Aldor crash

[pip88nl]

Hi,

today, I updated Aldor to work on Visual Studio 2008 as windows console
application. It works fine but behaves exactly like the one I built on
linux regarding crashes. Can anyone confirm that the following sequence
yields a segmentation fault:

  $ aldor -gloop
  Aldor

  Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).

  Release: Aldor(C) version 1.1.0 for LINUX(glibc-unknown)
  Type "#int help" for more details.
  %1 >> #include "aldor"
                                             Comp: 40 msec, Interp: 10 msec
  %2 >> #include "aldorio"
                                             Comp: 10 msec, Interp: 0 msec
  %3 >> import from Integer
                                             Comp: 10 msec, Interp: 0 msec
  %4 >> 3
  Program fault (segmentation violation).Program fault (segmentation violation).
  %4 >> 3

  #1 (Error) No meaning for identifier `print'.

  %5 >> 3
  Segmentation fault

To those who know the compiler internals:
 - gen0MakeApplyArgs calls gen0AbContextType in order to receive a TForm
 - gen0AbContextType calls gen0AbType for the TForm to be returned
 - gen0AbType calls abGetCategory for a TForm to be passed to
   tfDefineeType the return value of which is to be returned
 - abGetCategory sets the TForm it will return to tfUnknown
  - The function goes through several if/else conditions that might set
    the TForm variable, but none of them apply, the TForm returned is
    tfUnknown
 - after tfDefineeType processed the TForm, it is still tfUnknown
 - back up in the call tree in gen0MakeApplyArgs, the returned TForm is
   thus tfUnknown
 - in this function, the TForm is used as argument to tfMapArgc
  - after initial processing with tfFollow it calls tfFollowArg with our
    tfUnknown as first argument and 0 as second argument.
   - tfFollowArg checks whether i (which is 0) < tfArgc (tfUnknown)
     (which is also 0) and returns NULL if the condition is false. In
     this case, the condition is indeed false.
  - Now, with the NULL TForm we got from tfFollowArg, we call
    tfAsMultiArgc
   - This function calls tfFollow with the NULL TForm which behaves
     correctly, skipping any processing if tf == NULL
   - After that, the tfIsMulti macro is called on the NULL TForm and here
     is where the code breaks. The macro in turn calls tfTag on the TForm
     which is defined as (tf)->tag. It dereferences a NULL pointer...
     boom

Now, I know this is overly detailed but I thought by doing these
debugging steps myself, I could lift some work of the people who know
more about the compiler and motivate them to give it some thought. Maybe
some of you thinks "oh right, we need to do it like this...". Personally,
I have little ideas on this. I don't know what TForms really are and I
don't know how to code is supposed to behave in this case.

Although I can't imagine it would be helpful to anybody, here is the
AbSyn that was passed to gen0AbContextType, which returned tfUnknown as
an answer:
http://xinutec.org/~pippijn/files/txt/06f39ed2e3b1a89858491d955803e8d3.txt

I still have a lot to learn about the aldor internals. It is a rather
large piece of software and I must say, even though I have been fretting
about its instability, a sublime piece of art. I'm enjoying the read.

Regards,

--
Pippijn van Steenhoven


_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[Bill Page-7]

2008/6/17 Pippijn van Steenhoven wrote:
> Hi,
>
> today, I updated Aldor to work on Visual Studio 2008 as windows
> console application. It works fine but behaves exactly like the one
> I built on linux regarding crashes. Can anyone confirm that the following
> sequence yields a segmentation fault:
> ...

Yes, this error also happens for me, however using TextWriter avoids the error:

root@sage:/# aldor -gloop
Aldor

Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).

Release: Aldor(C) version 1.1.0 for LINUX(glibc2.3)
Type "#int help" for more details.
%1 >> #include "aldor"
                                           Comp: 70 msec, Interp: 0 msec
%2 >> #include "aldorio"
                                           Comp: 20 msec, Interp: 0 msec
%3 >> import from Integer
                                           Comp: 10 msec, Interp: 0 msec
%4 >> import from TextWriter
                                           Comp: 0 msec, Interp: 0 msec
%5 >> stdout << 3
3  () @ TextWriter
                                           Comp: 0 msec, Interp: 70 msec

--------

Perhaps this points at something specific to the interpreted mode?

Regards,
Bill Page.

