[S] 3.2A -> 3.2C frustration (gcc fix)

From: <cyrmiche_at_sidoci.qc.ca>
Date: Tue, 03 Oct 1995 15:24:57 cdt

Last week i wrote : (the fix for gcc is include at the end)

>We,ve just upgraded one of our dev machine from DEC OSF/1 3.2A to 3.2C
>and are experiencing a VERY frustrating situation. Because of gcc
>compile errors, we've decided to go back to 3.2A. This is NOT
>possible.The system does not reconize anymore the ADVFS partitions
>that were locally installed on that machine.

> ...

>Does this means that section 4.4.2.4 should specify that there is a
>change in the matadata structure in rel 3.2C ?

Don't worry ... the Metastructure of ADVFS did not change between 3.2a
and 3.2c ...

What happend is that the gentlement that switch back from 3.2c to 3.2c
did not wait for the disk to sync before he unpluged our BA350
hot-swappable disk. And replug the 3.2a ... and as you can imagine sync
3.2c stuff over 3.2a (OUTCH !). The person didn't told me and restore
only the root partition (ufs). Funny problem anyway.

I must apologize to the list for giving the impression that advfs was
wrong. For the first WE've made this one occur

I wish to tanks :

golden_at_falcon.invincible.com
olle_at_cb.uu.se (Olle Eriksson)
benites_at_burnsville.cs.unca.edu
Tim Mooney <mooney_at_dogbert.cc.ndsu.NoDak.edu>

For pointing the error in gcc and supplying a fix

----- Received mail -----

----- golden_at_falcon.invincible.com -----

*** config/alpha/alpha.h~ Thu Jun 15 09:08:45 1995
- --- config/alpha/alpha.h Fri Jun 23 13:09:21 1995
***************
*** 43,49 ****
  #define SIGNED_CHAR_SPEC "%{funsigned-char:-D__CHAR_UNSIGNED__}"
  
  /* No point in running CPP on our assembler output. */
! #define ASM_SPEC "-nocpp %{pg}"
  
  /* Under OSF/1, -p and -pg require -lprof1. */
  
- --- 43,49 ----
  #define SIGNED_CHAR_SPEC "%{funsigned-char:-D__CHAR_UNSIGNED__}"
  
  /* No point in running CPP on our assembler output. */
! #define ASM_SPEC "-g3 -nocpp %{pg}"
  
  /* Under OSF/1, -p and -pg require -lprof1. */
  
----- olle_at_cb.uu.se (Olle Eriksson) -----

A ggc-patch har been published about ten times in this group, much
easier to get that patch than to swap the OS


I received messages from a number of people experiencing the
same problem as well as two offering solutions. Jason Yanowitz
(yanowitz_at_eternity.cs.umass.edu) pointed out that in most cases
DEC's cc produces better code. Since I also need g++, I was
still stuck. The answer came from Dan Riley
<dsr_at_lns598.lns.cornell.edu>
who wrote the note below. This patch allowed me to build gcc
although I did have to start from a clean distribution. Thanks Dan.
Here is Dan's note:

> mips-tfile, /tmp/ccaaouEa.s:0 index 4294967295 is out of bounds
for lstrs,
max is 0, mips-tfile.c line# 4643

Apparently in 3.2c 'as' stopped emitting some debugging records when
debugging is not selected. The patch below should fix it. Please let
me know if it works or not--I haven't had a chance to try it on the
release 3.2c, and I'd like some confirmation that it works before
distributing it more widely.

--
Dan Riley                          Internet: 
dsr_at_lns598.lns.cornell.edu
Wilson Lab, Cornell University     HEPNET/SPAN: lns598::dsr
(44630::dsr)
              "Distance means nothing/To me." -Kate Bush
*** config/alpha/alpha.h~       Thu Jun 15 09:08:45 1995
--- config/alpha/alpha.h        Fri Jun 23 13:09:21 1995
***************
*** 43,49 ****
  #define SIGNED_CHAR_SPEC "%{funsigned-char:-D__CHAR_UNSIGNED__}"
  
  /* No point in running CPP on our assembler output.  */
! #define ASM_SPEC "-nocpp %{pg}"
  
  /* Under OSF/1, -p and -pg require -lprof1.  */
  
--- 43,49 ----
  #define SIGNED_CHAR_SPEC "%{funsigned-char:-D__CHAR_UNSIGNED__}"
  
  /* No point in running CPP on our assembler output.  */
! #define ASM_SPEC "-g3 -nocpp %{pg}"
  
  /* Under OSF/1, -p and -pg require -lprof1.  */
  
----- benites_at_burnsville.cs.unca.edu -----
Did you recreate the /etc/fdmns files?
----- Tim Mooney <mooney_at_dogbert.cc.ndsu.NoDak.edu> -----
When we last communicated, cyrmiche_at_sidoci.qc.ca wrote:
)Hello managers,
 
