Getting a segfault on vm built from sources on gentoo Linux on an amd64

 
Hi, I seem to be having trouble getting squeak to run on one of my machines.

It's an amd64 running gentoo Linux with x86_64 as the architecture.

If I build Squeak from the current sources I get the following segfault:

miles@salmon ~/src/Squeak-3.9-8 $ squeak Squeak3.9-final-7067.image

Segmentation fault

16405880 >?
16256940 SmalltalkImage>snapshot:andQuit:embedded:
16256824 SmalltalkImage>snapshot:andQuit:
16256396 SmalltalkImage>saveImageInFileNamed:
16246520 SmalltalkImage>saveAs
16246336 TheWorldMenu>doMenuItem:with:
16246244 [] in MenuItemMorph>invokeWithEvent:
16246152 BlockContext>ensure:
16246060 Cursor>showWhile:
16245968 MenuItemMorph>invokeWithEvent:
16245876 MenuItemMorph>mouseUp:
16245784 MenuItemMorph>handleMouseUp:
16245652 MouseButtonEvent>sentTo:
16245560 Morph>handleEvent:
16245468 MorphicEventDispatcher>dispatchDefault:with:
16245376 MorphicEventDispatcher>dispatchEvent:with:
16245284 Morph>processEvent:using:
16245192 MorphicEventDispatcher>dispatchDefault:with:
16245084 MorphicEventDispatcher>dispatchEvent:with:
16244952 Morph>processEvent:using:
16244860 Morph>processEvent:
16244768 MenuMorph>handleFocusEvent:
16244492 [] in HandMorph>sendFocusEvent:to:clear:
16244584 [] in PasteUpMorph>becomeActiveDuring:
16244400 BlockContext>on:do:
16244308 PasteUpMorph>becomeActiveDuring:
16244124 HandMorph>sendFocusEvent:to:clear:
16244032 HandMorph>sendEvent:focus:clear:
16243940 HandMorph>sendMouseEvent:
16243796 HandMorph>handleEvent:
16243540 HandMorph>processEvents
16243632 [] in WorldState>doOneCycleNowFor:
16243448 SequenceableCollection>do:
16243356 WorldState>handsDo:
16243264 WorldState>doOneCycleNowFor:
16243172 WorldState>doOneCycleFor:
16243080 PasteUpMorph>doOneCycle
15589312 [] in >spawnNewProcess
15589496 [] in BlockContext>newProcess
Aborted
miles@salmon ~/src/Squeak-3.9-8 $


If I try using the i686-pc-linux-gnu download, I have to force it to
use that architecture, and then I get the following error:

miles@salmon ~/src/Squeak-3.9-8 $ squeak Squeak3.9-final-7067.image
could not find display driver vm-display-X11; either:
  - check that /usr/local/lib/squeak/3.9-8/vm-display-X11.so exists, or
  - use the '-plugins <path>' option to tell me where it is, or
  - remove DISPLAY from your environment.
Aborted

I do have the file at /usr/local/lib/squeak/3.9-8/vm-display-X11 but
it doesn't end in .so  I'm not sure if that's OK or not.

if it is needed, config.guess outputs:
salmon Squeak-3.9-8 # ./config.guess
x86_64-unknown-linux-gnu


On some tangent questions:

Also, is inisqueak now obsolete?  The i686 installer didn't include
one (but it did include the man page for it) and the source install
built one but didn't install it.  And if I try to run it from the
build directory it doesn't behave as described in the man file.


I also get this error when running make uninstall:

salmon bld # make uninstall
if test -n ""; then \
          /bin/sh
/home/miles/src/Squeak-3.9-8/platforms/unix/config/uninstall ; \
        fi
rm -f /usr/local/squeak/lib/squeak/3.9-8/squeak
rm -f /usr/local/squeak/bin/squeak
rmdir /usr/local/squeak/bin
/bin/sh: -c: line 6: syntax error: unexpected end of file
make: *** [uninstall-plugins] Error 2
salmon bld #

Hopefully this information is useful.  If you would like me to do
anything to give you more information about any of the situations
mentioned above, just let me know.

 
On Wed, Dec 19, 2007 at 01:47:36AM -0800, Miles Georgi wrote:
>  
> Hi, I seem to be having trouble getting squeak to run on one of my machines.
>
> It's an amd64 running gentoo Linux with x86_64 as the architecture.

Here is what I do:
  http://lists.squeakfoundation.org/pipermail/vm-dev/2007-December/001703.html

Use the most up to date sources from Subversion on squeakvm.org also.

> If I build Squeak from the current sources I get the following segfault:
>
> miles@salmon ~/src/Squeak-3.9-8 $ squeak Squeak3.9-final-7067.image
>
> Segmentation fault
>
> 16405880 >?
> 16256940 SmalltalkImage>snapshot:andQuit:embedded:

If you follow the recipe above, it should work fine. However, if you
are still getting seg faults, try running your vm under a debugger
(gdb, ddd, etc). When it dies, the debugger will show you where it's
happening.  Sometimes that is enough to point you to the problem.

Dave

 
David, thanks, I'll give that a try

I did use a debugger to see where it died and it didn't really reveal
much.  It actually looked quite odd that it died where and why it did.

I won't bother posting information about that until I have given your
approach a try

Thanks again!

MIles



On 12/19/07, David T. Lewis <[hidden email]> wrote:

 
David, the link to the final attatchment in the archive you listed
above for step 7
(http://lists.squeakfoundation.org/pipermail/vm-dev/2007-November/001675.html)
looks garbled.

I'm assuming I can get the file by just replacing the filename in the
previous url with
platforms-unix-plugins-AsynchFilePlugin-sqUnixAsynchFile.c.diff but I
figured I'd give you a heads up incase you hadn't noticed that before.

Oh, and building from the svn source got me passed all of the
segfaults I was having before, thanks again!

On 12/19/07, David T. Lewis <[hidden email]> wrote: