Re: [Various] libc/1161: Unable to get nscd to work on alpha

[Thorsten Kukuk <kukuk@suse.de>]

Hello,

On Sat, Jun 12, Andreas Jaeger wrote:

> 
> Is there anything we can do to fix nscd on alpha?  I fear this is a
> compiler bug:-(.  Richard, can you help?

It's an older bug which I have already send twice to the egcs people
without getting an reaction.


egcs miscompiles the following lines from nscd/cache.c:

  /* Put the new entry in the first position.  */
  do
    newp->next = table->array[hash];
  while (! compare_and_swap ((volatile long int *) &table->array[hash],


This lines alone works, so there happens something other. I couldn't
identify the other pieces of code where it crashes yet. This is not
the only one.

  Thorsten


> 
> Andreas
> 

> Date: Sat Jun 12 21:31:40 1999
> From: Various
> Subject: libc/1161: Unable to get nscd to work
> 
> Topics:
>    libc/1161: Unable to get nscd to work
>    Re: libc/1161: Unable to get nscd to work
>    Re: libc/1161: Unable to get nscd to work
> 
> 
> ----------------------------------------------------------------------
> 
> Date: Sat, 12 Jun 1999 14:41:31 -0400
> From: sba@srl.caltech.edu
> To: bugs@gnu.org
> Subject: libc/1161: Unable to get nscd to work
> Message-Id: <199906121841.OAA06324@delysid.gnu.org>
> 
> 
> >Number:         1161
> >Category:       libc
> >Synopsis:       Unable to get nscd to work
> >Confidential:   no
> >Severity:       critical
> >Priority:       medium
> >Responsible:    libc-gnats
> >State:          open
> >Class:          sw-bug
> >Submitter-Id:   unknown
> >Arrival-Date:   Sat Jun 12 14:50:00 EDT 1999
> >Last-Modified:
> >Originator:     sba@srl.caltech.edu
> >Organization:
> net
> >Release:        2.1.1
> >Environment:
> Dec Alpha (LX164) running RH6.0
> >Description:
> After attempting to compile glibc-2.1.1 with several versions of egcs
> (1.1.2 and later), I have been unable to get a version of nscd that works.
> Either I attempt to compile glibc-2.1.1 with -O3 which fails to even compile
> with most (all?) versions of egcs between 1.1.2 and the latest 2.95 prerelease,
> or I compile with -O and get a copy of nscd that has increadibly poor
> performance:
> 
> > If I un-tar a file like glibc-2.1.1pre3.tar.gz which has files
> > owned by unknown users (on my system), nscd -d goes into a loop generating:
> > 
> > 852:    GETPWBYNAME (drepper)
> > 852: handle_request: request received (Version = 2)
> > 852:    GETPWBYNAME (drepper)
> > 852: handle_request: request received (Version = 2)
> > 852:    GETPWBYNAME (drepper)
> > 852: handle_request: request received (Version = 2)
> > 852:    GETPWBYNAME (drepper)
> > ...
> > 
> > until eventually after ~1min the first file is written to disk,
> > and everything repeats for the next file, i.e., it is generating
> > thousands of these messages per file rather than just 1.
> > 
> > With several copies of nscd running in the background (without -d) you
> > can see them all (and tar) eating up all the CPU cycles just trying to
> > extract the first file from the tarball.
> >How-To-Repeat:
> Attempt to compile glibc-2.1.1 on a RH6.0 alpha system with -O3 and watch it crash-n-burtn
> 
> Try running nscd compiled with -O and watch it go into a spin loop when it
> tries to cache and unknown user name when untarring some tar file from the net
> >Fix:
> >Audit-Trail:
> >Unformatted:
> 
> 
> ------------------------------
> 
> Date: Sat, 12 Jun 1999 15:20:01 -0400
> From: Andreas Jaeger <aj@arthur.rhein-neckar.de>
> To: libc-gnats@gnu.org
> Cc: gnats-admin@gnu.org
> Subject: Re: libc/1161: Unable to get nscd to work
> Message-Id: <199906121920.PAA03961@mescaline.gnu.org>
> 
> The following reply was made to PR libc/1161; it has been noted by GNATS.
> 
> From: Andreas Jaeger <aj@arthur.rhein-neckar.de>
> To: sba@srl.caltech.edu
> Cc: bugs@gnu.org
