You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Mark Phippard <ma...@gmail.com> on 2008/09/09 13:53:10 UTC
Weird JavaHL problem
I have had two different Subclipse on OSX users have a problem that
the JVM crashes as soon as Subclipse touches JavaHL. In both cases,
the error is this:
% gdb eclipse
(gdb) run
... lots of warnings ...
Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0xf1f21bc8
[Switching to process 644 thread 0x2c903]
0x92f850a8 in Pool::~Pool ()
(gdb) bt
#0 0x92f850a8 in Pool::~Pool ()
#1 0x317ee050 in SVNClient::setConfigDirectory ()
#2 0x317fe9e4 in
Java_org_tigris_subversion_javahl_SVNClient_setConfigDirectory ()
We are most likely passing a NULL for the path of the config
directory. It was actually a problem in 1.4 that this would get an
error, but it was fixed in 1.5. The thing that is really odd, is that
I do all my development on OSX and do not see this, as does the other
Subclipse developer. I think we are all using the binaries that
Jeremy Whitlock produces. So it just seems odd that the behavior
would vary.
I thought that maybe the first user I helped was somehow getting a
1.4.x version of the library loaded but that turned out not to be the
case. I also had this user specify ~/.subversion in Subclipse so that
we were passing a String and not a NULL.
Does anyone have any ideas what could cause this or what to look at?
--
Thanks
Mark Phippard
http://markphip.blogspot.com/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Weird JavaHL problem
Posted by Jeremy Whitlock <jc...@gmail.com>.
> Jeremy, the user is on PPC (OSX 10.4.11 (Tiger) on a PPC G4 (32-bit)).
> Any chance the problem is just with the PPC build of the JavaHL
> library? I know you have to build and then stitch them together into
> Universal Binary. Is it possible a 1.4.x version of library is being
> stitched in? I only ask that because the error that comes out is what
> you would see with 1.4.
There was a problem with the 1.4.x JavaHL. When we came out with
1.5.0-2, we fixed this.
--
Take care,
Jeremy Whitlock
http://www.thoughtspark.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Weird JavaHL problem
Posted by Mark Phippard <ma...@gmail.com>.
On Tue, Sep 9, 2008 at 11:27 AM, Jeremy Whitlock <jc...@gmail.com> wrote:
>> Does anyone have any ideas what could cause this or what to look at?
>
> The only thing I could think of would be to rebuild with debugging
> symbols on to get a better stack trace. Then you could do variable
> analysis to see which variable is NULL. I could even build a special
> developer build of the binary with debug on so that we could make sure
> it isn't a problem with the binary. Anything I can do to help, please
> let me know.
Jeremy, the user is on PPC (OSX 10.4.11 (Tiger) on a PPC G4 (32-bit)).
Any chance the problem is just with the PPC build of the JavaHL
library? I know you have to build and then stitch them together into
Universal Binary. Is it possible a 1.4.x version of library is being
stitched in? I only ask that because the error that comes out is what
you would see with 1.4.
--
Thanks
Mark Phippard
http://markphip.blogspot.com/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Weird JavaHL problem
Posted by Jeremy Whitlock <jc...@gmail.com>.
> Does anyone have any ideas what could cause this or what to look at?
The only thing I could think of would be to rebuild with debugging
symbols on to get a better stack trace. Then you could do variable
analysis to see which variable is NULL. I could even build a special
developer build of the binary with debug on so that we could make sure
it isn't a problem with the binary. Anything I can do to help, please
let me know.
--
Take care,
Jeremy Whitlock
http://www.thoughtspark.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org