From mikets at frontline-pcb.com  Wed Apr  1 09:33:11 2009
From: mikets at frontline-pcb.com (Michael Tseitlin)
Date: Wed, 1 Apr 2009 09:33:11 +0300
Subject: Tcsh.exe crashes on multicore computers
Message-ID: <830C82B0B852B9488F0A944853D4D01205F2931F@uganda.FRONTLINE-PCB.COM>

Hi,
 
I've reported several time to this list about intermittent crashes of
tcsh.exe that I experience for a long time. Before the crashes tcsh
usually reports about memory inconsistency etc.
 
Now I've figured out that if I define a CPU affinity for tcsh.exe to run
on a single CPU - the crashes don't happen any more.
The simplest way to do it is using Task Manager. More advanced tools are
those that monitor starting processes and set the affinity
automatically.
 
What I did finally - just put the following code into nt_init function
of win32/support.c:
 
SetProcessAffinityMask(GetCurrentProcess(), 1);
 
I might be a bit more creative and set a random CPU in the range of
existing ones.
 
The best way is certainly to solve the real memory problems, but I'm not
proficient enough in tcsh code to figure it out. The above solution is
at least sufficient for me.
 
Regards,
 
Michael
 

From trineo at gmail.com  Wed Apr  1 23:56:29 2009
From: trineo at gmail.com (Neon)
Date: Wed, 1 Apr 2009 16:56:29 -0400
Subject: Subscript out of range error on existing working script
Message-ID: <ab88c5be0904011356u69235b28if0972157bed5bb37@mail.gmail.com>

Just downloaded tcsh 6.10.0 source tgz and looks like this bug is fixed.

Will there be a patch release to 6.14.xx with this fix?

I am in a situation where I do not control the tcsh versions on the
target machines.
Looks like Red Hat Enterprise Linus (RHEL) 5.3 includes tcsh 6.14.00
with this problem.
If there was a version 6.14.01 (Red Hat and others may be more
inclined to include it in a patch fix for there OS)

Is there a regression test for this bug in tcsh? Where is your bug database?

Regards,

Neon

> Can you try 6.16?
>
> christos
>
| I have an existing script that worked fine under tcsh 6.12.00 which no
| longer works in version 6.14.00
|
| It errors out with "Subscript is out of range." message in version 6.12.00
|
| Here is a test script to show the problem.
|
| -- begin test.tcsh --
| #!/bin/tcsh
| set second_til_last = "${argv[2-]}"
| echo "2nd to last arguments = $second_til_last"
| -- end of test.tcsh ---
|
| This originally is a csh script which would set the second_til_last
| variable to a blank string (i.e. "")


From trineo at gmail.com  Thu Apr  2 00:31:06 2009
From: trineo at gmail.com (Neon)
Date: Wed, 1 Apr 2009 17:31:06 -0400
Subject: Subscript out of range error on existing working script
In-Reply-To: <ab88c5be0904011356u69235b28if0972157bed5bb37@mail.gmail.com>
References: <ab88c5be0904011356u69235b28if0972157bed5bb37@mail.gmail.com>
Message-ID: <ab88c5be0904011431m3631ad15q2f027b897bb002cf@mail.gmail.com>

I have to take that back what I said in the previous post, this also
fails in tcsh 6.16.00
my test script was referring to /bin/tcsh (v6.12.00) where this used
to work when I ran that.

Interesting thing is that I also compiled 6.14.00 from a refresh src
tgz and that seems to work on my local box.
Under further investigation.

Here are the tcsh versions with this subscript error for my test script:
tcsh 6.14.00 (Astron) 2005-03-25 (i386-intel-linux) options
wide,nls,dl,al,kan,sm,rh,color,filec
tcsh 6.16.00 (Astron) 2008-09-30 (i386-intel-linux) options
wide,nls,dl,al,kan,sm,rh,color,filec

Here are the tcsh versions without the subscript error for my test script:
tcsh 6.14.00 (Astron) 2005-03-25 (i386-intel-linux) options
wide,nls,dl,al,kan,rh,color,filec
tcsh 6.12.00 (Astron) 2002-07-23 (i386-intel-linux) options
8b,nls,dl,al,kan,rh,color,dspm,filec

So, it looks like with the "sm" option in version 6.14.00 and 6.16.00
we have this subscript error for my test script.

