oslib-devel — discussions about oslib development/bugs/ideas, and so on...

[Oslib-devel] diskdemo int86x offset don't thrust int86x

[Paul D. <du...@ho...>]

Well, Luca suggested that the registers could be passed wrong.
First in diskdemo.c, I believe the offset of the dos block allocated is 
passed in bx,
and it should be di (es:di). Found it by a test in debug... I had not the 
doc under
hands.
But I don't think it is the problem. Since I use __BIOS_DEBUG__ to see the 
registers,
and bx and di were 0, so it is ok. :-) And also changing this did not fix 
the problem.

On the other hand, I DONT THRUST int86x, I remember having found a bug with
segment register passing of Borland int86x, in 1992 or 1993. And this 
version I (we) use
is of 1991, so I expect it to be buged. I even think it was Exactly with 
this
int 13h es:di I had found it by the time.

_________________________________________________________________
MSN Search, le moteur de recherche qui pense comme vous !  
http://fr.ca.search.msn.com/

Re: [Oslib-devel] With X_call only floppy access is screw up

[Luca A. <luc...@em...>]

On Fri, 2003-07-11 at 09:46, Paul Dufresne wrote:
> Ok, I dedided to try also with grub rather than x version 5.2.
> 
> So I keep x_calls to read parameter, and to read the floppy.
> 
> But surprised, on grub it is unable to go upto read the parameters, as under 
> x
> it is able (I restested with the same diskdemo.xtn).
> 
> Here what happenned under grub:
> c on grub
> kernel=(hd0,5)/oslib/oslib/examples/diskdemo.xtn
> boot
> 
> the result:
> ...
> features #3: 0x383f9ff
> Starting CX=F8
> PM reentry linear address = 0x4800
> Exc 13 occured
> General Protection Fault
> DS:30 CS:38
> Actual stack: 328694
> TR F8
> SS:SP AF25:85A5A5A5
> Stack 0: 26a5 a5b5ad85
> Stack 1: a5a5 85ac85a7
> stack 2: a585 a5a1a5a5
> 
> cs:aca5 ds:25a5
> 
> Desc [F8] InfoBase: 32c6e4 Lim: d8 acc: 8b Gran: 40
> 
> Abort5!!! Halt called
Paul, 

X_callBIOS returns to real mode to call the BIOS. For doing this, it
uses a call gate that is created by the x.exe extender, but not by GRUB.
Hence, X_callBIOS() cannot be used with grub, but only with x.exe.

The fact that a BIOS call fails even with X_callBIOS() means that we are
probably calling INT 0x13 improperly... I suspect that not all the
registers are set up in the proper way, or that some othe BIOS call has
to be invoked first... In the weekend, I'll have a look to the RBIL, to
see if I discover something useful.

BTW, I suggest you not to do experiments with the hd, if you have
important data on it... ;-)

Hence, it seems that the vm86_callBIOS() mechanism and the reflection
code are innocent, right?

Regarding my today's cvs commit... I realize that the anonymous cvs is
updated with a 24 hours delay, hence you will see it tomorrow :(
As soon as I will find the time, I'll setup a cvs log mailing list, so
that the cvs commits are immediately visible.

			Luca

-- 
_____________________________________________________________________________
 Copy this in your signature, if you think it is important:
                                N O    W A R ! ! !



--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
«Io non ci credevo. E invece l'amore vero esiste. Grazie Meetic.it!»
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=1572&d=11-7

Re: [Oslib-devel] With X_call only floppy access is screw up

[Paul D. <du...@ho...>]

Ok, I dedided to try also with grub rather than x version 5.2.

So I keep x_calls to read parameter, and to read the floppy.

But surprised, on grub it is unable to go upto read the parameters, as under 
x
it is able (I restested with the same diskdemo.xtn).

