Conference smurf::buildhelp

1236.0. "fwd: status re: v32support (ase patch) build problem" by AOSG::FILTER (Automatic Posting Software - mail to flume::puck) Thu Feb 02 1995 17:08

Date Of Receipt: 	 1-FEB-1995 01:09:50.09
From: 	SMURF::FLAMBE::jmf "Joshua M. Friedman OSF/UNIX SDE  01-Feb-1995 0108"
To: 	buildhelp@DEC:.zko.flambe, larson@DEC:.zko.flambe
CC: 	
Subj: 	fwd: status re: v32support (ase patch) build problem

Mike & Buildhelp - never mind - the problem was system related...

------- Forwarded Message

Return-Path: jmf 
Delivery-Date: Wed, 01 Feb 95 01:07:18 -0500
Return-Path: jmf
Received: from localhost by flambe.zk3.dec.com; (5.65/1.1.8.2/30Mar94-0502PM)
	id AA05308; Wed, 1 Feb 1995 01:07:12 -0500
Message-Id: <[email protected]>
To: meg, metsch
Cc: sestokas, zap, tresvik, jmf, reops
Subject: status re: v32support (ase patch) build problem
Date: Wed, 01 Feb 95 01:07:12 -0500
From: "Joshua M. Friedman, OSF/UNIX SDE 381-1548" <jmf>
X-Mts: smtp

Jim & Meg,

The issue with the v32supportos build dying has been determined to be
system troubles, alas. It clearly had nothing to do with Mike's submit.

Darrell told me the problem was reminiscent of a build problem once
seen caused by an advfs corruption over nfs.  This v32 build was on an
advfs fileset (locally mounted).  Indeed, we discovered obvious
corruption in some build files, including some build logs and the file
pointed to in the kernel build error log.

The build system, useg02, is running V2.0.  We had been planning an
upgrade to the system soon after the V3.2 SSB ship.  It has been
functioning normally, building other pools on advfs for many months.

Darrell Dunnuck, John Flanagan and I were investigating the problem,
examining corrupted files and system logs, during which time the system
panicked.  A crash dump was produced.  (During the panic, there was
also an advfs dump going on over an rsh at this same time, in the same
domain, copying v30bsupportos.bld.)

To attempt to repair the filesystem and prevent the problem, Russell
Laliberte destroyed and recreated the advfs domain which was exhibiting
the problem, and I setup the build to run again this evening (it is
running now and appears normal).

Hopefully we'll have a good build tonight...	-josh

- ------- Forwarded Message

Return-Path: jmf 
Delivery-Date: Tue, 31 Jan 95 16:46:25 -0500
Return-Path: jmf
Received: from localhost by flambe.zk3.dec.com; (5.65/1.1.8.2/30Mar94-0502PM)
	id AA18509; Tue, 31 Jan 1995 16:45:50 -0500
Message-Id: <[email protected]>
To: buildhelp
Cc: zap, meg, tresvik, jmf
Subject: Kernel build help needed w/ v32support (ase patch) problem
Date: Tue, 31 Jan 95 16:45:50 -0500
From: "Joshua M. Friedman, OSF/UNIX SDE 381-1548" <jmf>
X-Mts: smtp

Build Helpers... 

If anyone can offer any help here, that would be great.  The
v32supportos pool is a mirror of goldminos.bl5, and it has built
successfully (with nothing submitted) on 1/25.  Yesterday we took in
one submit, from Mike Larson, of
    ./kernel/io/cam/spo/simport_htm.c.

Last night the build failed (error attached).  Mike has indicated that
the failure is unrelated to his submit.  The only other change is that
several advfs fixes when into goldminos.bl5 (and built ok in goldminos),
since the last successful v32supportos build:
    ./kernel/msfs/bs/bs_access.c
    ./kernel/msfs/bs/bs_buffer2.c
    ./kernel/msfs/fs/fs_read_write.c
    ./kernel/msfs/osf/msfs_vfsops.c
    ./kernel/msfs/osf/msfs_vnops.c

There were no other changes that I know of to the build environment.

The error logs are in v32supportos/Logs; if you need to examine the
build area (v32supportos.bld), it's on host useg02 and I can provide
access.

thank you very much...			-josh


- - ------- Forwarded Messages

Return-Path: jmf 
Delivery-Date: Tue, 31 Jan 95 13:50:34 -0500
Return-Path: jmf
Received: from localhost by flambe.zk3.dec.com; (5.65/1.1.8.2/30Mar94-0502PM)
	id AA28069; Tue, 31 Jan 1995 13:50:25 -0500
Message-Id: <[email protected]>
To: larson
Cc: zap, jmf, metsch, sestokas
Subject: kernel build error with v32supportos
Date: Tue, 31 Jan 95 13:50:25 -0500
From: "Joshua M. Friedman, OSF/UNIX SDE 381-1548" <jmf>
X-Mts: smtp

Mike, the v32supportos build failed last night in the kernel; here are the
errors from:
	/usr/sde/osf1/build/v32supportos/Logs/kernel.bld.log.errors

The following errors occurred in the kernel build
++++++++++++++++++++++++++++++
MATCH in /usr/sde/osf1/build/v32supportos.bld/Logs/kernel.bld.log at line 3063
/usr/sde/osf1/build/v32supportos.bld/tools/alpha_OSF1/alpha/acc/usr/lib/cmplrs/cc/cfe: Error: ./mach/machUser.c, line 2477: member of structure or union required
        InP->init_port_setType.msg_type_long_number  =  init_port_setCnt;
        -----^
>> *** Exit 1


Can you please look into this problem?  Thanks...	-josh


- - ------- Message 2

Return-Path: larson 
Delivery-Date: Tue, 31 Jan 95 15:04:31 -0500
Return-Path: larson
Received: from wasted.zk3.dec.com by flambe.zk3.dec.com; (5.65/1.1.8.2/30Mar94-0502PM)
	id AA06022; Tue, 31 Jan 1995 15:04:28 -0500
Received: from localhost by wasted.zk3.dec.com; (5.65/1.1.8.2/22Dec94-1214PM)
	id AA09537; Tue, 31 Jan 1995 15:04:27 -0500
Message-Id: <[email protected]>
To: jmf
Cc: zap, metsch, sestokas, meg
Subject: Re: kernel build error with v32supportos 
Date: Tue, 31 Jan 95 15:04:27 -0500
From: larson
X-Mts: smtp


Josh,

The error is in a file called machUser.c. This file gets generated
by MIG (the Mach Interface Generator) from some other files. So
something is wrong in either one of the mach definition files, or
in MIG itself. Since I don't have access to v32supportos.bld, I
can't tell for sure.

BTW - this error has nothing to do with the submit that I made
last evening.



					Mike Larson

- - ------- End of Forwarded Messages



- ------- End of Forwarded Message


------- End of Forwarded Message