)We,ve just upgraded one of our dev machine from DEC OSF/1 3.2A to 3.2C
)and are experiencing a VERY frustrating situation. Because of gcc
)compile errors, we've decided to go back to 3.2A.
This is a (somewhat) known problem that was caused by a change in the
format
of the assembler output.  There is a patch for gcc, to make it
understand,
and it's very short.
) This is NOT possible.
)The system does not reconize anymore the ADVFS partitions tha were
)locally installed on that machine.
That I'm not so sure on; we've only had OSF for a little while and I've
not
performed any upgrade installations -- just overwrite installations.
)Does anyone hace experienced the same problems ?
With gcc yes.  I can tell you that it's an easy fix.
)
)touch stmp-int-hdrs
)./xgcc -B./  -DIN_GCC   -g -I./include    -DNO_MEM -DNO_LONG_DOUBLE_IO
)-O0 -I. -
)c ./enquire.c
)mips-tfile, /tmp/ccaawsfa.s:0 index 4294967295 is out of bounds for
)lstrs, max i
)s 0, mips-tfile.c line# 4642
)*** Exit 1
)Stop.
)*** Exit 1
)Stop.
Tim
-- 
Tim Mooney                              mooney_at_toons.cc.ndsu.NoDak.edu
Information Technology Services         (701) 231-1076 (Voice)
Room 242-J6, IACC Bldg.                 (701) 231-8541 (FAX)
North Dakota State University, Fargo, ND 58105
-----------------------------------------------------------------------
-----
>From 
ns1.nodak.edu!netnews1.nwnet.net!news.u.washington.edu!uw-beaver!uhog.m
t.edu!ne
ws.kei.com!travelers.mail.cornell.edu!newsstand.cit.cornell.edu!lnsnews
lns.corn
ell.edu!news Sun Sep 10 09:33:17 1995
Xref: ns1.nodak.edu comp.unix.osf.osf1:10940 comp.sys.dec:34176
Path: 
ns1.nodak.edu!netnews1.nwnet.net!news.u.washington.edu!uw-beaver!uhog.m
t.edu!ne
ws.kei.com!travelers.mail.cornell.edu!newsstand.cit.cornell.edu!lnsnews
lns.corn
ell.edu!news
From: dsr_at_lns598.lns.cornell.edu (Daniel S. Riley)
Newsgroups: comp.unix.osf.osf1,comp.sys.dec
Subject: Re: dUnix 3.2c and gcc 2.7
Date: 06 Sep 1995 15:14:50 -0400
Organization: Wilson Lab, Cornell U., Ithaca, NY  14853
Lines: 35
Sender: dsr_at_lns100.lns.cornell.edu
Message-ID: <shu46qj7px.fsf_at_lns100.lns.cornell.edu>
References: <42kfn0$8qj_at_amhux3.amherst.edu>
NNTP-Posting-Host: lns100.lns.cornell.edu
In-reply-to: pdchapin_at_news.amherst.edu's message of 6 Sep 1995 15:45:36
GMT
X-Newsreader: Gnus v5.0
In article <42kfn0$8qj_at_amhux3.amherst.edu>,
pdchapin_at_news.amherst.edu (PAUL D. CHAPIN) writes:
pdchapin> mips-tfile, /tmp/ccaaFfda.s:0 index 4294967295 is out of
bounds for
pdchapin>       lstrs, max is 0, mips-tfile.c line# 4643
pdchapin> *** Exit 1
3.2c changed the behavior of the assembler slightly--it no longer
includes some debugging records that it used to by default.  The
following patch should fix it--apply the patch, do a 'make clean' and
rebuild.
*** config/alpha/alpha.h~       Thu Jun 15 09:08:45 1995
--- config/alpha/alpha.h        Fri Jun 23 13:09:21 1995
***************
*** 43,49 ****
  #define SIGNED_CHAR_SPEC "%{funsigned-char:-D__CHAR_UNSIGNED__}"
  
  /* No point in running CPP on our assembler output.  */
! #define ASM_SPEC "-nocpp %{pg}"
  
  /* Under OSF/1, -p and -pg require -lprof1.  */
  
--- 43,49 ----
  #define SIGNED_CHAR_SPEC "%{funsigned-char:-D__CHAR_UNSIGNED__}"
  
  /* No point in running CPP on our assembler output.  */
! #define ASM_SPEC "-g3 -nocpp %{pg}"
  
  /* Under OSF/1, -p and -pg require -lprof1.  */
  
Thanks again to the list !!! 
Michel Cyr
SIDOCI enr.
IST-Integration inc.
(514)252-4333 ex 258
Michel.Cyr_at_sidoci.qc.ca

Received on Tue Oct 03 1995 - 21:00:18 NZDT