Regards,

Neon

On Wed, Apr 1, 2009 at 4:56 PM, Neon <trineo at gmail.com> wrote:
> Just downloaded tcsh 6.10.0 source tgz and looks like this bug is fixed.
>
> Will there be a patch release to 6.14.xx with this fix?
>
> I am in a situation where I do not control the tcsh versions on the
> target machines.
> Looks like Red Hat Enterprise Linus (RHEL) 5.3 includes tcsh 6.14.00
> with this problem.
> If there was a version 6.14.01 (Red Hat and others may be more
> inclined to include it in a patch fix for there OS)
>
> Is there a regression test for this bug in tcsh? Where is your bug database?
>
> Regards,
>
> Neon
>
>> Can you try 6.16?
>>
>> christos
>>
> | I have an existing script that worked fine under tcsh 6.12.00 which no
> | longer works in version 6.14.00
> |
> | It errors out with "Subscript is out of range." message in version 6.12.00
> |
> | Here is a test script to show the problem.
> |
> | -- begin test.tcsh --
> | #!/bin/tcsh
> | set second_til_last = "${argv[2-]}"
> | echo "2nd to last arguments = $second_til_last"
> | -- end of test.tcsh ---
> |
> | This originally is a csh script which would set the second_til_last
> | variable to a blank string (i.e. "")


From christos at zoulas.com  Thu Apr  2 00:56:05 2009
From: christos at zoulas.com (Christos Zoulas)
Date: Wed, 1 Apr 2009 17:56:05 -0400
Subject: Subscript out of range error on existing working script
In-Reply-To: <ab88c5be0904011431m3631ad15q2f027b897bb002cf@mail.gmail.com>
	from Neon (Apr  1,  5:31pm)
Message-ID: <20090401215605.1EF8656550@rebar.astron.com>

On Apr 1,  5:31pm, trineo at gmail.com (Neon) wrote:
-- Subject: Re: Subscript out of range error on existing working script

| I have to take that back what I said in the previous post, this also
| fails in tcsh 6.16.00
| my test script was referring to /bin/tcsh (v6.12.00) where this used
| to work when I ran that.
| 
| Interesting thing is that I also compiled 6.14.00 from a refresh src
| tgz and that seems to work on my local box.
| Under further investigation.
| 
| Here are the tcsh versions with this subscript error for my test script:
| tcsh 6.14.00 (Astron) 2005-03-25 (i386-intel-linux) options
| wide,nls,dl,al,kan,sm,rh,color,filec
| tcsh 6.16.00 (Astron) 2008-09-30 (i386-intel-linux) options
| wide,nls,dl,al,kan,sm,rh,color,filec
| 
| Here are the tcsh versions without the subscript error for my test script:
| tcsh 6.14.00 (Astron) 2005-03-25 (i386-intel-linux) options
| wide,nls,dl,al,kan,rh,color,filec
| tcsh 6.12.00 (Astron) 2002-07-23 (i386-intel-linux) options
| 8b,nls,dl,al,kan,rh,color,dspm,filec
| 
| So, it looks like with the "sm" option in version 6.14.00 and 6.16.00
| we have this subscript error for my test script.

Hmm, this should only be in 6.16... It is fix 24 in the fixes file:

 24. $x[A-B] did not complain for A out of range. (Cai Xianchao)

Yes, this is a behavior change, but it was done for symmetry.

christos


From avg at icyb.net.ua  Thu Apr 23 12:25:16 2009
From: avg at icyb.net.ua (Andriy Gapon)
Date: Thu, 23 Apr 2009 12:25:16 +0300
Subject: periodic annoyance with history -S
In-Reply-To: <46C1A41D.20303@icyb.net.ua>
References: <20070811085747.25F9D56407@rebar.astron.com>
	<46C1A41D.20303@icyb.net.ua>
Message-ID: <49F033FC.2070404@icyb.net.ua>

on 14/08/2007 15:46 Andriy Gapon said the following:
> Christos,
> http://bugs.gw.com/view.php?id=58
> 
> Thank you!

It's been a long while - I now found some time to analyze this issue
more and I put some additional comments in the bug tracker.


-- 
Andriy Gapon


From christos at zoulas.com  Fri Apr 24 17:35:36 2009
From: christos at zoulas.com (Christos Zoulas)
Date: Fri, 24 Apr 2009 10:35:36 -0400
Subject: periodic annoyance with history -S
In-Reply-To: <49F033FC.2070404@icyb.net.ua> from Andriy Gapon (Apr 23, 12:25pm)
Message-ID: <20090424143536.E379F5654E@rebar.astron.com>

On Apr 23, 12:25pm, avg at icyb.net.ua (Andriy Gapon) wrote:
-- Subject: Re: periodic annoyance with history -S

| on 14/08/2007 15:46 Andriy Gapon said the following:
| > Christos,
| > http://bugs.gw.com/view.php?id=58
| > 
| > Thank you!
| 
| It's been a long while - I now found some time to analyze this issue
| more and I put some additional comments in the bug tracker.

