[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: apr/javahl crash on Win32 (was Re: JavaHL 1.2 Problem -- Test Case attached)

From: Patrick Mayweg <mayweg_at_qint.de>
Date: 2005-05-02 09:01:11 CEST

Hi Mark,

Mark Phippard wrote:

>Thanks for testing and for catching the true/false mistake. dlr actually
>wanted me to commit it to trunk. I wanted to see if I could find another
>example, where the checkout of the WC would be smaller. If I can, then I
>will probably commit that.
>Neither Subversion or APR call that apr_get_username function. I suspect
>the reference to that, while possibly a clue, is also likely a red-herring.
>Maybe the memory being overwritten is in that location?
I would not think to much about it. If you are running out of stack
space, the stack may not be reliable.

>I also wonder what the reference to 0x47 means. I thought maybe that is an
>ordinal number, but 0x47 in libapr.dll is not this function. Again,
>perhaps another clue?
That is the byte offset of the start of the function.

>I think it is going to take someone capable of running this in the debugger
>to figure it out.
My main thought right now is that the stack space of a java thread may
be quiet small and you are really running out of stack space. No endless
recurse or any of that. The main thread of an application has a quiet
big stack. The http protocol uses neon and may be the heaviest on the
stack. On different platform the default stack space for a java thread
may be larger.

As said before, I am on restricted resources till sunday. If that
problem does not get fixed till then, it will be my first priority.

To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon May 2 09:03:17 2005

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.