Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.sys.acorn.programmer > #371 > unrolled thread

Porting 'zip' - CC fatal error msg query

Back to article view | Back to comp.sys.acorn.programmer

Contents

  Porting 'zip' - CC fatal error msg query Howard Dawson <hsdawson@ormail.co.uk> - 2011-05-30 23:30 +0100
    Re: Porting 'zip' - CC fatal error msg query Steve Fryatt <news@stevefryatt.org.uk> - 2011-05-30 23:53 +0100
      Re: Porting 'zip' - CC fatal error msg query Matthew Phillips <mnews@sinenomine.freeserve.co.uk> - 2011-05-31 07:30 +0100
        Re: Porting 'zip' - CC fatal error msg query Howard Dawson <hsdawson@ormail.co.uk> - 2011-05-31 16:30 +0100
          Re: Porting 'zip' - CC fatal error msg query David Pitt <news@pittdj.co.uk> - 2011-05-31 17:22 +0100
            Re: Porting 'zip' - CC fatal error msg query cferris@freeRemoveuk.com.invalid - 2011-05-31 19:13 +0100
              Re: Porting 'zip' - CC fatal error msg query Howard Dawson <hsdawson@ormail.co.uk> - 2011-05-31 21:00 +0100
                Re: Porting 'zip' - CC fatal error msg query David Pitt <news@pittdj.co.uk> - 2011-06-01 08:17 +0100
                Re: Porting 'zip' - CC fatal error msg query "Ste (news)" <steve@revi11.plus.com> - 2011-06-02 01:04 +0100
                  Re: Porting 'zip' - CC fatal error msg query Howard Dawson <hsdawson@ormail.co.uk> - 2011-06-06 20:34 +0100
              Re: Porting 'zip' - CC fatal error msg query David Pitt <news@pittdj.co.uk> - 2011-05-31 21:05 +0100
                Re: Porting 'zip' - CC fatal error msg query cferris@freeRemoveuk.com.invalid - 2011-06-01 09:45 +0100
                  Re: Porting 'zip' - CC fatal error msg query David Pitt <news@pittdj.co.uk> - 2011-06-01 11:04 +0100
                  Re: Porting 'zip' - CC fatal error msg query Harriet Bazley <bazley@feathermail.co.uk> - 2011-06-13 22:41 +0100
    Re: Porting 'zip' - CC fatal error msg query Erik G <erikg@noname.invalid> - 2011-06-08 14:42 +0200
      Re: Porting 'zip' - CC fatal error msg query Theo Markettos <theom+news@chiark.greenend.org.uk> - 2011-06-08 17:09 +0100
      Re: Porting 'zip' - CC fatal error msg query Howard Dawson <hsdawson@ormail.co.uk> - 2011-06-08 21:15 +0100
        Re: Porting 'zip' - CC fatal error msg query "Ste (news)" <steve@revi11.plus.com> - 2011-06-08 21:58 +0100
          Re: Porting 'zip' - CC fatal error msg query Howard Dawson <hsdawson@ormail.co.uk> - 2011-06-09 15:37 +0100
      Re: Porting 'zip' - CC fatal error msg query Martin Wuerthner <spamtrap@mw-software.com> - 2011-06-10 20:07 +0200

#371 — Porting 'zip' - CC fatal error msg query

Hi everyone,

I have a 2-part question.

Firstly:
Does anyone know if the Info-zip 'zip' cmd-line routine has yet been
recompiled to be 32-bit compatible yet?
I have got a 32-bit version of 'unzip' that was done by Jeffrey Lee in the
early days of the Beagleboard, but haven't yet found any evidence of the zip
routine having been updated.

Secondly:
After a recent surprising success at my first attempt at porting, producing
a 32-bit 'BSpice' from the Unix sources, I downloaded the source from
Info-zip to have a go myself. This is more Risc OS-friendly and even
contains AMU-compatible makefiles.

One file "c.util" steadfastly refuses to compile with the following fatal
error from CC.

"c.util", line 486: Fatal internal error: ALIVE empty block L113
Internal inconsistency: etc......