Here what happenned under grub:
c on grub
kernel=(hd0,5)/oslib/oslib/examples/diskdemo.xtn
boot

the result:
...
features #3: 0x383f9ff
Starting CX=F8
PM reentry linear address = 0x4800
Exc 13 occured
General Protection Fault
DS:30 CS:38
Actual stack: 328694
TR F8
SS:SP AF25:85A5A5A5
Stack 0: 26a5 a5b5ad85
Stack 1: a5a5 85ac85a7
stack 2: a585 a5a1a5a5

cs:aca5 ds:25a5

Desc [F8] InfoBase: 32c6e4 Lim: d8 acc: 8b Gran: 40

Abort5!!! Halt called

_________________________________________________________________
MSN Search, le moteur de recherche qui pense comme vous !  
http://fr.ca.search.msn.com/

[Oslib-devel] With X_call only floppy access is screw up

[Paul D. <du...@ho...>]

Ok, I came back to using x_call.
And I removed the first call to reset the floppy controller, keeping only 
get disk parameters
call.

It worked! : Drives=1 Type=0x4 Cyl: 80 Sect: 18

But if I tried to continue with the call to read a sector...
Floppy motor turns, but it never returned.

So I tried to read on hard disk, dl=0x80, it works!
Meaningless stuff read, but it works.

Here my hypothesis:
Reset the controller have to do IO on the floppy, so it hangs.
Read disk parameter, only read the info somewhere in RAM, it works!
Reading floppy disk works, but not hard disk:
Either because the floppy disk have to setup DMA, not hard disk.
Or IO permissions for hard disk is Ok, not for floppy.

_________________________________________________________________
MSN Messenger : discutez en direct avec vos amis !  
http://messenger.fr.msn.ca/

[Oslib-devel] diskdemo results (with DEBUG)

[Paul D. <du...@ho...>]

First, Luca said that on cvs vm86.c had changed.
I did not seems to be the case, so I manually removed the "offending" lines.

Replacing vm86_ to x_ calls, have made the floppy disk motor to spin. (At 
least first time
I tried). But then hangs there, without ever writing reset finished, so it 
must be
on the reset call.

I have made some changes, to remove warnings, and also to  comment out
the reading of a sector, keeping only the call to reset disk, and read disk
parameters.

Don't apply this blindly, just to show you! :
H:\oslib\oslib\examples>cvs -z3 diff diskdemo.c
Index: diskdemo.c
===================================================================
RCS file: /cvsroot/oslib/oslib/examples/diskdemo.c,v
retrieving revision 1.1
diff -r1.1 diskdemo.c
39a40
>#include <ll/i386/x-dosmem.h>
40a42
>#include <ll/ll.h>
50c52
<     ctx = context_save();
---
>     ctx = ll_context_save();
70c72
<     ctx = context_save();
---
>     ctx = ll_context_save();
95,96c97
<          message("Error: Trying to reflect 0x%lx (%d)!!!\n",
<                          intnum, intnum);
---
>            message("Error: Trying to reflect (%ld)!!!\n",intnum);
124c125
<     ir.h.dl = 0;
---
>     ir.h.dl = 0x00;
132c133
<     ir.h.dl = 0;
---
>     ir.h.dl = 0x00;
147,148c148,149
<
< #if 0
---
>/*
>if 0
150c151
<     ir.h.al = 8;       /* Let's verify 8 sectors... */
---
>     ir.h.al = 8;        Let's verify 8 sectors...
152,155c153,156
<     ir.h.ch = 0;       /* Cylonder 0 */
<     ir.h.cl = 1;       /* Sector 0 */
<     ir.h.dh = 0;       /* Head 0 */
< #else
---
>     ir.h.ch = 0;        Cylonder 0
>     ir.h.cl = 1;        Sector 0
>     ir.h.dh = 0;        Head 0
>else
160,163c161,164
<     ir.h.ch = 0; /* C */
<     ir.h.cl = 1; /* S */
<     ir.h.dh = 0; /* H */
<     ir.h.dl = 0;
---
>     ir.h.ch = 0;  C
>     ir.h.cl = 1;  S
>     ir.h.dh = 0;  H
>     ir.h.dl = 0x00;
166c167
< #endif
---
>endif
169c170
<     vm86_callBIOS(0x13, &ir, &or, &sr);
---
>     X_callBIOS(0x13, &ir, &or, &sr);
180a182
>*/