You nailed it!

Index: sh.c
===================================================================
RCS file: /p/tcsh/cvsroot/tcsh/sh.c,v
retrieving revision 3.142
diff -u -u -r3.142 sh.c
--- sh.c	25 Feb 2009 22:54:53 -0000	3.142
+++ sh.c	24 Apr 2009 14:35:00 -0000
@@ -139,6 +139,7 @@
     Char	  HIST;
     int	  cantell;
     struct Bin	  B;
+    int		  justpr;
 };
 
 static	int		  srccat	(Char *, Char *);
@@ -1470,6 +1471,7 @@
     st->alvec		= alvec;
     st->onelflg		= onelflg;
     st->enterhist	= enterhist;
+    st->justpr		= justpr;
     if (hflg)
 	st->HIST	= HIST;
     else
@@ -1568,6 +1570,7 @@
 	HIST	= st->HIST;
     enterhist	= st->enterhist;
     cantell	= st->cantell;
+    justpr	= st->justpr;
 
     if (st->argv != NULL)
 	setq(STRargv, st->argv, &shvhed, VAR_READWRITE);


From avg at icyb.net.ua  Fri Apr 24 17:44:53 2009
From: avg at icyb.net.ua (Andriy Gapon)
Date: Fri, 24 Apr 2009 17:44:53 +0300
Subject: periodic annoyance with history -S
In-Reply-To: <20090424143536.E379F5654E@rebar.astron.com>
References: <20090424143536.E379F5654E@rebar.astron.com>
Message-ID: <49F1D065.9080500@icyb.net.ua>

on 24/04/2009 17:35 Christos Zoulas said the following:
> On Apr 23, 12:25pm, avg at icyb.net.ua (Andriy Gapon) wrote:
> -- Subject: Re: periodic annoyance with history -S
> 
> | on 14/08/2007 15:46 Andriy Gapon said the following:
> | > Christos,
> | > http://bugs.gw.com/view.php?id=58
> | > 
> | > Thank you!
> | 
> | It's been a long while - I now found some time to analyze this issue
> | more and I put some additional comments in the bug tracker.
> 
> You nailed it!

Christos, thank you very much for the fix!

> Index: sh.c
> ===================================================================
> RCS file: /p/tcsh/cvsroot/tcsh/sh.c,v
> retrieving revision 3.142
> diff -u -u -r3.142 sh.c
> --- sh.c	25 Feb 2009 22:54:53 -0000	3.142
> +++ sh.c	24 Apr 2009 14:35:00 -0000
> @@ -139,6 +139,7 @@
>      Char	  HIST;
>      int	  cantell;
>      struct Bin	  B;
> +    int		  justpr;
>  };
>  
>  static	int		  srccat	(Char *, Char *);
> @@ -1470,6 +1471,7 @@
>      st->alvec		= alvec;
>      st->onelflg		= onelflg;
>      st->enterhist	= enterhist;
> +    st->justpr		= justpr;
>      if (hflg)
>  	st->HIST	= HIST;
>      else
> @@ -1568,6 +1570,7 @@
>  	HIST	= st->HIST;
>      enterhist	= st->enterhist;
>      cantell	= st->cantell;
> +    justpr	= st->justpr;
>  
>      if (st->argv != NULL)
>  	setq(STRargv, st->argv, &shvhed, VAR_READWRITE);


-- 
Andriy Gapon


From christos at zoulas.com  Fri Apr 24 18:38:55 2009
From: christos at zoulas.com (Christos Zoulas)
Date: Fri, 24 Apr 2009 11:38:55 -0400
Subject: periodic annoyance with history -S
In-Reply-To: <49F1D065.9080500@icyb.net.ua> from Andriy Gapon (Apr 24,  5:44pm)
Message-ID: <20090424153855.74EEE5654E@rebar.astron.com>

On Apr 24,  5:44pm, avg at icyb.net.ua (Andriy Gapon) wrote:
-- Subject: Re: periodic annoyance with history -S

| Christos, thank you very much for the fix!

You did all the work :-)

christos