This is the mail archive of the java@gcc.gnu.org mailing list for the Java project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: 'thread stack pointer out of range'? -- solved?


Adam King <aking@dreammechanics.com> writes:
> > This is on Win98
> ..
> running Win98 SE.

Yeeks, mine is SE as well. Interesting.

> > This leads me to believe that it may be the socket read()/write()
> > routines that are doing this.

> hmm.. that's interesting - it occured for us while the client was
> downloading data from the server - so you might be onto something
> there.  

[[ Adam M. googles furiously ]]

Egads, I think I found it -- from http://www.uae.de/bsdsocket.txt

> Some in-depth research on the Windows 9x incompatibility problem
> reveals that the majority of the crashes were caused by the WSP/NSP
> starting up in a "magic stack" context, which was easily fixed by
> making a dummy WinSock call after initializing the socket layer with
> a regular stack pointer. Unfortunately, some WinSock functions would
> crash nevertheless (e.g. connect() and sendto()). I kludge-fixed
> connect() by running it in an extra thread, but then I got tired of
> wasting my time working around Microsoft's sloppy programming.

This suggests that Hans' aforementioned fix which works when SP is off
by "a lot" should be sufficient -- WinSock is flipping over to an
entirely different stack. I'm not quite sure what the author means
here by "making a dummy WinSock call", though. I'll try to contact him.

  - a

-- 
Sick of HTML user interfaces?
www.xwt.org

"disappointing retail sales figures suggest to some that America's love
 affair with retail therapy may be over."    -- _The Economist_, 19-Jun


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]