H:\oslib\oslib\examples>

Here what I got with this:

Stuff scrolled up...
Executing code: CD 48
reset finished.
Stack: 209D0 ss:1F00 SP:13CA
CS:70 IP:3EE
Switching to F8
Real-Mode Address is CS=2000 IP=5D0
Linear=205D0
Stack Frame: 69 328A2F
5FF000FE ...
...
...
I am back...
TSS CS=38 IP=3219E0
PM reentry linear address=   There was nothing there!
Executing code: CD 48
Drives:0
Drive Type: 0x32
Cylinders: 518
Sectors: 41
... End reached

I tried to use X_calls with DL=0x80 (hard disk rather than floppy), but this 
didnt helped.
Hanging before writing reset finished.

_________________________________________________________________
MSN Messenger : discutez en direct avec vos amis !  
http://messenger.fr.msn.ca/

Re: [Oslib-devel] trying the examples

[Paolo G. <pao...@ti...>]

> I am a little bit surprised by the fact that the 5.2 binary downloaded
> from sf works well here (both on my computer and in bochs), but maybe
I
> am just lucky ;-)

me too... it does happen with some new machines... basically it runs on
many PC here at the ReTiS, but only with simple shark examples... when
there are some more complex demos ... boom!!!

In Pavia, where a lot of PCs are newer than here, it happens almost
always :-(

bye

PJ

-- 
-----------------------------------------------------------------------
Paolo Gai - ReTiS Lab - PhD Student

Scuola Superiore S. Anna        Tel       : +39 050 883 451 
Polo S. Anna Valdera            Fax       : +39 050 883 452
viale Rinaldo Piaggio 34        e-mail    : pj...@ga...
56025 - Pontedera (PI) - ITALY  home page : http://feanor.sssup.it/~pj
------------------------------------------------------------------------
Per favore non mandatemi allegati in Word o PowerPoint.
(Please avoid sending me Word or PowerPoint attachments.)
Si veda (See) http://www.fsf.org/philosophy/no-word-attachments.html
------------------------------------------------------------------------

Re: [Oslib-devel] trying the examples

[Luca A. <luc...@em...>]

Hi all,

On Fri, 2003-07-11 at 07:09, Paolo Gai wrote:
> Hi!
> 
> We had a similar problem (the same?) with the compiled version of X 5.2
> that Luca put on the web a few months ago... we tried for a few weeks
> both in Pisa and in Pavia to guess why it didn't work, and at the end we
> recompiled the CVS source and it worked!... 
> 
> We sent our compiled version to Luca, anyway I do not know really if he
> replaced the buggy one which was on the sourceforge web site...
Ok, I'll put the new binary on the sf download page...
I am a little bit surprised by the fact that the 5.2 binary downloaded
from sf works well here (both on my computer and in bochs), but maybe I
am just lucky ;-)

				Luca
-- 
_____________________________________________________________________________
Copy this in your signature, if you think it is important:
                               N O    W A R ! ! !



--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Tariffe promozionali per viaggiare in tutto il mondo, clicca e scopri dove prenotare
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=1258&d=11-7

Re: [Oslib-devel] In diskdemo, motor don't turn

[Luca A. <luc...@em...>]

Hi Paul,

> Just to make clear, that I checked, and in the diskdemo,
> the floppy doesn't seems to be really accessed.
> I means, the light don't comes on, it goes too fast,
> and I don't hear the motor turn.
This is not very surprising... As seen in the previous mail, all the
return values are completely screwed up. Hence, there is something that
is really not working properly.

Can you try to change vm86_callBIOS() in X_callBIOS() in diskdemo.c and
see if it behaves better?

			Thanks,
				Luca
-- 
_____________________________________________________________________________
Copy this in your signature, if you think it is important:
                               N O    W A R ! ! !