Can anyone shed light on the last part of the first line of this message as
it's a new one on me. After a quick look I have some doubts about the curly
bracket matching in the offending source file but can't be 100% sure at this
stage.

TIA
Howard.


-- 
-------------
Howard Dawson
 
hsdawson@ormail.co.uk

http://hd-audio.orpheusweb.co.uk/
---------------------------------

[toc] | [next] | [standalone]

#372

On 31 May, Howard Dawson wrote in message
    <na.1eb3bf51db.a80520hsdawson@ormail.co.uk>:

> Firstly: Does anyone know if the Info-zip 'zip' cmd-line routine has yet
> been recompiled to be 32-bit compatible yet? I have got a 32-bit version
> of 'unzip' that was done by Jeffrey Lee in the early days of the
> Beagleboard, but haven't yet found any evidence of the zip routine having
> been updated.

Presumably, since Harriet Bazley's InfoZip frontend has been using them both
on at least one Iyonix for many years.

  http://www.starfighter.acornarcade.com/mysite/utilities.htm

-- 
Steve Fryatt - Leeds, England

http://www.stevefryatt.org.uk/

[toc] | [prev] | [next] | [standalone]

#375

In message <mpro.lm169h0jwwwdc01j3.news@stevefryatt.org.uk>
 on 30 May 2011 Steve Fryatt  wrote:

> On 31 May, Howard Dawson wrote in message
>     <na.1eb3bf51db.a80520hsdawson@ormail.co.uk>:
> 
> > Firstly: Does anyone know if the Info-zip 'zip' cmd-line routine has yet
> > been recompiled to be 32-bit compatible yet? I have got a 32-bit version
> > of 'unzip' that was done by Jeffrey Lee in the early days of the
> > Beagleboard, but haven't yet found any evidence of the zip routine having
> > been updated.
> 
> Presumably, since Harriet Bazley's InfoZip frontend has been using them
> both on at least one Iyonix for many years.
> 
>   http://www.starfighter.acornarcade.com/mysite/utilities.htm

Are you wanting a Beagleboard-compatible binary, Howard, rather than just
32-bit?  (The work that Jeffrey Lee did was for Beagleboard: as Steve says
there's been a 32-bit version for ages.)

I have a copy of a "zip" which I believe is meant to be Beagleboard
compatible which was sent to me by Michael Gerbracht when I was doing some
testing of Luafox on the Beagleboard.  I have not used it with Harriet's
front end, or on the command line, but only as part of whatever Luafox uses
it for.

I'm not sure whether it has been put on general distribution and I cannot
remember where Michael got it from.

-- 
Matthew Phillips
Durham

[toc] | [prev] | [next] | [standalone]

#377

In article <49c0dfdb51.Matthew@sinenomine.freeserve.co.uk>, Matthew Phillips
<mnews@sinenomine.freeserve.co.uk> wrote:
> Are you wanting a Beagleboard-compatible binary, Howard, rather than just
> 32-bit?  (The work that Jeffrey Lee did was for Beagleboard: as Steve says
> there's been a 32-bit version for ages.)
> 
> I have a copy of a "zip" which I believe is meant to be Beagleboard
> compatible which was sent to me by Michael Gerbracht when I was doing some
> testing of Luafox on the Beagleboard.  I have not used it with Harriet's
> front end, or on the command line, but only as part of whatever Luafox
> uses
> it for.
> 
> I'm not sure whether it has been put on general distribution and I cannot
> remember where Michael got it from.
> 
> -

Sorry, yes it's for BeagleBoard, I should have made that clearer.

