[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference smurf::buildhelp

Title:USG buildhelp questions/answers
Moderator:SMURF::FILTER
Created:Mon Apr 26 1993
Last Modified:Mon Jan 20 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2763
Total number of notes:5802

338.0. "Libc build problem in my sandbox" by SMURF::FILTER (Automatic Posting Software - mail to flume::puck) Fri Sep 24 1993 14:01

Date Of Receipt: 	24-SEP-1993 12:30:25.38
From: 	QUARRY::njo "Neil O'Brien USG/DE"
To: 	odehelp@DEC:.zko.quarry
CC: 	brett njo tomp
Subj: 	Libc build problem in my sandbox

Hi,

I just discovered today that, when I do a build of libc.a, I get the
following messages at the end of the build:


>**Warning**: The following non-ANSI, unreserved global symbols do not have
>weak/strong counterparts and may cause name space pollution:
>
>itrunc
>
>**Warning**: Found references to the following non-ANSI,
>unreserved global symbols:
>
>trunc
>
>These references should be replaced with reserved versions which do not
>cause name space pollution.


For some reason, itrunc.o is being built from a source module located
in /usr/sde/osf1/build/agosmaint.bl6/src/usr/ccs/lib/libc/itrunc.c.
                       ^^^^^^^^^^^^^

I recently submitted a change to libc to move the itrunc.c module from
src/usr/ccs/lib/libc to src/usr/ccs/lib/libc/alpha.  To do that, I did 
the following:

    1.  Defuncted src/usr/ccs/lib/libc/itrunc.c

    2.  Deleted the itrunc.o entry from src/usr/ccs/lib/libc/Makefile

    3.  Bcreated src/usr/ccs/lib/libc/alpha/itrunc.c

    4.  Added the itrunc.o entry to src/usr/ccs/lib/libc/machdep.mk


Did I do somethinng wrong here that would cause itrunc.c to be picked
up from agosmaint.bl6/src/usr/ccs/lib/libc instead of agosminor?

Thanks.

Neil


PS:  At Brett Sampson's suggestion, I checked the libc build log 
usr_ccs_lib_libc.log in /share/buffer/build/submits.dsk1/agosminor/Logs
and the problem doesn't show up there.  It only shows up when I do a
libc build in my sandbox.

T.RTitleUserPersonal
Name
DateLines
338.1Re: Libc build problem in my sandboxSMURF::FILTERAutomatic Posting Software - mail to flume::puckFri Sep 24 1993 15:0571
Date Of Receipt: 	24-SEP-1993 13:22:24.42
From: 	FLUME::"[email protected]" "Grant Van Dyck"
To: 	Neil O'Brien USG/DE <[email protected]>
CC: 	[email protected], [email protected], [email protected]
Subj: 	Re: Libc build problem in my sandbox

Try again. I believe you got caught in a network glitch, where the 
rc_files from the bld disk were moved to nightly, but the update process
crashed before they could be modified to reflect the nightly tree.

Therefore your SB expects its backing tree to be agosminor.bld, a tree that
you cannot mount. And when that fails, the BACKED_SOURCEDIR variable takes
you to the next tree down the pipe, the backingtree for agosminor, which is 
agosmaint.bl6.

We've fixed this and hopefully you'll be OK now.

	-Grant


| Hi,
| 
| I just discovered today that, when I do a build of libc.a, I get the
| following messages at the end of the build:
| 
| 
| >**Warning**: The following non-ANSI, unreserved global symbols do not have
| >weak/strong counterparts and may cause name space pollution:
| >
| >itrunc
| >
| >**Warning**: Found references to the following non-ANSI,
| >unreserved global symbols:
| >
| >trunc
| >
| >These references should be replaced with reserved versions which do not
| >cause name space pollution.
| 
| 
| For some reason, itrunc.o is being built from a source module located
| in /usr/sde/osf1/build/agosmaint.bl6/src/usr/ccs/lib/libc/itrunc.c.
|                        ^^^^^^^^^^^^^
| 
| I recently submitted a change to libc to move the itrunc.c module from
| src/usr/ccs/lib/libc to src/usr/ccs/lib/libc/alpha.  To do that, I did 
| the following:
| 
|     1.  Defuncted src/usr/ccs/lib/libc/itrunc.c
| 
|     2.  Deleted the itrunc.o entry from src/usr/ccs/lib/libc/Makefile
| 
|     3.  Bcreated src/usr/ccs/lib/libc/alpha/itrunc.c
| 
|     4.  Added the itrunc.o entry to src/usr/ccs/lib/libc/machdep.mk
| 
| 
| Did I do somethinng wrong here that would cause itrunc.c to be picked
| up from agosmaint.bl6/src/usr/ccs/lib/libc instead of agosminor?
| 
| Thanks.
| 
| Neil
| 
| 
| PS:  At Brett Sampson's suggestion, I checked the libc build log 
| usr_ccs_lib_libc.log in /share/buffer/build/submits.dsk1/agosminor/Logs
| and the problem doesn't show up there.  It only shows up when I do a
| libc build in my sandbox.
| 

338.2Re: Libc build problem in my sandboxSMURF::FILTERAutomatic Posting Software - mail to flume::puckFri Sep 24 1993 16:1218
Date Of Receipt: 	24-SEP-1993 14:29:40.06
From: 	FLUME::jmcg "Jim McGinness"
To: 	flume::njo, flume::odehelp
CC: 	flume::brett, flume::tomp
Subj: 	Re:  Libc build problem in my sandbox

This looks like a bug in the VPATH handling stuff.  It should not, but
apparently does, preferrentially look up things in backing trees
before looking along the VPATH.

The fastest workaround I can think of is to create a symbolic link
itrunc.c -> alpha/itrunc.c in your sandbox to head off the error.

The nightly build has a truncated backing tree path, so it wouldn't
see this problem.

	-- jmcg

338.3Re: Libc build problem in my sandboxSMURF::FILTERAutomatic Posting Software - mail to flume::puckMon Sep 27 1993 10:0110
Date Of Receipt: 	24-SEP-1993 16:33:29.67
From: 	FLUME::jmcg "Jim McGinness"
To: 	flume::jmcg, flume::njo, flume::odehelp
CC: 	flume::brett, flume::tomp
Subj: 	Re:  Libc build problem in my sandbox

I like Grant's answer better.

	-- jmcg