Startup failure

Startup failure

[Klaus Schilling]

I can't get the freshly compiled 2.0.5 to start up.
The debug version says the following:

pessoa@debian:~/guile-2.0.5/meta$ ./gdb-uninstalled-guile
GNU gdb 6.3-debian
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and
you are
welcome to change it and/or distribute copies of it under certain
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for
details.
This GDB was configured as "i386-linux"...Using host libthread_db
library "/lib/libthread_db.so.1".

(gdb) r
Starting program: /home/pessoa/guile-2.0.5/libguile/.libs/lt-guile
[Thread debugging using libthread_db enabled]
[New Thread 16384 (LWP 11502)]
Backtrace:
In ice-9/boot-9.scm:
 149: 2 [catch #t #<catch-closure 81e8220> ...]
 157: 1 [#<procedure 81b38c0 ()>]
In unknown file:
   ?: 0 [catch-closure]

ERROR: In procedure catch-closure:
ERROR: Throw to key `decoding-error' with args `("scm_from_stringn"
"input locale conversion error" 22 #vu8(47 104 111 109 101 47 112 101
115 115 111 97 47 103
117 105 108 101 45 50 46 48 46 53 47 108 105 98 103 117 105 108 101 47
46 108 105 98 115 47 108 116 45 103 117 105 108 101))'.

Program exited with code 01.

The long character vector is probably decimal ascii for /home/pessoa/guile-2.0.5/libguile/.libs/lt-guile

The checks in the standalone directory also fail, maybe for the same reason.

make  check-TESTS
make[5]: Entering directory
`/home/pessoa/guile-2.0.5/test-suite/standalone'
Backtrace:
In ice-9/boot-9.scm:
 149: 12 [catch #t #<catch-closure 81e8220> ...]
 157: 11 [#<procedure 81b38c0 ()>]
In unknown file:
   ?: 10 [catch-closure]
In ice-9/boot-9.scm:
  63: 9 [call-with-prompt prompt0 ...]
In ice-9/eval.scm:
 407: 8 [eval # #]
In ice-9/boot-9.scm:
2111: 7 [save-module-excursion #<procedure 81a1060 at
ice-9/boot-9.scm:3646:3 ()>]
3651: 6 [#<procedure 81a1060 at ice-9/boot-9.scm:3646:3 ()>]
1380: 5 [%start-stack load-stack ...]
1385: 4 [#<procedure 81fc4c8 ()>]
In unknown file:
   ?: 3 [primitive-load
"/home/pessoa/guile-2.0.5/test-suite/standalone/test-system-cmds"]
In ice-9/eval.scm:
 392: 2 [eval # #]
 368: 1 [eval # #]
In unknown file:
   ?: 0 [system* "guile" "-c" "(exit 42)"]

ERROR: In procedure system*:
ERROR: Stack overflow
Backtrace:
In ice-9/boot-9.scm:
 149: 14 [catch #t #<catch-closure 81e8220> ...]
 157: 13 [#<procedure 81b38c0 ()>]
In unknown file:
   ?: 12 [catch-closure]
In ice-9/boot-9.scm:
  63: 11 [call-with-prompt prompt0 ...]
In ice-9/eval.scm:
 407: 10 [eval # #]
In ice-9/boot-9.scm:
2111: 9 [save-module-excursion #<procedure 81a1060 at
ice-9/boot-9.scm:3646:3 ()>]
3651: 8 [#<procedure 81a1060 at ice-9/boot-9.scm:3646:3 ()>]
1380: 7 [%start-stack load-stack ...]
1385: 6 [#<procedure 81fc4c8 ()>]
In unknown file:
   ?: 5 [primitive-load
"/home/pessoa/guile-2.0.5/test-suite/standalone/test-system-cmds"]
In ice-9/eval.scm:
 392: 4 [eval # #]
 368: 3 [eval # #]
In unknown file:
   ?: 2 [system* "guile" "-c" "(exit 42)"]
In ice-9/boot-9.scm:
 105: 1 [#<procedure 81a1280 at ice-9/boot-9.scm:97:6 (thrown-k
. args)> stack-overflow ...]
  65: 0 [abort-to-prompt catch4 stack-overflow #f "Stack overflow" #f
#f]

ice-9/boot-9.scm:65:2: In procedure abort-to-prompt:
ice-9/boot-9.scm:65:2: Stack overflow

at this point, `make check' hangs indefinitely.


Klaus Schilling

Re: Startup failure

[Ludovic Courtès]

Hi Klaus,

Klaus Schilling <schilling.klaus@web.de> skribis:

> ERROR: Throw to key `decoding-error' with args `("scm_from_stringn"
> "input locale conversion error" 22 #vu8(47 104 111 109 101 47 112 101
> 115 115 111 97 47 103
> 117 105 108 101 45 50 46 48 46 53 47 108 105 98 103 117 105 108 101 47
> 46 108 105 98 115 47 108 116 45 103 117 105 108 101))'.

Can you check whether this is the same problem as
<http://lists.gnu.org/archive/html/bug-guile/2012-02/msg00082.html>, and
whether the workaround applies?

Thanks,
Ludo’.