--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Vuoi un notebook potente, funzionale ideale per il lavoro?
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=763&d=11-7

Re: [Oslib-devel] the fact of X 5.1 and 5.2 don't works bothers me

[Luca A. <luc...@em...>]

On Thu, 2003-07-10 at 20:52, Paul Dufresne wrote:
> I downloaded the sources of X.
> He, this is quite old, 1996!
Are you sure? The 5.2 version should be less than 1 year old...


				Luca
-- 
_____________________________________________________________________________
Copy this in your signature, if you think it is important:
                               N O    W A R ! ! !



--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Richiedi il tuo artigiano di fiducia su QxService! Potrai vincere il nuovo Nec e606 UMTS e ricariche telefoniche! Clicca qui!
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=1413&d=11-7

Re: [Oslib-devel] diskdemo results (without LL_DEBUG)

[Luca A. <luc...@em...>]

Hi Paul,

I am a little bit confused by all these messages that I am receiving
from the ml, anyway... ;-)

readdemo uses DOS_*, that in turn use X_callBIOS()... This function
switches back to RM to call the bios functions.
On the other hand, diskdemo uses vm86_callBIOS(), that tries to use the
vm86 mode instead or switching back to RM.

Hence, we can say that there is a problem in the vm86 code...

[...]
> reset finished.
> Drives: 0
It finds 0 drives??? I'd say that we are in big trouble ;-)

> Drive type: 0x32
The drive type is also looking very strange...

> Cylingers: 519
> Sectors: 9
...and this is definitly one of the most funny floppy geometries that
I've ever seen ;-)

I am suspecting that there is some bug in the code that copies back the
return values from the cpu registers... It is working well for me, but
maybe it is just a case ;-)
I'll try to have a better look at vm86_return() in xlib/vm86.c. In the
meanwhile, you can try to enable debugging and to see if the returned
values make any sense...
Also, I am not sure if the bios services are called in the proper way (I
will have to check if the buffers are properly aligned, if all the
registers are filled with the correct values, and so on...).

Finally, can you try to change vm86_callBIOS() with X_callBIOS() in
diskdemo.c, and to see the result? Just to verify that the problem
actually is in vm86_callBIOS().

			Thanks,
				Luca
-- 
_____________________________________________________________________________
Copy this in your signature, if you think it is important:
                               N O    W A R ! ! !



--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Con KLM puoi risparmiare fino a 20 Euro sul tuo biglietto aereo prenotando on line per Usa, Europa e il resto del mondo
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=1259&d=11-7

Re: [Oslib-devel] LL DEBUG option triggers a bug

[Luca A. <luc...@em...>]

Hi Paul,

On Thu, 2003-07-10 at 16:04, Paul Dufresne wrote:
> In xlib:
> vm86.c:338: #error Fix the following: use global_regs->xxx ???
Should be fixed in cvs now... Can you try to cvs update and let me know
if it compiles? If it does not, simply comment out that part of code
(from my new #if 0 to #endif).

			Thanks,
				Luca
-- 
_____________________________________________________________________________
Copy this in your signature, if you think it is important:
                               N O    W A R ! ! !



--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Su eBay compri e vendi qualsiasi oggetto al prezzo che decidi tu
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=1647&d=11-7

Re: [Oslib-devel] trying the examples

[Paolo G. <pao...@ti...>]

Hi!

We had a similar problem (the same?) with the compiled version of X 5.2
that Luca put on the web a few months ago... we tried for a few weeks
both in Pisa and in Pavia to guess why it didn't work, and at the end we
recompiled the CVS source and it worked!... 

We sent our compiled version to Luca, anyway I do not know really if he
replaced the buggy one which was on the sourceforge web site...

anyway, now it works, and I'm happy you can play with it!

