am-utils and NIS

Anyone have any experience with either of these under Linux/SGI? In trying to rebuild the am-utils package to support NIS maps, I keep getting this situation -- happens each time I try it :-(

[...]
gcc -shared -Wl,-soname -Wl,libamu.so.1 -o .libs/libamu.so.1.0.1 misc_rpc.lo mount_fs.lo mtab.lo nfs_prot_xdr.lo util.lo wire.lo xdr_func.lo xutil.lo transputil.lo mtabutil.lo mountutil.lo umount_fs.lo
collect2: ld terminated with signal 6 [Aborted]
make[2]: *** [libamu.la] Error 1
make[2]: Leaving directory `/usr/src/redhat/BUILD/am-utils-6.0a16/libamu'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/usr/src/redhat/BUILD/am-utils-6.0a16'
make: *** [all-recursive-am] Error 2

I'm building this from the SRPM, and I actually haven't done any modifications to it yet -- it's just the am-utils-6.0a16-1.src.rpm from ftp.linux.sgi.com. (built with 'rpm -bc am-utils.spec'). Oddly enough, it doesn't appear that NIS support was detected in the am-utils configure script (although the machine was bound to a domain at the time). I won't spam the entire build log to the list, but I have it if anyone has any leads on this and could benefit from it.

I don't know how related this is, but while ypmatch-type calls work OK, ypcat generally returns errors like this:

green:~$ ypcat group 
eegrad:*:4000:test
eeugrad:*:6000:
faculty:*:2000:
staff:*:10:azam,harper,marchany,sysacct,clepple
eestaff:*:3000:azam,harper,marchany,sysacct
yp_all: clnt_call: RPC: Program not registered
[above repeats 3 to 5 times, then terminates with...]
No such map group.byname. Reason: RPC failure on NIS operation

Sorry to bug you all about this -- I know that NIS probably isn't high on everyone's priority list...

Thanks in advance,

Charles Lepple
System Administrator, Virginia Tech EE Workstation Labs
[email protected] || http://www.foo.tho.org/charles/

[I'm cc'ing this to the list in case anyone else has seen this behaviour in strace]

Ralf Baechle wrote:
[stuff about ypcat failing]

One of these days I'll manage to compile something without incident...

I grabbed the strace CVS module, and tried to build it. Configure identified a 'mips-unknown-linux', and things went fine until time.c:

gcc -Wall -DHAVE_CONFIG_H -I. -Ilinux/mips -I./linux/mips -Ilinux
-I./linux -g -O2 -c time.c /usr/include/asm/timex.h: In function `get_cycles':
In file included from /usr/include/linux/timex.h:138,

                 from time.c:36:

/usr/include/asm/timex.h:36: warning: implicit declaration of function `read_32bit_cp0_register'
/usr/include/asm/timex.h:36: `CP0_COUNT' undeclared (first use this function)
/usr/include/asm/timex.h:36: (Each undeclared identifier is reported only once
/usr/include/asm/timex.h:36: for each function it appears in.)
make: *** [time.o] Error 1
<<<<

So I looked up the offending function, and added '#include <asm/mipsregs.h>' before the other Linux-specific includes. This produced an executable, but when I ran 'strace -o /tmp/ypcat -f ypcat group', the strace output ended like this:

6341  gettimeofday({925689555, 977502}, NULL) = 0
6341  getpid()                          = 6341
6341  socket(PF_INET, SOCK_DGRAM, IPPROTO_UDP) = 4
6341  getpid()                          = 6341
6341  bind(4, {sin_family=AF_INET, sin_port=htons(1005),

sin_addr=inet_addr("0.0
.0.0")}, 16) = 0
6341 ioctl(4, 0x667e, 0x7ffffb18) = 0
<<<<

No exit indication, and no output from the ypcat program, either. The error code from strace was zero.

The kernel source and strace were both pulled from CVS about an hour ago.

Any ideas?

Charles Lepple
System Administrator, Virginia Tech EE Workstation Labs
[email protected] || http://www.foo.tho.org/charles/

I don't know if you've solved this problem, but I bet you're using binutils 2.9.1, install the old 2.8.1 and everything will (probably) work just fine.

I'm trying to get back to the cutting edge of my mailbox today :-)

- Ulf