Re: Startup failure

[Klaus Schilling]

From: ludo@gnu.org (Ludovic Courtès)
Subject: Re: Startup failure
Date: Fri, 01 Jun 2012 18:26:26 +0200

Hi Ludovic,

> Hi Klaus,
> 
> Klaus Schilling <schilling.klaus@web.de> skribis:
> 
> > ERROR: Throw to key `decoding-error' with args `("scm_from_stringn"
> > "input locale conversion error" 22 #vu8(47 104 111 109 101 47 112 101
> > 115 115 111 97 47 103
> > 117 105 108 101 45 50 46 48 46 53 47 108 105 98 103 117 105 108 101 47
> > 46 108 105 98 115 47 108 116 45 103 117 105 108 101))'.
> 
> Can you check whether this is the same problem as
> <http://lists.gnu.org/archive/html/bug-guile/2012-02/msg00082.html>, and
> whether the workaround applies?

It is the same problem, and the workaround does apply, but still leads
to a nonfatal warning:
warning: failed to install locale: Invalid argument

Make check still hangs indefinitely at the same spot.

Thanks, 
Klaus Schilling

Re: Startup failure

[Ludovic Courtès]

Hi Klaus,

Klaus Schilling <schilling.klaus@web.de> skribis:

>> Klaus Schilling <schilling.klaus@web.de> skribis:
>> 
>> > ERROR: Throw to key `decoding-error' with args `("scm_from_stringn"
>> > "input locale conversion error" 22 #vu8(47 104 111 109 101 47 112 101
>> > 115 115 111 97 47 103
>> > 117 105 108 101 45 50 46 48 46 53 47 108 105 98 103 117 105 108 101 47
>> > 46 108 105 98 115 47 108 116 45 103 117 105 108 101))'.
>> 
>> Can you check whether this is the same problem as
>> <http://lists.gnu.org/archive/html/bug-guile/2012-02/msg00082.html>, and
>> whether the workaround applies?
>
> It is the same problem, and the workaround does apply, but still leads
> to a nonfatal warning:
> warning: failed to install locale: Invalid argument

Yes, it’s non-fatal.  It means that the locale you chose isn’t available
on your system.

> Make check still hangs indefinitely at the same spot.

This one?

> ERROR: In procedure system*:
> ERROR: Stack overflow

Could it be that Guile was compiled with -O0?  It should be compiled
with -O2.

Thanks,
Ludo’.

Re: Startup failure

[Klaus Schilling]

From: ludo@gnu.org (Ludovic Courtès)
Subject: Re: Startup failure
Date: Sat, 02 Jun 2012 14:40:46 +0200

Hi Ludovic,

> Hi Klaus,
> 
> 
> This one?
> 
> > ERROR: In procedure system*:
> > ERROR: Stack overflow
> 
> Could it be that Guile was compiled with -O0?  It should be compiled
> with -O2.
> 
> Thanks,
> Ludo’.

The Makefile says 
CFLAGS = -g -O2

Is there a way for gcc to mess this up and compile without
optimisation as a fallback? It's easy to miss out on some warnings
during the compilation.

Other procedures from that fail in a similar manner are sigaction with some
signalnumbers (I had to fumble with top-repl.scm in order to make the
repla start up, it contains a sigaction for SIGINT, so disabled the
check (provided? 'posix) and continued automatically with the non
posix version which does not manipulate and restore signal handlers
within a dynamic-wind).  The same problem was already in 2.0.0. In
1.8.8, system* and sigaction work flawlessly. Also, system without a
star works well even with 2.0.5. So the problem rests with some, but
not all posix procedures.

Klaus Schilling

Re: Startup failure

[Ludovic Courtès]

Hi,

Sorry for the late reply.

Klaus Schilling <schilling.klaus@web.de> skribis:

> From: ludo@gnu.org (Ludovic Courtès)
> Subject: Re: Startup failure
> Date: Sat, 02 Jun 2012 14:40:46 +0200

[...]

>> > ERROR: In procedure system*:
>> > ERROR: Stack overflow

Could you check whether this still occurs with 2.0.6?

If it does, could you run:

  ./meta/gdb-uninstalled-guile

and in GDB:

  b scm_error
  run

then at the REPL:

  (system "guile -c '(exit 42)'")

Thanks,
Ludo’.

Re: Startup failure

[Klaus Schilling]

From: ludo@gnu.org (Ludovic Courtès)
Subject: Re: Startup failure
Date: Mon, 09 Jul 2012 11:39:29 +0200

> Hi,
> 
> Sorry for the late reply.
> 
> Klaus Schilling <schilling.klaus@web.de> skribis:
> 
> > From: ludo@gnu.org (Ludovic Courtès)
> > Subject: Re: Startup failure
> > Date: Sat, 02 Jun 2012 14:40:46 +0200
> 
> [...]
> 
> >> > ERROR: In procedure system*:
> >> > ERROR: Stack overflow
> 
> Could you check whether this still occurs with 2.0.6?
> 
> If it does, could you run:
> 
>   ./meta/gdb-uninstalled-guile
> 
> and in GDB:
> 
>   b scm_error
>   run
> 
> then at the REPL:
> 
>   (system "guile -c '(exit 42)'")
> 
> Thanks,
> Ludo’.

It was still not working, and the repl does not even show up. The
stack overflow occurs already prior to that. Thus it is not possible
to do the last step. 

However, when configuring --without-thread, nothing like that occurs
and guile starts up regularly. In that case, the last expression
returns something like	`$1 = 10752'.

Again, disabling threads was not necessary in guile-1.8.

Thanks, 
Klaus Schilling