byebye

PJ

On Fri, 2003-07-11 at 04:26, Paul Dufresne wrote:
> I am playing with my rebuild X.
> It's fun because for most examples, it works well.
> In fact, I would say that only diskdemo don't work.
> Even readdemo works... I rebooted the computer, and it did print what I had
> put in my test file!
> 
> Some examples, are not able to return correctly to DOS.
> DOS complains that the drive is invalid.
> 
> So...
> So it seems it is mostly a problem with either int13h or maybe more 
> precisely,
> the floppy disk.
> 
> I'll try to figure out which is which tomorrow.
> 
> But I am surprised, I guess readdemo calls int21h in vm86 mode, and that
> int21h call int13h !
> 
> _________________________________________________________________
> MSN Search, le moteur de recherche qui pense comme vous !  
> http://fr.ca.search.msn.com/
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email sponsored by: Parasoft
> Error proof Web apps, automate testing & more.
> Download & eval WebKing and get a free book.
> www.parasoft.com/bulletproofapps1
> _______________________________________________
> Oslib-devel mailing list
> Osl...@li...
> https://lists.sourceforge.net/lists/listinfo/oslib-devel
-- 
-----------------------------------------------------------------------
Paolo Gai - ReTiS Lab - PhD Student

Scuola Superiore S. Anna        Tel       : +39 050 883 451 
Polo S. Anna Valdera            Fax       : +39 050 883 452
viale Rinaldo Piaggio 34        e-mail    : pj...@ga...
56025 - Pontedera (PI) - ITALY  home page : http://feanor.sssup.it/~pj
------------------------------------------------------------------------
Per favore non mandatemi allegati in Word o PowerPoint.
(Please avoid sending me Word or PowerPoint attachments.)
Si veda (See) http://www.fsf.org/philosophy/no-word-attachments.html
------------------------------------------------------------------------

[Oslib-devel] trying the examples

[Paul D. <du...@ho...>]

I am playing with my rebuild X.
It's fun because for most examples, it works well.
In fact, I would say that only diskdemo don't work.
Even readdemo works... I rebooted the computer, and it did print what I had
put in my test file!

Some examples, are not able to return correctly to DOS.
DOS complains that the drive is invalid.

So...
So it seems it is mostly a problem with either int13h or maybe more 
precisely,
the floppy disk.

I'll try to figure out which is which tomorrow.

But I am surprised, I guess readdemo calls int21h in vm86 mode, and that
int21h call int13h !

_________________________________________________________________
MSN Search, le moteur de recherche qui pense comme vous !  
http://fr.ca.search.msn.com/

[Oslib-devel] X rebuild from sources (ver 5.2) works fine for me!

[Paul D. <du...@ho...>]

X rebuild from sources (ver 5.2) works fine for me!

_________________________________________________________________
MSN Messenger : discutez en direct avec vos amis !  
http://messenger.fr.msn.ca/

[Oslib-devel] In diskdemo, motor don't turn

[Paul D. <du...@ho...>]

Just to make clear, that I checked, and in the diskdemo,
the floppy doesn't seems to be really accessed.
I means, the light don't comes on, it goes too fast,
and I don't hear the motor turn.

_________________________________________________________________
MSN Search, le moteur de recherche qui pense comme vous !  
http://fr.ca.search.msn.com/

[Oslib-devel] found the january 31 203 snapshot

[Paul D. <du...@ho...>]

Oh, I see, on the sourceforge site.
http://sourceforge.net/project/showfiles.php?group_id=14468

But like I said, I was able to get cvs work.

_________________________________________________________________
MSN Messenger : discutez en direct avec vos amis !  
http://messenger.fr.msn.ca/

[Oslib-devel] the fact of X 5.1 and 5.2 don't works bothers me

[Paul D. <du...@ho...>]

I downloaded the sources of X.
He, this is quite old, 1996!
Ok, I look the source code of X.

I was please with the code to switch A20 on (good macro to wait, not just 
some jmp or
nop ).

I was also please to see that there is a nice A20query, to test if A20 lines 
are on or off.

But the fact that when I tried to load a .xtn file, it makes my computer 
reboots...
that's an interesting bug to search for. Since X is the key were BIOS calls 
are
made, I guess we should make sure to find where it is bugging with my 
computer.
Which is a Duron.

So I guess now I should go and find Turbo/C to compile X.

