headers
Chisheng Huang | 7 Jun 2005 20:27

segmentation error with cmucl 19b pre1 and cmucl-2005-06 (x86, linux)

Hi,

The following pcore-2005-06 was generated by
  0. Run CMUCL (2005-06, x86, Linux).
  1. Compile and load a bunch of files into Lisp
  2. Quit Lisp.
  3. Run Lisp again.
  4. Load the fasl files into Lisp
  5. (ext:save-lisp "pcore-2005-06").

One of the packages defined in pcore-2005-06 is "V".

I get a segmentation violation error with this core file if I enter
"(in-package :v)" at the REPL right after I fire up CMUCL.

I do not have this problem with previous CMUCL snapshots (2005-05 and earlier).
Both CMUCL 19b pre1 and 2005-06 give me this error.   If the attached backtrace
does not ring a bell with anybody, I'll try to distill a simple test case to
reproduce the error.

Best,

-cph

abalone_104% lisp -noinit -core pcore-2005-06
CMU Common Lisp Snapshot (19B), running on abalone
With core: /home/cph/v/pcore-2005-06
Dumped on: Mon, 2005-06-06 21:42:03-07:00 on abalone
See <http://www.cons.org/cmucl/> for support information.
Loaded subsystems:
(Continue reading)

Permalink | Reply | 0 comments |
headers
Eric Marsden | 7 Jun 2005 23:43
Picon
Favicon

please help test the CMUCL 19b release


[message also posted to comp.lang.lisp]

CMUCL version 19b will be released in July. We would appreciate help
in testing the prerelease binaries that are available for various
platforms. Here's how you can help test:

   - download the pre-release binaries for your platform
   - read the release notes 
   - try to compile and run your favorite applications (note that FASL
     files from this release are binary-incompatible with previous
     CMUCL releases; you will need to recompile all source files)
   - report any problems you encounter to the cmucl-help mailing list
     (see <URL:http://www.cons.org/cmucl/support.html> for details of our
     mailing lists)

This release will include a number of significant changes from CMUCL 19a,
that are detailed in the release notes:

   - a port to the MacOS X platform
   - a port to NetBSD on x86 systems
   - improved bignum performance
   - fixes to the simple-streams implementation, and implementation of
     bivalent streams
   - many ANSI compliance problems identified by Paul Dietz's test suite
     have been fixed
   - many bugfixes

Prerelease binaries for x86/Linux, x86/FreeBSD, SPARC/Solaris and
PowerPC/Darwin (MacOS X) can be downloaded from the following URLs
(Continue reading)

Permalink | Reply | 28 comments |
headers
Paolo Amoroso | 8 Jun 2005 13:54
Picon
Favicon

Re: please help test the CMUCL 19b release


Eric Marsden <eric.marsden <at> free.fr> writes:

> CMUCL version 19b will be released in July. We would appreciate help
> in testing the prerelease binaries that are available for various

I use CMUCL Snapshot (19B) under Slackware Linux 10.0 with a
customized 2.6.10 kernel.

>    - try loading shared object files and ".o" object files using the
>      function EXT:LOAD-FOREIGN, and check that you can resolve symbols
>      in these libraries and call the corresponding functions. For
>      example:
>
>      ,----
>      | CL-USER> (ext:load-foreign "/usr/lib/libcrypt.so")
[...]

OK.

>    - try dumping an customized CMUCL image (using the function
>      EXT:SAVE-LISP) and restarting it.

I do this with McCLIM, and it works fine.

>    - try connecting to your X11 display:
>
>      ,----
>      | CL-USER> (require :clx)
>      | ; Loading #p"/opt/cmucl-19b-pre/lib/cmucl/lib/subsystems/clx-library.x86f".
(Continue reading)

Permalink | Reply | 14 comments |
headers
Raymond Toy | 8 Jun 2005 16:27
Picon
Favicon

Re: please help test the CMUCL 19b release


>>>>> "Paolo" == Paolo Amoroso <amoroso <at> mclink.it> writes:

    >> - try using the Motif debugger and inspector:
    >> 
    >> ,----
    >> | CL-USER> (require :clm)
    >> | ; Loading
    Paolo> #p"/opt/cmucl-19b-pre/lib/cmucl/lib/subsystems/clm-library.x86f".
    >> | ("CLM")

    Paolo> OK.

