gnupic: Thread: Re: [gnupic] gpsim

polly wrote:
> Hello,
>   I'm trying to run gpsim on a simple cod file output from
> SDCC on a pic18f252. I'm seeing inconsistent effects:
> 
>  From the command line the .cod file sometimes loads, sometimes
> does not. From the gpsim prompt, attempting to load the .cod file
> results in an immediate segfault.

Could you show us the command you used?

> On the few occasions the code succeeds in loading, the effect
> of breakpoints seems entirely arbitrary: the PC stops at points
> where no breakpoint is set, other breakpoints are ignored, stepping
> after a breakpoint works or doesn't, depending.

Are you sure that the .cod file and the source are from the same build? 
Are you experiencing WDT timeouts?

>                                     There's a console message
> about 'HLL not supported yet' ?? Should I be able to step through a simple
> C app built on SDCC ?

I removed support for HLL debugging because it is broken. This feature 
is not currently supported in gpsim.

> 
> The RAM view of eg T0CON registers seems unaffected by instructions setting
> and clearing bits in the registers. Are internal regs in 18F252 simulated ?

What does the trace show? Could it be that your bank registers are 
pointing to the wrong place?

> I cannot get rid of console messages assuring me that symbols are not  
> supported
> in the font style chosen, I think I've tried every font setting offered.
> Is this affecting whether or not the code module is loaded ? Where can
> I find hints about Locale problems ?

This is an issue with parts of the gui still using older GTK APIs. I've 
fixed the source browser and the register window, but not the symbol viewer.

> 
>   I have been running the CVS version of gpsim but I'm not able to get an
> SVN update;  "SSL not supported" .. browsing on the web suggests I have to
> build not only SVN but some app called Neon ?  Where can I find  
> instructions
> for getting SVN running ?

I'd start off with google.

http://www.google.com/search?hl=en&q=subversion&btnG=Google+Search

The first link is to:

http://subversion.tigris.org/

Which has everything you need.

Also, you can find gpsim-specific subversion notes here:
http://sourceforge.net/svn/?group_id=2341


> Can anyone offer clues for building the pty code needed in order for 
> the  USART
> funtions to be available on systems other than BSD .. like core 4 ? I  
> promise
> to share ......

Robert pointed this out to me the other day. I didn't know this was 
still around. If you only need USART functionality, then you can use the 
built in gpsim usart module. This module lacks the capability of 
communicating with the kernel.

Scott

####@####.#### wrote:
> I am new to gpsim. Got the zip packags from SourceForge. Built gpsim & graphicled without
> problem. lcd-0.2.9 complains about not finding install.sh.
>   
Strange. I don't believe install-sh is even used...

However, you may be able to get around this by re-creating the whole new 
install script suite. Try running these commands in sequence:

If there's a directory called 'autom4te.cache', remove it:

$ rm -rf autom4te.cache

Then run these commands

$ aclocal
$ libtoolize --force
$ autoheader
$ automake --add-missing
$ autoconf
$ ./configure
$ make
$ make install

BTW, if you're getting code from SVN, then this is the recommended 
procedure for customizing the install scripts. However, the tar ball 
should have a 'configure' script that be all that you need.

> I have a small project that will be using a 128x160 graphic lcd and the 18f252. When I execute the example I get a seqmentation fault when expanding the breadboard window.
>   
I don't see this behavior at all. Are you sure it's the breadboard 
resizing that's the problem and not the register window resizing? The 
reason I ask is  because there's a bug in 'gtk-extra' (this package 
supplies the widget used in the register window) that causes a segv 
whenever the register window is resized.

Scott

Scott,
I just repeated the install and get identical failures. System is a Centrino Duo running
and up to date Fedora Core 5.

1. ./configure in lcd is what fails looking for install-sh or install.sh.
2. Using a slider on the breadboad window in graphiclcd example segfaults. Also get warnings
    about missing fonts. The console shows 
                  0: p18f452 Execution done(0x298)
                             Message:***PASSED LCD test
                  1: p18f452 Execution at 0x29c
                             Message:***FAILED LCD test

George

-------------- Original message -------------- 
From: Scott Dattalo ####@####.#### 