Howard


-- 
-------------
Howard Dawson
 
hsdawson@ormail.co.uk

http://hd-audio.orpheusweb.co.uk/
---------------------------------

[toc] | [prev] | [next] | [standalone]

#378

Howard Dawson, on 31 May, wrote:

> In article <49c0dfdb51.Matthew@sinenomine.freeserve.co.uk>, Matthew
> Phillips <mnews@sinenomine.freeserve.co.uk> wrote:
> > Are you wanting a Beagleboard-compatible binary, Howard, rather than
> > just 32-bit?  (The work that Jeffrey Lee did was for Beagleboard: as
> > Steve says there's been a 32-bit version for ages.)
> > 
> > I have a copy of a "zip" which I believe is meant to be Beagleboard
> > compatible which was sent to me by Michael Gerbracht when I was doing
> > some testing of Luafox on the Beagleboard.  I have not used it with
> > Harriet's front end, or on the command line, but only as part of
> > whatever Luafox uses it for.
> > 
> > I'm not sure whether it has been put on general distribution and I
> > cannot remember where Michael got it from.
> > 
> > -
> 
> Sorry, yes it's for BeagleBoard, I should have made that clearer.

Is this any good :-

http://www.riscos.info/packages/UtilitiesDetails.html#zip

In the most superficial of tests, with InfoZip, it worked here on an ARMini
but exceptions must be off.


-- 
David Pitt

[toc] | [prev] | [next] | [standalone]

#380

In message <mpro.lm2it100000wn01w0.news@pittdj.co.uk>
          David Pitt <news@pittdj.co.uk> wrote:

> Howard Dawson, on 31 May, wrote:
> 
> > In article <49c0dfdb51.Matthew@sinenomine.freeserve.co.uk>, Matthew
> > Phillips <mnews@sinenomine.freeserve.co.uk> wrote:
[snip]

> > Sorry, yes it's for BeagleBoard, I should have made that clearer.
> 
> Is this any good :-
> 
> http://www.riscos.info/packages/UtilitiesDetails.html#zip
> 
> In the most superficial of tests, with InfoZip, it worked here on an
> ARMini but exceptions must be off.
> 
I think Howard wanted to find out what was up with the code he was
trying to compile.

With ref to your version of '!InfoZip' what is the error address given
when the exceptions are on?

-- 
Colin Ferris Cornwall UK

[toc] | [prev] | [next] | [standalone]

#381

In article <9a1c20dc51.cferris@cferris.freeuk.com>,
cferris@freeRemoveuk.com.invalid wrote:
> I think Howard wanted to find out what was up with the code he was
> trying to compile.
> 
> 

Thanks chaps, exceptions off is not a big issue. I'll have a look at that.

I'd still be curious about the meaning of my fatal error msg from CC though,
as detailed in my original posting as I'd just like to understand the
problem with the source file in question.
Emboldened by the Spice success, I'd still like to try and compile it on the
Beagle if only for the experience and learning it would provide.

Cheers
Howard.


-- 
-------------
Howard Dawson
 
hsdawson@ormail.co.uk

http://hd-audio.orpheusweb.co.uk/
---------------------------------

[toc] | [prev] | [next] | [standalone]

#385

Howard Dawson, on 31 May, wrote:

[snip]
 
> I'd still be curious about the meaning of my fatal error msg from CC
> though, as detailed in my original posting as I'd just like to understand
> the problem with the source file in question. 

Some you win some you lose. I have managed to build zip30, using the 'acorn'
settings, on the ARMini with cc vsn 5.69, without the cryptic fatal error
reported at the start of this thread. The downside, if one could call it
that, was that the resulting zip absolute crashes on the ARMini, with either
exceptions on or off. The same absolute does run on the Iyonix and can get
as far as reporting its version, I have not tested further.  

> Emboldened by the Spice
> success, I'd still like to try and compile it on the Beagle if only for
> the experience and learning it would provide.

Likewise, I have found a few old faithfuls in need of a rebuild, it would be
useful to know how to. 


-- 
David Pitt

[toc] | [prev] | [next] | [standalone]

#388

In article <na.9f29e451dc.a80520hsdawson@ormail.co.uk>,
   Howard Dawson <hsdawson@ormail.co.uk> wrote:
> Thanks chaps, exceptions off is not a big issue.

Are you sure about that? If it is provoking exceptions when they are
enabled, that means the CPU is executing undefined instructions when those
exceptions are disabled. Given that this is (un)packing zip archives, the
results could be subtly but fatally broken.