_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[pip88nl]

On Tue, Jun 17, 2008 at 07:53:59PM -0400, Bill Page wrote:
> 2008/6/17 Pippijn van Steenhoven wrote:
> Yes, this error also happens for me, however using TextWriter avoids the error:
>
> %5 >> stdout << 3
> 3  () @ TextWriter

I am aware that it works with the TextWriter, but the point of the loop
is that you can analyse the return values of certain expressions (such as
3). I expected something like this:

%5 >> 3
3  () @ AldorInteger

I might have dreamed it, but I seem to vaguely remember that I once tried
this and it worked..

--
Pippijn van Steenhoven


_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[Bill Page-7]

2008/6/17 Pippijn van Steenhoven <pip88nl@...>:

>
> I am aware that it works with the TextWriter, but the point of the loop
> is that you can analyse the return values of certain expressions (such
> as 3). I expected something like this:
>
> %5 >> 3
> 3  () @ AldorInteger
>
> I might have dreamed it, but I seem to vaguely remember that I once tried
> this and it worked..
>

root@sage:/# aldor -gloop
Aldor

Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).

Release: Aldor(C) version 1.1.0 for LINUX(glibc2.3)
Type "#int help" for more details.
%1 >> #include "aldor"
                                           Comp: 60 msec, Interp: 10 msec
%2 >> import from AldorInteger
                                           Comp: 10 msec, Interp: 0 msec
%3 >> 3
  () @ AldorInteger
                                           Comp: 0 msec, Interp: 10 msec

???

Regards,
Bill Page

_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[Ralf Hemmecke]

On 06/18/2008 01:28 AM, Pippijn van Steenhoven wrote:

> Hi,
>
> today, I updated Aldor to work on Visual Studio 2008 as windows console
> application. It works fine but behaves exactly like the one I built on
> linux regarding crashes. Can anyone confirm that the following sequence
> yields a segmentation fault:
>
>   $ aldor -gloop
>   Aldor
>
>   Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).
>
>   Release: Aldor(C) version 1.1.0 for LINUX(glibc-unknown)
>   Type "#int help" for more details.
>   %1 >> #include "aldor"
>                                              Comp: 40 msec, Interp: 10 msec
>   %2 >> #include "aldorio"
>                                              Comp: 10 msec, Interp: 0 msec
>   %3 >> import from Integer
>                                              Comp: 10 msec, Interp: 0 msec
>   %4 >> 3
>   Program fault (segmentation violation).Program fault (segmentation violation).
>   %4 >> 3
>
>   #1 (Error) No meaning for identifier `print'.
>
>   %5 >> 3
>   Segmentation fault

You are doing the wrong thing.

 >aldor -gloop
Aldor

Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).

Release: Aldor(C) version 1.1.0 (trunk-23 AXL_EDIT_1_1_13_18=0) for
LINUX(glibc2.3)
Type "#int help" for more details.
%1 >> #include "aldor"
                                   Comp: 50 msec, Interp: 0 msec
%2 >> #include "aldorinterp"
                                   Comp: 50 msec, Interp: 0 msec
%3 >> import from Integer
                                   Comp: 20 msec, Interp: 0 msec
%4 >> 3
3 @ AldorInteger
                                   Comp: 0 msec, Interp: 140 msec
%5 >>

The sequence of your step shows a bug in Aldor since segfaults should
never happen, but the right way currently is to replace "aldorio" by
"aldorinterp".

Look at
https://svn.origo.ethz.ch/algebraist/trunk/aldor/lib/aldor/include/aldorinterp.as
https://svn.origo.ethz.ch/algebraist/trunk/aldor/lib/aldor/src/aldor_gloop.as
for more details.

Oooops, https://aquarium.aldor.csd.uwo.ca/svn/trunk is down again.

All that seems to have historical reasons, since the standard library
was axllib. And this works...

woodpecker:~/scratch>aldor -gloop
Aldor

Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).

Release: Aldor(C) version 1.1.0 (trunk-23 AXL_EDIT_1_1_13_18=0) for
LINUX(glibc2.3)
Type "#int help" for more details.
%1 >> #include "axllib"
                                   Comp: 140 msec, Interp: 10 msec
%2 >> import from Integer
                                   Comp: 30 msec, Interp: 0 msec
%3 >> 3
3 @ Integer
                                   Comp: 0 msec, Interp: 50 msec

The library libaldor does not know of a function "print". But such a
function is required by "aldor -gloop".