> ####@####.#### wrote: 
> > I am new to gpsim. Got the zip packags from SourceForge. Built gpsim & 
> graphicled without 
> > problem. lcd-0.2.9 complains about not finding install.sh. 
> > 
> Strange. I don't believe install-sh is even used... 
> 
> However, you may be able to get around this by re-creating the whole new 
> install script suite. Try running these commands in sequence: 
> 
> If there's a directory called 'autom4te.cache', remove it: 
> 
> $ rm -rf autom4te.cache 
> 
> Then run these commands 
> 
> $ aclocal 
> $ libtoolize --force 
> $ autoheader 
> $ automake --add-missing 
> $ autoconf 
> $ ./configure 
> $ make 
> $ make install 
> 
> BTW, if you're getting code from SVN, then this is the recommended 
> procedure for customizing the install scripts. However, the tar ball 
> should have a 'configure' script that be all that you need. 
> 
> > I have a small project that will be using a 128x160 graphic lcd and the 
> 18f252. When I execute the example I get a seqmentation fault when expanding the 
> breadboard window. 
> > 
> I don't see this behavior at all. Are you sure it's the breadboard 
> resizing that's the problem and not the register window resizing? The 
> reason I ask is because there's a bug in 'gtk-extra' (this package 
> supplies the widget used in the register window) that causes a segv 
> whenever the register window is resized. 
> 
> Scott 
> 
> --------------------------------------------------------------------- 
> To unsubscribe, e-mail: ####@####.#### 
> For additional commands, e-mail: ####@####.#### 
> 

####@####.#### wrote:
> Scott,
> I just repeated the install and get identical failures. System is a Centrino Duo running
> and up to date Fedora Core 5.
>   
Hi George,

I was able to duplicate *some* of this behavior and have determined that 
in my case the problem is with gtkextra. In my case, I just installed 
FC6 and used a copy of gtkextra from one of my old linux setups. This 
built and installed just fine, but gpsim segfaults whenever the register 
window is resized. So, I removed this version of gtkextra and installed 
the one Borut create for gpsim Win32:

http://gpsim.sourceforge.net/gpsimWin32/gpsimWin32.html
and in particular:
http://gpsim.sourceforge.net/gpsimWin32/packages/gtk+extra-2.1.1-src-20060611.zip

Unfortunately, the text files use DOS end-of-line termination. So after 
unzipping the file I ran dos2unix in the gtkextra directory. Here's a 
copy of my history:

  7XX  mkdir gtk+extra-2.1.1
 <snip>
  715  mv gtk+extra-2.1.1-src-20060611.zip gtk+extra-2.1.1
  716  cd gtk+extra-2.1.1
  717  ls
  718  unzip gtk+extra-2.1.1-src-20060611.zip
<snip>
  729  dos2unix *
  730  ls
  731  ./configure
  732  make
  733  ls
  734  aclocal
  735  libtoolize --force
  736  autoheader
  737  automake
  738  automake --add-missing
  739  autoconf
  740  ls
  741  rm d2utmppUXY2J
  742  ./configure
  743  make
  744  make install

The command at 741 is removing the failed result of dos2unix trying to 
convert a directory.

After this I rebuilt gpsim and the register window resizes correctly.

Now I realize that you're having problems with the breadboard resizing 
and not the register window. So here are a couple of things to try:

1) remove the file ~/.gpsim
This removes the history of the window states.
2) Invoke gpsim with no PIC source. Then open the breadboard window and 
resize it to the size you like. Quit gpsim and then restart it with the 
graphics LCD code. This should at least allow you to use gpsim.

If (1) fixes then I'll say your .gpsim file got corrupted. Solution (2) 
is only a work around. I'd still recommend installing win32 version of 
gtkextra though.

----
I don't know what to say about the 'install.sh' problem. Did you try 
recreating configure.in using the commands I wrote in the previous message?

Scott

Scott Dattalo wrote:
>
>  <snip>
>
> The command at 741 is removing the failed result of dos2unix trying to 
> convert a directory.
Instead of "dos2unix *" you can run:

find . -type f -exec dos2unix {} \;

which converts only files.


Borut

Scott,

The first suggestion of removing .gpsim, resizing the breadboard and then running the lcd
sim segfaults immediately on starting my sim run with lcd.  Will try second scheme tonight.