    >> | CL-USER> (inspect *package*)

    Paolo> * (inspect *package*)
    Paolo> Starting server:
    Paolo>    will_fork  = False
    Paolo>    will_trace = False
    Paolo>    No Inet domain socket created.
    Paolo>    path       = /tmp/.motif_socket-p4705
    Paolo> Waiting for connection.
    Paolo> Accepting client on Unix socket.
    Paolo> Accepted client on fd 4
    Paolo> Server not forking.
    Paolo> Warning:  Motif server died.
    Paolo> Status = :SIGNALED, exit code = 11.

Aargh.  I can confirm this problem, even on the machine I used to
build the binaries.
(Continue reading)

Permalink | Reply | 0 comments |
headers
GP lisper | 8 Jun 2005 20:20

Re: please help test the CMUCL 19b release

On Wed, 08 Jun 2005 13:54:54 +0200, <amoroso <at> mclink.it> wrote:
> Eric Marsden <eric.marsden <at> free.fr> writes:
>
>> CMUCL version 19b will be released in July. We would appreciate help
>> ...
>>    - try connecting to your X11 display:
>>
>>      ,----
>>      | CL-USER> (require :clx)
>>      | ; Loading #p"/opt/cmucl-19b-pre/lib/cmucl/lib/subsystems/clx-library.x86f".
>>      | ("CLX")
>
> ; Loading #P"/usr/local/cmucl-2005-06/lib/cmucl/lib/subsystems/clx-library.x86f".
> Warning:  XLIB previously used the following packages:
>   (#<The LISP package, 1833/3461 internal, 978/1227 external>)
> ("CLX")
> *

I note that generally you type 'OK', but didn't happen to do so here.  I get:

CL-USER> ("CLX")

; In: "CLX"

;   ("CLX")
; Error: Illegal function call.
Invoking debugger...
Starting server:
   will_fork  = False
   will_trace = False
(Continue reading)

Permalink | Reply | 3 comments |
headers
Paolo Amoroso | 8 Jun 2005 21:39
Picon
Favicon

Re: please help test the CMUCL 19b release


GP lisper <spambait <at> CloudDancer.com> writes:

> On Wed, 08 Jun 2005 13:54:54 +0200, <amoroso <at> mclink.it> wrote:
>> Eric Marsden <eric.marsden <at> free.fr> writes:
>>
>>> CMUCL version 19b will be released in July. We would appreciate help
>>> ...
>>>    - try connecting to your X11 display:
>>>
>>>      ,----
>>>      | CL-USER> (require :clx)
>>>      | ; Loading #p"/opt/cmucl-19b-pre/lib/cmucl/lib/subsystems/clx-library.x86f".
>>>      | ("CLX")
>>
>> ; Loading #P"/usr/local/cmucl-2005-06/lib/cmucl/lib/subsystems/clx-library.x86f".
>> Warning:  XLIB previously used the following packages:
>>   (#<The LISP package, 1833/3461 internal, 978/1227 external>)
>> ("CLX")
>> *
>
> I note that generally you type 'OK', but didn't happen to do so here.  I get:

In this case I forgot the OK.  Except for the warning, which I have
been noticing since pre-19a snapshots, CLX loads and works fine: I
heavily use it with McCLIM.

Paolo
--

-- 
Lisp Propulsion Laboratory log - http://www.paoloamoroso.it/log
(Continue reading)

Permalink | Reply | 2 comments |
headers
GP lisper | 8 Jun 2005 22:30

Re: please help test the CMUCL 19b release

On Wed, 08 Jun 2005 21:39:07 +0200, <amoroso <at> mclink.it> wrote:
>> On Wed, 08 Jun 2005 13:54:54 +0200, <amoroso <at> mclink.it> wrote:
>>> Eric Marsden <eric.marsden <at> free.fr> writes:
>>>
>>>> CMUCL version 19b will be released in July. We would appreciate help
>>>> ...
>>>>    - try connecting to your X11 display:
>>>>
>>>>      ,----
>>>>      | CL-USER> (require :clx)
>>>>      | ; Loading #p"/opt/cmucl-19b-pre/lib/cmucl/lib/subsystems/clx-library.x86f".
>>>>      | ("CLX")
>>>
>>> ; Loading #P"/usr/local/cmucl-2005-06/lib/cmucl/lib/subsystems/clx-library.x86f".
>>> Warning:  XLIB previously used the following packages:
>>>   (#<The LISP package, 1833/3461 internal, 978/1227 external>)
>>> ("CLX")
>>> *
>>
> In this case I forgot the OK.  Except for the warning, which I have
> been noticing since pre-19a snapshots, CLX loads and works fine: I
> heavily use it with McCLIM.