Ralf

> To those who know the compiler internals:
>  - gen0MakeApplyArgs calls gen0AbContextType in order to receive a TForm
>  - gen0AbContextType calls gen0AbType for the TForm to be returned
>  - gen0AbType calls abGetCategory for a TForm to be passed to
>    tfDefineeType the return value of which is to be returned
>  - abGetCategory sets the TForm it will return to tfUnknown
>   - The function goes through several if/else conditions that might set
>     the TForm variable, but none of them apply, the TForm returned is
>     tfUnknown
>  - after tfDefineeType processed the TForm, it is still tfUnknown
>  - back up in the call tree in gen0MakeApplyArgs, the returned TForm is
>    thus tfUnknown
>  - in this function, the TForm is used as argument to tfMapArgc
>   - after initial processing with tfFollow it calls tfFollowArg with our
>     tfUnknown as first argument and 0 as second argument.
>    - tfFollowArg checks whether i (which is 0) < tfArgc (tfUnknown)
>      (which is also 0) and returns NULL if the condition is false. In
>      this case, the condition is indeed false.
>   - Now, with the NULL TForm we got from tfFollowArg, we call
>     tfAsMultiArgc
>    - This function calls tfFollow with the NULL TForm which behaves
>      correctly, skipping any processing if tf == NULL
>    - After that, the tfIsMulti macro is called on the NULL TForm and here
>      is where the code breaks. The macro in turn calls tfTag on the TForm
>      which is defined as (tf)->tag. It dereferences a NULL pointer...
>      boom
>
> Now, I know this is overly detailed but I thought by doing these
> debugging steps myself, I could lift some work of the people who know
> more about the compiler and motivate them to give it some thought. Maybe
> some of you thinks "oh right, we need to do it like this...". Personally,
> I have little ideas on this. I don't know what TForms really are and I
> don't know how to code is supposed to behave in this case.
>
> Although I can't imagine it would be helpful to anybody, here is the
> AbSyn that was passed to gen0AbContextType, which returned tfUnknown as
> an answer:
> http://xinutec.org/~pippijn/files/txt/06f39ed2e3b1a89858491d955803e8d3.txt
>
> I still have a lot to learn about the aldor internals. It is a rather
> large piece of software and I must say, even though I have been fretting
> about its instability, a sublime piece of art. I'm enjoying the read.
>
> Regards,

_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[pip88nl]

On Wed, Jun 18, 2008 at 09:50:53AM +0200, Ralf Hemmecke wrote:
> The library libaldor does not know of a function "print". But such a
> function is required by "aldor -gloop".

Okay, that makes sense, but it makes me wonder why this works:

[...]
Type "#int help" for more details.
%1 >> #include "aldor"
                                           Comp: 40 msec, Interp: 10 msec
%2 >> #include "aldorio"
                                           Comp: 10 msec, Interp: 0 msec
%3 >> stdout
  () @ TextWriter
                                           Comp: 0 msec, Interp: 0 msec

Is there a function "print" for TextWriter but not for Integer?

--
Pippijn van Steenhoven


_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[pip88nl]

Oh.. never mind the last mail, the reason it works for TextWriter is that
the type is not printable. It all makes sense now. I'll look into this in
more detail later. Now that I know exactly what is wrong, I might even be
able to fix it.

--
Pippijn van Steenhoven


_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[Ralf Hemmecke]

On 06/18/2008 10:27 AM, Pippijn van Steenhoven wrote:
> Oh.. never mind the last mail, the reason it works for TextWriter is that
> the type is not printable. It all makes sense now. I'll look into this in
> more detail later. Now that I know exactly what is wrong, I might even be
> able to fix it.

That would be great. And while you are at it... also check out the bug...

aldor -gloop
Aldor

Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).

Release: Aldor(C) version 1.1.0 (trunk-23 AXL_EDIT_1_1_13_18=0) for
LINUX(glibc2.3)
Type "#int help" for more details.
%1 >> Use '#quit' to quit Axiomxl.
Segmentation fault

The segfault just appears by pressing Ctrl-C.

Ralf

_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[pip88nl]

On Wed, Jun 18, 2008 at 01:12:11PM +0200, Ralf Hemmecke wrote:

> aldor -gloop
> Aldor
>
> Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).
>
> Release: Aldor(C) version 1.1.0 (trunk-23 AXL_EDIT_1_1_13_18=0) for
> LINUX(glibc2.3)
> Type "#int help" for more details.
> %1 >> Use '#quit' to quit Axiomxl.
> Segmentation fault
>
> The segfault just appears by pressing Ctrl-C.

pippijn@osiris aldor $ aldor -gloop
Aldor

Copyright © 1990-2007 Aldor Software Organization Ltd (Aldor.org).