I think I have a problem with problems. Somehow, up to now, I feeled 
compulse to
find a solution, to the fact that I was unable to rebuild the oslib.
And now I am realizing that I don't really know how to continue improving 
the situation.
Now I can compile oslib. But my previous idea about A20 line being badly 
done is gone. Sorry. :-)
So now, I begin to be interested to X. I guess I will slow down a bit, see 
your suggestions.
And get back to real life a bit... :-)

_________________________________________________________________
MSN Messenger : discutez en direct avec vos amis !  
http://messenger.fr.msn.ca/

[Oslib-devel] diskdemo results (without LL_DEBUG)

[Paul D. <du...@ho...>]

First, I had to add diskdemo.xtn to the Makefile. As it was not there.
Indeed, it generates some warnings.

So here the results:
...
Got: 110000
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0x0
0x0 Starting... CX=F8
reset finished.
Drives: 0
Drive type: 0x32
Cylingers: 519
Sectors: 9
... read a sector: Bytes:
Ok!!!
End reached!
... (stack were equals)

There was no bytes of the sector written after Bytes:

What's drive is this?
My logical DOS partition or my floppy?

Anyway, does this helps?
I'll look at diskdemo.c to understand what I see.

_________________________________________________________________
MSN Search, le moteur de recherche qui pense comme vous !  
http://fr.ca.search.msn.com/

[Oslib-devel] x version 5.1 and 5.2 reboots the computer for me

[Paul D. <du...@ho...>]

I have tried x from your links, renaming them to x51.exe and x52.exe.
But reboot computer. (tried with diskdemo.xtn)

But the one from freedos32, that I renamed to xn, work well.

_________________________________________________________________
MSN Search, le moteur de recherche qui pense comme vous !  
http://fr.ca.search.msn.com/

[Oslib-devel] LL DEBUG option triggers a bug

[Paul D. <du...@ho...>]

In xlib:
vm86.c:338: #error Fix the following: use global_regs->xxx ???

So I am unable to build with LL_DEBUG option.

_________________________________________________________________
MSN Messenger : discutez en direct avec vos amis !  
http://messenger.fr.msn.ca/

Re: [Oslib-devel] preliminary results of biosdem and vmdemo

[Luca A. <luc...@em...>]

Hi Paul,

[...]
> I'll try with this debug option in config.mk at my next try.
> But for now, here what happened with biosdemo.xtn and vmdemo.xtn.
> 
> BiosDemo.xtn: a multicolored line from upper left to middle of screen have 
> appeared.
> I guess this is the wanted result.
Yes... Hence, it seems that vm86_callBIOS() is working... ;-)
 
> VmDemo.xtn:
> Some stuff all followed by: DONE
> Then:
> Multiboot Loading
> Getting area of size 2179904
> Got: 110 000
> 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0x0
> 0x0 Unhandled Exc or Int 0 occured!!!
> Halt called
Ok... There is some problem when the BIOS is used for accessing the
floppy... I'll have a look at the sources tonight, to see if these
information can help in finding the bug. Meanwhile, you can try to
enable the debugging.

> By the way, all the examples I tried finish by Halt called.
> Was'nt the idea of X to try stuff, and be able to return to DOS to
> try other stuff without having to reboot computer?
Yes... You are simply using the wrong x.exe version ;-)
Try the latest one from
http://sourceforge.net/projects/oslib
(it should be 5.2). Someone reported that the 5.2 x.exe binary give
problems... If you have problems with it, try 5.1.

				Luca
-- 
_____________________________________________________________________________
Copy this in your signature, if you think it is important:
                               N O    W A R ! ! !



--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Miotti.it: qualità e assortimento online di Ottica F.lli Miotti-Milano.
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=1452&d=10-7

[Oslib-devel] preliminary results of biosdem and vmdemo

[Paul D. <du...@ho...>]

Luca said:
>Thanks for doing this: it will help a lot!!! You may want to try the
>biosdemo and diskdemo examples, and to enable the debugging in xlib (by
>adding a -D__LL_DEBUG__ in C_OPT).

I'll try with this debug option in config.mk at my next try.
But for now, here what happened with biosdemo.xtn and vmdemo.xtn.

BiosDemo.xtn: a multicolored line from upper left to middle of screen have 
appeared.
I guess this is the wanted result.