I think that warning turns out to be important.  There is a mismatch
in the libs, you uncover it later with the inspector.  I used to have
a similar problem with the debugger, until I compiled CMUCL myself.

Have you built your own CMUCL?  Just rebuild an image, but remove the
-static and -dynamic from Config.x86 first and build the extras.  The
docs work fine, except that you need to start with a simple image,
(Continue reading)

Permalink | Reply | 1 comment |
headers
Antonio Menezes Leitao | 8 Jun 2005 23:49
Picon

Strange behavior in do-http-request

Hi,

I'm sending this message both to SBCL and CMUCL mailing lists as the
problem I will describe happens in both implementations.

While using the portable allegro serve and, more particularly, the
do-http-request function to invoke a web service that returns a XML
form, I noticed that the xml parser sometimes (most of the times, in
fact) couldn't parse the result.  The exact same invocation works
perfectly in Allegro 7.0.

The reason why the xml parser fails is the fact that, most of the
times, the do-http-request call returns a string that contains one
extra character in the middle of a xml tag, thus making it invalid.

After much testing, I noticed that the problem occurs more or less
randomly.  Sometimes, it works, sometimes it doesn't (i.e., sometimes
there's no extra character, sometimes there's one extra character but,
as I mentioned previously, the non-working cases are much more
frequent).  I'm absolutely sure the server always returns the exact
same answer so the blame is on the client.

After digging on the aserve source files, I noticed that the
do-http-request function calls 

(client-request-read-sequence ans creq :start start)

in a loop to consume the server answer.  The reason why (sometimes)
there's an extra character in the do-http-request result is because
(sometimes) this function returns a sequence that contains the extra
(Continue reading)

Permalink | Reply | 2 comments |
headers
Raymond Toy | 9 Jun 2005 03:20
Picon
Favicon

Re: please help test the CMUCL 19b release


Paolo Amoroso wrote:
> Eric Marsden <eric.marsden <at> free.fr> writes:
> 
> 
>>CMUCL version 19b will be released in July. We would appreciate help
>>in testing the prerelease binaries that are available for various

Can you try the 19b-pre1-a linux binary in the experimental directory?  I 
rebuilt it on my home machine with motifd dynamically linked to motif 2.1 
(Lesstif 0.92.6).

I hope it works.

Ray
Permalink | Reply | 8 comments |
headers
Hoehle, Joerg-Cyril | 9 Jun 2005 13:49
Favicon

fwd: bug: def-alien-routine :out #return values with void return type


[Personal opinion: the extra nil in the (values nil a i) is quite unfortunate in the light of interfacing to
libraries like Pascal/Ada et.al. where most "functions" are procedures that you interface to via :out
and :in-out parameters, and void return type. cmucl/sbcl's documentation and clisp do TRT.]

-----Ursprüngliche Nachricht-----
To: 'Raymond Toy'; 'Christophe Rhodes'
Cc: 'haible <at> ilog.fr'; 'Peter Van Eynde'

Hi,

Here's a bug I consider very serious, affecting sbcl and cmucl.
a. mismatch between documentation and observable behaviour
b. regression against past behaviour.

The documentation for sbcl and for cmucl says in §8.7.3
"     (def[ine]-alien-routine "cfoo" void
       (str c-string)
       (a char :in-out)
       (i int :out))
The Lisp function cfoo will have two arguments (str and a) and two return values (a and i)."

It provides three return values in cmucl-19a-release-2004-07-28-9 and sbcl-0.8.17.4-1 (from
Ubuntu/Hoary for i386)!

W.r.t point b), I'm not sure, but my memory is that cmucl matched its documentation 10 years ago (when we used
cmucl-15b upto 17c in our project). So I'm not 100% its a regression against cmucl-17c.

Giving the above CLISP syntax, CLISP returns 2 values in this case, as cmucl documents. I'm not sure whether
clisp was modeled after cmucl back in 1995, we'd have to ask Bruno.
(Continue reading)

Permalink | Reply | 2 comments |

Gmane