Release: Aldor(C) version 1.1.0 for LINUX(glibc-unknown)
Type "#int help" for more details.
%1 >> Use '#quit' to quit the interactive loop.
Use '#quit' to quit the interactive loop.
Use '#quit' to quit the interactive loop.
Use '#quit' to quit the interactive loop.
Use '#quit' to quit the interactive loop.
[...]

It doesn't happen here. Can you run it in gdb like so:

pippijn@osiris devel $ gdb --args aldor -gloop
[..header..]
(gdb) handle SIGINT nostop pass
SIGINT is used by the debugger.
Are you sure you want to change it? (y or n) y
Signal        Stop      Print   Pass to program Description
SIGINT        No        Yes     Yes             Interrupt
(gdb) r
Starting program: /home/pippijn/rona/devel/aldor/obj/bin/aldor -gloop
Aldor

Copyright © 1990-2007 Aldor Software Organization Ltd (Aldor.org).

Release: Aldor(C) version 1.1.0 for LINUX(glibc-unknown)
Type "#int help" for more details.
%1 >>
Program received signal SIGINT, Interrupt.
Use '#quit' to quit the interactive loop.

And then, when it crashes, type in:
(gdb) bt
[.. lots of output ..]
(gdb) bt full
[.. even more output ..]

--
Pippijn van Steenhoven


_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[Ralf Hemmecke]

Hi Pippijn,

On 06/18/2008 02:12 PM, Pippijn van Steenhoven wrote:

> On Wed, Jun 18, 2008 at 01:12:11PM +0200, Ralf Hemmecke wrote:
>> aldor -gloop
>> Aldor
>>
>> Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).
>>
>> Release: Aldor(C) version 1.1.0 (trunk-23 AXL_EDIT_1_1_13_18=0) for
>> LINUX(glibc2.3)
>> Type "#int help" for more details.
>> %1 >> Use '#quit' to quit Axiomxl.
>> Segmentation fault
>>
>> The segfault just appears by pressing Ctrl-C.
>
> pippijn@osiris aldor $ aldor -gloop
> Aldor
>
> Copyright © 1990-2007 Aldor Software Organization Ltd (Aldor.org).
>
> Release: Aldor(C) version 1.1.0 for LINUX(glibc-unknown)
> Type "#int help" for more details.
> %1 >> Use '#quit' to quit the interactive loop.
> Use '#quit' to quit the interactive loop.
> Use '#quit' to quit the interactive loop.
> Use '#quit' to quit the interactive loop.
> Use '#quit' to quit the interactive loop.
> [...]
>
> It doesn't happen here.

Cool... I now realize, that it doesn't happen with Aldor 1.0.3. Hmmmm....

But look for "Ctrl-C at the prompt" in

http://mlblog.osdir.com/lang.aldor/2005-06/index.shtml

> Can you run it in gdb like so:

 >gdb --args aldor -gloop
GNU gdb 6.4.90-debian
Copyright (C) 2006 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 "i486-linux-gnu"...Using host libthread_db
library "/lib/tls/i686/cmov/libthread_db.so.1".

(gdb) handle SIGINT nostop pass
SIGINT is used by the debugger.
Are you sure you want to change it? (y or n) y

Signal        Stop      Print   Pass to program Description
SIGINT        No        Yes     Yes             Interrupt
(gdb) r
Starting program: /home/hemmecke/software/aldor/linux/1.1.0/bin/aldor -gloop
Failed to read a valid object file image from memory.
Aldor

Copyright (c) 1990-2007 Aldor Software Organization Ltd (Aldor.org).

Release: Aldor(C) version 1.1.0 (trunk-23 AXL_EDIT_1_1_13_18=0) for
LINUX(glibc2.3)
Type "#int help" for more details.
%1 >>
Program received signal SIGINT, Interrupt.
Use '#quit' to quit Axiomxl.

Program received signal SIGSEGV, Segmentation fault.
0x0000000d in ?? ()
(gdb) bt
#0  0x0000000d in ?? ()
#1  0x00000002 in ?? ()
#2  0x00000033 in ?? ()
#3  0x00000000 in ?? ()
(gdb) bt full
#0  0x0000000d in ?? ()
No symbol table info available.
#1  0x00000002 in ?? ()
No symbol table info available.
#2  0x00000033 in ?? ()
No symbol table info available.
#3  0x00000000 in ?? ()
No symbol table info available.