George

-------------- Original message -------------- 
From: Scott Dattalo ####@####.#### 

> ####@####.#### wrote: 
> > Scott, 
> > I just repeated the install and get identical failures. System is a Centrino 
> Duo running 
> > and up to date Fedora Core 5. 
> > 
> Hi George, 
> 
> I was able to duplicate *some* of this behavior and have determined that 
> in my case the problem is with gtkextra. In my case, I just installed 
> FC6 and used a copy of gtkextra from one of my old linux setups. This 
> built and installed just fine, but gpsim segfaults whenever the register 
> window is resized. So, I removed this version of gtkextra and installed 
> the one Borut create for gpsim Win32: 
> 
> http://gpsim.sourceforge.net/gpsimWin32/gpsimWin32.html 
> and in particular: 
> http://gpsim.sourceforge.net/gpsimWin32/packages/gtk+extra-2.1.1-src-20060611.zi 
> p 
> 
> Unfortunately, the text files use DOS end-of-line termination. So after 
> unzipping the file I ran dos2unix in the gtkextra directory. Here's a 
> copy of my history: 
> 
> 7XX mkdir gtk+extra-2.1.1 
> 
> 715 mv gtk+extra-2.1.1-src-20060611.zip gtk+extra-2.1.1 
> 716 cd gtk+extra-2.1.1 
> 717 ls 
> 718 unzip gtk+extra-2.1.1-src-20060611.zip 
> 
> 729 dos2unix * 
> 730 ls 
> 731 ./configure 
> 732 make 
> 733 ls 
> 734 aclocal 
> 735 libtoolize --force 
> 736 autoheader 
> 737 automake 
> 738 automake --add-missing 
> 739 autoconf 
> 740 ls 
> 741 rm d2utmppUXY2J 
> 742 ./configure 
> 743 make 
> 744 make install 
> 
> The command at 741 is removing the failed result of dos2unix trying to 
> convert a directory. 
> 
> After this I rebuilt gpsim and the register window resizes correctly. 
> 
> Now I realize that you're having problems with the breadboard resizing 
> and not the register window. So here are a couple of things to try: 
> 
> 1) remove the file ~/.gpsim 
> This removes the history of the window states. 
> 2) Invoke gpsim with no PIC source. Then open the breadboard window and 
> resize it to the size you like. Quit gpsim and then restart it with the 
> graphics LCD code. This should at least allow you to use gpsim. 
> 
> If (1) fixes then I'll say your .gpsim file got corrupted. Solution (2) 
> is only a work around. I'd still recommend installing win32 version of 
> gtkextra though. 
> 
> ---- 
> I don't know what to say about the 'install.sh' problem. Did you try 
> recreating configure.in using the commands I wrote in the previous message? 
> 
> Scott 
> 
> 
> --------------------------------------------------------------------- 
> To unsubscribe, e-mail: ####@####.#### 
> For additional commands, e-mail: ####@####.#### 
> 

Scott,
Same result with win32 version of gtk+extra. Segfault on resizing the breadboard.
George

####@####.#### wrote:
> Scott,
> Same result with win32 version of gtk+extra. Segfault on resizing the breadboard.
> George
>   
Bummer...

But I think I've reproduced the problem

Which version of gtk are you using?

$ pkg-config --modversion gtk+-2.0

I've tested 2.8.?? and 2.10.4 with no problems. But I just tried an old 
system with 2.6.7 and whammo, it segfaults whenever the graphic lcd is 
exposed in the breadboard window. This is good clue and I'll explore it 
this weekend (probably).

Scott

Scott,

I have 2.8.20

George

-------------- Original message -------------- 
From: Scott Dattalo ####@####.#### 

> ####@####.#### wrote: 
> > Scott, 
> > Same result with win32 version of gtk+extra. Segfault on resizing the 
> breadboard. 
> > George 
> > 
> Bummer... 
> 
> But I think I've reproduced the problem 
> 
> Which version of gtk are you using? 
> 
> $ pkg-config --modversion gtk+-2.0 
> 
> I've tested 2.8.?? and 2.10.4 with no problems. But I just tried an old 
> system with 2.6.7 and whammo, it segfaults whenever the graphic lcd is 
> exposed in the breadboard window. This is good clue and I'll explore it 
> this weekend (probably). 
> 
> Scott 
> 
> --------------------------------------------------------------------- 
> To unsubscribe, e-mail: ####@####.#### 
> For additional commands, e-mail: ####@####.#### 
> 

On Fri, 2007-01-12 at 23:00 +0000, ####@####.#### wrote:
> Scott,
>
> I have 2.8.20

Hmm. I have no idea why this would be failing for you (since gtk 2.8.20
works for me). However, I do think there's something bogus in the graphic
LCD. I just check in some code that renders the graphic LCD in its own
window. This behavior is now identical to the text LCD.

I added a '#define IN_BREADBOARD 0' to the graphic LCD code. If this is
changed to 1 then the LCD is rendered in the breadboard window (like it
used to be). At the moment, it is set to 0

George, can you try this new code and let me know how it works?

Thanks,
Scott