gdb in particular can't follow the child
> easily.
>
> > -For the panic option that brings up gdb, from Andrew Esh:
> >
> > gnome-terminal --command 'gdb /proc/%d/exe %d'"
> >
> > What do I use for the %d %d parts?
>
> Thats smb.conf magic.  %d is subsituted for the current PID inside samba
> - this allows gdb to pop up in a window with itself in a debugger.
> Quite neat.  (I use a similar thing, but I add '|| /bin/sleep 9000' to
> catch processes that dont have access to my X display.)
>
> > -BTW the code I am running is from CVS head using the -r ...samba17
> > version.
>
> Then you are using stale code.  We make snaphots and tag them, but we
> don't update them.  Try again with current HEAD, as there is some
> suspicion that this bug has been fixed.
>
> Andrew Bartlett