_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[pip88nl]

On Wed, Jun 18, 2008 at 04:18:39PM +0200, Ralf Hemmecke wrote:
> Cool... I now realize, that it doesn't happen with Aldor 1.0.3. Hmmmm....

I am using vanilla Aldor from the Algebraist SVN repository and it
doesn't happen there, either (this should be 1.1.0 or something).

> But look for "Ctrl-C at the prompt" in
>
> http://mlblog.osdir.com/lang.aldor/2005-06/index.shtml

Interesting. By the way: the compiler seems to be a little bit more
stable than the interpreter, but they are mostly equal. They even produce
the exact same crashes (the ones with NULL-closures).

> Program received signal SIGSEGV, Segmentation fault.
> 0x0000000d in ?? ()
> (gdb) bt
> #0  0x0000000d in ?? ()
> #1  0x00000002 in ?? ()
> #2  0x00000033 in ?? ()
> #3  0x00000000 in ?? ()
> (gdb) bt full
> #0  0x0000000d in ?? ()
> No symbol table info available.
> #1  0x00000002 in ?? ()
> No symbol table info available.
> #2  0x00000033 in ?? ()
> No symbol table info available.
> #3  0x00000000 in ?? ()
> No symbol table info available.

Thanks for the attempt, but you seem to have a stripped, non-debug
version of Aldor. That doesn't really help :-\

--
Pippijn van Steenhoven


_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[Bill Page-7]

On Wed, Jun 18, 2008 at 11:22 AM, Ralf Hemmecke wrote:

> ...
> Anyway, should we use
>
> http://algebraist.origo.ethz.ch/issues
>
> for tracking bugs?
> Bill, what do you think? Or would
> http://axiom-wiki.newsynthesis.org/IssueTracker
> also be appropriate?
>

I guess we have too many possible tools... There is also

http://www.aldor.org/mantis/login_page.php

which was setup by the aldor.org project. I just signed up and it
seems easy. However it seems that currently there are no reports filed
at this site and I am not sure whether anyone at aldor.org is still
monitoring it. Of course it would be nice to get some feedback, but do
we really care if it is currently actively used or not?

I think that tracker at:

http://algebraist.origo.ethz.ch/node/add/issue

is more directly linked to the algebraist clone repository at the same
site - that might be a good thing if developers actually start using
the integrated development tools available at this site.

The tracker at:

http://axiom-wiki.newsynthesis.org/IssueTracker

is currently (mostly) focused on Axiom, but it could also easily be
adapted for Aldor. To the extent that Aldor bugs can be illustrated
via the Axiom-Aldor interface, this might provide some advantage.

I am willing to consider using any or all of these.

Regards,
Bill Page.

_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[Ralf Hemmecke]

> I guess we have too many possible tools... There is also

> http://www.aldor.org/mantis/login_page.php

Well, would you rely on that? Try to say

svn list https://aquarium.aldor.csd.uwo.ca/svn

I currently get

svn: PROPFIND request failed on '/svn'
svn: PROPFIND of '/svn': could not connect to server
(https://aquarium.aldor.csd.uwo.ca)

> http://algebraist.origo.ethz.ch/node/add/issue
>
> is more directly linked to the algebraist clone repository at the same
> site - that might be a good thing if developers actually start using
> the integrated development tools available at this site.

OK, then maybe we should report issues there.

Ralf

_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[Bill Page-7]

On Wed, Jun 18, 2008 at 11:58 AM, Ralf Hemmecke wrote:

>...
> Bill Page wrote:
>> http://algebraist.origo.ethz.ch/node/add/issue
>>
>> is more directly linked to the algebraist clone repository at the
>> same site - that might be a good thing if developers actually start
>> using the integrated development tools available at this site.
>
> OK, then maybe we should report issues there.
>

Ok, let's do that. Let me know if there are any technical issues about
how to start.

Regards,
Bill Page.

_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org

[Martin Rubey]

"Bill Page" <bill.page@...> writes:

> The tracker at:
>
> http://axiom-wiki.newsynthesis.org/IssueTracker
>
> is currently (mostly) focused on Axiom, but it could also easily be
> adapted for Aldor. To the extent that Aldor bugs can be illustrated
> via the Axiom-Aldor interface, this might provide some advantage.

I'd prefer this one.  I'm still hoping for a common Aldor-Axiom world.

Martin


_______________________________________________
Aldor-l mailing list
Aldor-l@...
http://aldor.org/mailman/listinfo/aldor-l_aldor.org