Just a heads-up; some people might not be aware of the true meaning of those
pesky exceptions and might assume that disabling them means the problem is
fixed :) But you may already know all that.

Ta,

Steve

-- 
Steve Revill @ Home
Note: All opinions expressed herein are my own.

[toc] | [prev] | [next] | [standalone]

#401

In article <51dcc40698steve@revi11.plus.com>, "Ste (news)"
<steve@revi11.plus.com> wrote:
> Are you sure about that? If it is provoking exceptions when they are
> enabled, that means the CPU is executing undefined instructions when those
> exceptions are disabled. Given that this is (un)packing zip archives, the
> results could be subtly but fatally broken.
> 
> Just a heads-up; some people might not be aware of the true meaning of
> those
> pesky exceptions and might assume that disabling them means the problem is
> fixed :) But you may already know all that.
> 
> Ta,
> 
> Steve
> 
> 

Thanks point taken, all the more reason to pursue a proper Beagle-friendly
port in the longer term.

For now at least the version from David's earlier link seems to do the job
with exceptions off and no problems encountered so far using it with my
!Zipper front end which is very similar to Harriet's except for being
written in C.

Thanks again for all the help.

Howard.




-- 
-------------
Howard Dawson
 
hsdawson@ormail.co.uk

http://hd-audio.orpheusweb.co.uk/
---------------------------------

[toc] | [prev] | [next] | [standalone]

#382

cferris@freeRemoveuk.com.invalid, on 31 May, wrote:

[snip]

> With ref to your version of '!InfoZip' what is the error address given
> when the exceptions are on?
> 
&24E50004

-- 
David Pitt

[toc] | [prev] | [next] | [standalone]

#386

In message <mpro.lm2t5w00000aj028k.news@pittdj.co.uk>
          David Pitt <news@pittdj.co.uk> wrote:

> cferris@freeRemoveuk.com.invalid, on 31 May, wrote:
> 
> [snip]
> 
> > With ref to your version of '!InfoZip' what is the error address
> > given when the exceptions are on?
> > 
> &24E50004
> 

Thanks for that - but as you know - !InfoZip consists of a BASIC
!Runimage - loading a code file - plus a Module.

Which calls a 'Absolute' file which does the main work.

Does the BASIC bit load ok on the Icon Bar?

If that is ok - Do you know which of the 'Absolute' files is called?

Does '*Where' & '*ShowRegs' give any clues about the error?


Thanks
-- 
Colin Ferris Cornwall UK

[toc] | [prev] | [next] | [standalone]

#387

cferris@freeRemoveuk.com.invalid, on 1 Jun, wrote:

> In message <mpro.lm2t5w00000aj028k.news@pittdj.co.uk>
>           David Pitt <news@pittdj.co.uk> wrote:
> 
> > cferris@freeRemoveuk.com.invalid, on 31 May, wrote:
> > 
> > [snip]
> > 
> > > With ref to your version of '!InfoZip' what is the error address given
> > > when the exceptions are on?
> > > 
> > &24E50004
> > 
> 
> Thanks for that - but as you know - !InfoZip consists of a BASIC !Runimage
> - loading a code file - plus a Module.

You are right, I do know.
> 
> Which calls a 'Absolute' file which does the main work.
> 
> Does the BASIC bit load ok on the Icon Bar?
> 
> If that is ok - Do you know which of the 'Absolute' files is called?
> 
> Does '*Where' & '*ShowRegs' give any clues about the error?

Currently with zip 3.0, compiled with gcc 4.1.1, !InfoZip stalls on zipping 
with a segmentation fault recorded in the taskwindow. !InfoZip itself has
not reported an error and is still on the iconbar.

*where
Address 00021270 is not in any module
*showregs
Register dump (stored at &2000E8AC) is:
R0  = 0000000B R1  = 00000080 R2  = 0006D31D R3  = 0006D313
R4  = 0007D310 R5  = 00000004 R6  = 00000000 R7  = 0006D41F
R8  = 0000003E R9  = 0000000A R10 = 0007D558 R11 = 0005D2FC
R12 = 0000000B R13 = 0007E048 R14 = 0006D308 R15 = 00021270
Mode USR32 flags set: nzCvqjggggeAift        PSR = 20000110
*

I get the same error calling zip from the command line so that is !InfoZip
off the hook.

I am going to have to come back to this, when I am better organised on this
ARMini. The latest hazard is StrongED SE4.69a4 crashing, the good news is
that DeskEdit4 4.01/32 is looking hopeful. 


-- 
David Pitt

[toc] | [prev] | [next] | [standalone]

#427

On 1 Jun 2011 as I do recall,
          cferris@freeRemoveuk.com.invalid wrote:

> In message <mpro.lm2t5w00000aj028k.news@pittdj.co.uk>
>           David Pitt <news@pittdj.co.uk> wrote:
>
> > cferris@freeRemoveuk.com.invalid, on 31 May, wrote:
> >
> > [snip]
> >
> > > With ref to your version of '!InfoZip' what is the error address
> > > given when the exceptions are on?
> > >
> > &24E50004
> >
>
> Thanks for that - but as you know - !InfoZip consists of a BASIC
> !Runimage - loading a code file - plus a Module.
>
> Which calls a 'Absolute' file which does the main work.
>
The Module is for heap management within the Wimpslot; the code file is
simply a bit of assembler to speed up BASIC's string handling.

The only part of the program which interacts with the Absolute file is
the BASIC program, and all that does is issue command-line strings:  I
won't guarantee that !Infozip can't cause memory exceptions on its own
due to sloppy assembler, but at least those bits don't have anything to
do with calling compiled C....

-- 
Harriet Bazley                     ==  Loyaulte me lie ==

Mate, this parrot wouldn't VOOM if you put four million volts through it!

[toc] | [prev] | [next] | [standalone]

#402

On 31-5-2011 0:30, Howard Dawson wrote:
> After a recent surprising success at my first attempt at porting, producing
> a 32-bit 'BSpice' from the Unix sources, I downloaded the source from
> Info-zip to have a go myself. This is more Risc OS-friendly and even
> contains AMU-compatible makefiles.
> 
> One file "c.util" steadfastly refuses to compile with the following fatal
> error from CC.
> 
> "c.util", line 486: Fatal internal error: ALIVE empty block L113
> Internal inconsistency: etc......
> 
> Can anyone shed light on the last part of the first line of this message as
> it's a new one on me. After a quick look I have some doubts about the curly
> bracket matching in the offending source file but can't be 100% sure at this
> stage.

This is an internal compiler error. That means that there is a bug
somewhere in the compiler itself. While building and processing an
internal representation of the source file, a sanity check inside the
compiler found that one part of the internal representation was
inconsistent with another part. So the compiler gives up, throwing out a
final error message and quitting.

This type of error should not happen, even if the source file contains
horrible (or maybe very subtle) C errors. The problem does not lie in
the c.util source, but in the compiler.

The eventual solution to this problem has to be worked out by the makers
of the compiler. I suggest you send the error message report and
information on where to find the source file (and the header files that
are included) to them. There is probably a proper channel for such bug
reports, such as an email address or a bug tracking system. The
documentation of the compiler may include instructions on how to report
the problem.

In the meantime your options are limited.

- If you are very lucky, the problem was already found and solved by the
makers of the compiler. See if there is a newer version of the compiler
available, and if this still contains the problem.

- You could try to use a different compiler. This is usually impractical.

- You could try to work around the problem by tweaking the c.util source
file. By changing the source so that it essentially does the same but
with slightly different coding, you may get lucky and avoid the bug in
the compiler.

-- 
Erik G.
From address is fake
See http://www.xs4all.nl/~erikgrnh/

[toc] | [prev] | [next] | [standalone]

#403

Erik G <erikg@noname.invalid> wrote:
> - You could try to use a different compiler. This is usually impractical.