> Subject: Re: libc/1161: Unable to get nscd to work
> Date: 12 Jun 1999 21:03:14 +0200
> 
>  >>>>> sba  writes:
>  
>  >> Number:         1161
>  >> Category:       libc
>  >> Synopsis:       Unable to get nscd to work
>  
>  Which Linux kernel version are you running?
>  
>  >> Description:
>  sba> After attempting to compile glibc-2.1.1 with several versions of egcs
>  sba> (1.1.2 and later), I have been unable to get a version of nscd that works.
>  egcs has AFAIK some bugs on alpha and so far nobody looked into fixing
>  them :-(
>  
>  sba> Either I attempt to compile glibc-2.1.1 with -O3 which fails to even compile
>  sba> with most (all?) versions of egcs between 1.1.2 and the latest 2.95 prerelease,
>  sba> or I compile with -O and get a copy of nscd that has increadibly poor
>  sba> performance:
>  
>  >> If I un-tar a file like glibc-2.1.1pre3.tar.gz which has files
>  >> owned by unknown users (on my system), nscd -d goes into a loop generating:
>  Can you reproduce this in your ix86 machine?
>  
>  Andreas
>  -- 
>   Andreas Jaeger   aj@arthur.rhein-neckar.de    jaeger@informatik.uni-kl.de
>    for pgp-key finger ajaeger@aixd1.rhrk.uni-kl.de
> 
> 
> ------------------------------
> Date: Sat, 12 Jun 1999 12:22:57 -0700 (PDT)
> From: sba@srl.caltech.edu (Stuart Anderson)
> To: Andreas Jaeger <aj@arthur.rhein-neckar.de>
> Subject: Re: libc/1161: Unable to get nscd to work
> Message-Id: <199906121922.MAA19003@jelly.srl>
> Content-Type: text/plain; charset=US-ASCII
> Content-Transfer-Encoding: 7bit
> 
> According to Andreas Jaeger:
> > >>>>> sba  writes:
> > 
> > >> Number:         1161
> > >> Category:       libc
> > >> Synopsis:       Unable to get nscd to work
> > 
> > Which Linux kernel version are you running?
> 
> 2.2.7
> 
> > 
> > >> Description:
> > sba> After attempting to compile glibc-2.1.1 with several versions of egcs
> > sba> (1.1.2 and later), I have been unable to get a version of nscd that works.
> > egcs has AFAIK some bugs on alpha and so far nobody looked into fixing
> > them :-(
> 
> bummer.
> 
> > 
> > sba> Either I attempt to compile glibc-2.1.1 with -O3 which fails to even compile
> > sba> with most (all?) versions of egcs between 1.1.2 and the latest 2.95 prerelease,
> > sba> or I compile with -O and get a copy of nscd that has increadibly poor
> > sba> performance:
> > 
> > >> If I un-tar a file like glibc-2.1.1pre3.tar.gz which has files
> > >> owned by unknown users (on my system), nscd -d goes into a loop generating:
> > Can you reproduce this in your ix86 machine?
> 
> No, this problem does NOT exist on my ix86 machines which where running
> as identical as is possible versions of RH/kernel/libc/egcs/...
> 
> A while back I received a response from Thorsten Kukuk (of Linux/NIS+ fame)
> that you had to compile glibc with -O3 to avoid some egcs bugs on the
> Alpha, but I haven't had as much luck as he has had with that. However,
> my last correspondence from him was that his nscd on an Alpha box was
> crashing as well.
> 
> Both of the bug reports I put in today are the result of trying to run
> both Intel and Alpha boxes as NIS+ clients on a predominantly Sun network.
> My only remaining problem is getting a good stable nscd on both platforms.
> 
> 
> - - 
> Stuart Anderson  sba@srl.caltech.edu  http://www.srl.caltech.edu/personnel/sba
> ------------------------------
> 
> 
> 
> 
> End of forwardK8Hes2 Digest
> ***************************

> 
> -- 
>  Andreas Jaeger   aj@arthur.rhein-neckar.de    jaeger@informatik.uni-kl.de
>   for pgp-key finger ajaeger@aixd1.rhrk.uni-kl.de


-- 
Thorsten Kukuk      http://www.suse.de/~kukuk/        kukuk@suse.de
SuSE GmbH           Schanzaeckerstr. 10             90443 Nuernberg
Linux is like a Vorlon.  It is incredibly powerful, gives terse,
cryptic answers and has a lot of things going on in the background.