VmDemo.xtn:
Some stuff all followed by: DONE
Then:
Multiboot Loading
Getting area of size 2179904
Got: 110 000
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0x0
0x0 Unhandled Exc or Int 0 occured!!!
Halt called

By the way, all the examples I tried finish by Halt called.
Was'nt the idea of X to try stuff, and be able to return to DOS to
try other stuff without having to reboot computer?

Ok, I added the LL_DEBUG option, I'll rebuild the system and try
biosdemo, diskdemo and vmdemo with it.

_________________________________________________________________
MSN Messenger : discutez en direct avec vos amis !  
http://messenger.fr.msn.ca/

[Oslib-devel] was able to make install under plain DOS!

[Paul D. <du...@ho...>]

I setup a DOS partition, copied djg2 directory, and booted with a DOS 
floppy.

Two small problems had happened:

1) The famous djg2/lib/dgjpp.ver with #include <sys/version.h>
I had to comment it out

2) Too many files open on : ar rs libhx.a xsys0.o ...
I changed the config.sys to change from files=10 buffers=10
to files=40 buffers=40

So at the end, I had the .xtn files in examples.
I have copied x.exe from the freedos32 floppy.

And I did a test:
x cpudemo.xtn

And the screen give nice info like my CPU have all the features.
At the end, the 2 stack pointers? were the same.

So now I am ready to chase my bug with BIOSDisk of freeDos32.
More precisely, VM86 mode.

Thanks for the suggestion of the patch for Cygwin, but I'll concentrate
on my bug with my BIOS versus OSLIB.

Conclusion: Avoid DJGPP and Windows XP!

_________________________________________________________________
MSN Messenger : discutez en direct avec vos amis !  
http://messenger.fr.msn.ca/

Re: [Oslib-devel] compiling ok on cygwin, but clean goes badly

[Luca A. <luc...@em...>]

Hi Paul,

as I previously wrote, cygwin is not supported (since I have not a
cygwin installation for testing).
There are some problems to be solved:
- cygwin uses a different executable format (not ELF nor COFF). Not a
big problem: we can generate a multiboot compliant image.
- the usual problem with the "_" prefix that some compilers attach to
all symbols as a prefix (this is the reason why you cannot use linux.mk,
as you already discovered)
- cygwin seems to introduce a call to alloca() in the startup of all
programs, and I do not know why...

Some time ago, I started to hack on it, and I wrote an patch:
http://sourceforge.net/tracker/index.php?func=detail&aid=718981&group_id=14468&atid=314468
(it basically introduces some new win32*.mk configuration files).
With such patch, it is possible to compile oslib with cygwin, but the
resulting executables _will not work_.
The patch is there just in case someone is willing to work on it and to
debug the problem (I think that the alloca() call is responsable for the
crash...).

If someone on the list has more knowledge about cygwin (in particular,
why is it insertig the malloca() call? Just to get the stack pointer?)
and is willing to help, let me know...

> I just had the idea of copying my oslib directory in cygwin, run a bash 
> shell
> and try make in the oslib directory...
> 
> It compiled without problems!
As you already discovered, it will not link :(
 
				Luca
-- 
_____________________________________________________________________________
Copy this in your signature, if you think it is important:
                               N O    W A R ! ! !



--
Email.it, the professional e-mail, gratis per te: http://www.email.it/f

Sponsor:
Sei un webmaster, vuoi guadagnare? Affiliati ad Artefiori. Clicca qui!
Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=751&d=10-7

Re: [Oslib-devel] cygwin with linux.mk as config.mk

[Paolo G. <pao...@ti...>]

On Thu, 2003-07-10 at 04:05, Paul Dufresne wrote:
> Slowly beginning to understand...
> I have copied linux.mk to ../config.mk to use cygwin.

maybe you can try the DJGPP version we are providing together with
shark, on http://shark.sssup.it/download.html

shark includes an oslib version that is something similar to a snapshot
taken in january, and it compiles both under djgpp and under gcc/linux
...

bye

PJ

-- 
Paolo Gai <pao...@ti...>
Scuola S. Anna