FWIW GCC should be able to compile this just fine.  In fact, GCCSDK builds
zip as a matter of course, so its ability to build zip is regularly tested. 
The version of GCC that runs on RISC OS shouldn't have any trouble either.

Theo

[toc] | [prev] | [next] | [standalone]

#406

In article <4def6e1b$0$11158$e4fe514c@dreader31.news.xs4all.nl>, Erik G
<erikg@noname.invalid> wrote:
> This type of error should not happen, even if the source file contains
> horrible (or maybe very subtle) C errors. The problem does not lie in
> the c.util source, but in the compiler.
> 
> 

OK thanks Erik, I am using an older version of CC so that explains that one 
nicely. A previous post suggests it compiles OK on the latest version so no
need to go any further - must upgrade soon (if only I could lay my hands on
the old Castle CD which seems to have gone AWOL in my workshop somewhere).

I hadn't got round to trying with GCC yet, but imagined it would be OK from
the zip documentation.

Cheers
Howard.



-- 
-------------
Howard Dawson
 
hsdawson@ormail.co.uk

http://hd-audio.orpheusweb.co.uk/
---------------------------------

[toc] | [prev] | [next] | [standalone]

#407

In article <na.e049f651e0.a80520hsdawson@ormail.co.uk>,
   Howard Dawson <hsdawson@ormail.co.uk> wrote:
> OK thanks Erik, I am using an older version of CC so that explains that
> one nicely. A previous post suggests it compiles OK on the latest version
> so no need to go any further - must upgrade soon (if only I could lay my
> hands on the old Castle CD which seems to have gone AWOL in my workshop
> somewhere).

The old CD might not make any difference to you - we're only offering
discounts to people who originally bought their tools CDs from ROOL (unless
you have one of the last batch that Castle made, which had a key starting
with DEV07...).

http://www.riscosopen.org/content/sales

We're soon going to discontinue discounts for _any_ CD not bought from ROOL,
seeing as we've been selling the tools for two years now - anyone who hasn't
upgraded by now will have missed the boat on that one - and we're doing a
lot of work on the toolchain, so ROOL at least should benefit if any of the
new features are enough to push someone with an older copy to upgrade.

Thanks,

Steve

-- 
Steve Revill @ Home
Note: All opinions expressed herein are my own.

[toc] | [prev] | [next] | [standalone]

#408

In article <51e04dd6c4steve@revi11.plus.com>, "Ste (news)"
<steve@revi11.plus.com> wrote:
> The old CD might not make any difference to you - we're only offering
> discounts to people who originally bought their tools CDs from ROOL
> (unless
> you have one of the last batch that Castle made, which had a key starting
> with DEV07...).
> 
> 
Ah OK, thanks Steve

My searching is in vain then as I'm fairly sure the key started with DEV04
(circa 2003 IIRC). No doubt it will turn up now!

Cheers
Howard.


-- 
-------------
Howard Dawson
 
hsdawson@ormail.co.uk

http://hd-audio.orpheusweb.co.uk/
---------------------------------

[toc] | [prev] | [next] | [standalone]

#414

In message <4def6e1b$0$11158$e4fe514c@dreader31.news.xs4all.nl>
          Erik G <erikg@noname.invalid> wrote:

> This is an internal compiler error. That means that there is a bug
> somewhere in the compiler itself. [...]

> The eventual solution to this problem has to be worked out by the makers
> of the compiler. I suggest you send the error message report and
> information on where to find the source file (and the header files that
> are included) to them.

For C compilers, it is common practice to send preprocessed sources 
(output of cc -E) when reporting bugs. That avoids having to mess 
about with header files, include paths, different versions of system 
header files, etc. It is easy enough to check whether the error still 
occurs when compiling the preprocessed source file (which is all the 
compiler proper sees anyway).

-- 
Martin
---------------------------------------------------------------------
Martin Wuerthner         MW Software      http://www.mw-software.com/
        RISC OS Software for Design, Printing and Publishing
---------------------------------------------------------------------

[toc] | [prev] | [standalone]

Back to top | Article view | comp.sys.acorn.programmer

csiph-web