wine users - Sep 2002 - Is wine *always* mapping threads onto processes? (Was: Re: multiple inst

Well then, there you go. I really hope that I remembered to apply
<stddisclaimer.h> to what I said.

/Fredrik
> 
> From: Kevin DeKorte <kdekorte@yahoo.com>
> Date: 2002/09/11 Wed PM 04:14:27 CEST
> To: Frank Joerdens <frank@joerdens.de>,  wine-users@winehq.com
> Subject: Re: Is wine *always* mapping threads onto processes? (Was: Re:
multiple instances  . . . )
> 
> Frank,
> 
> Linux does not have a lightweight thread model in the
> stable kernels. I believe that the dev kernels have
> implemented it but wine or glibc would have to be
> changed to take advantage of it I believe. 
> 
> Actually the memory you are seeing is probably NOT
> 32MB per process. The memory probably overlaps
> somewhat in that the code that is not being changed is
> not being copied only the data would be written again.
> So of that 32MB probably 30MB of it is shared across
> all the processes with only about 2MB being unique to
> each process. The memory in use being reported is a
> side affect of the Linux thread=pid model. This is
> pretty normal behavior for a Linux box. Look at a java
> program that has multiple threads and you will see the
> same behavior. 
> 
> 
> --- Frank Joerdens <frank@joerdens.de> wrote:
> > On Wed, Sep 11, 2002 at 04:59:35AM +0200, Sylvain
> > Petreolle wrote:
> > > it's probably a unique instance of the app with 5
> > threads in it.
> > > each windows thread matches a unix-like process.
> > 
> > Is that the general way wine translates windows
> > threads, mapping them
> > onto processes? Would that really be a good idea?
> > Usually the argument
> > goes that designing an app in a multithreaded rather
> > than multiprocess
> > fashion is better, more resource friendly, albeit
> > harder to do (e.g.
> > because inter-thread communication is much easier to
> > do than
> > inter-process communication).
> > 
> > In my particular case, if the application were to
> > spawn a few more
> > threads, I would run out of physical memory pretty
> > quick, with each
> > one of those threads/processes consuming 32 megs.
> > 
> > Regards, Frank
> > _______________________________________________
> > wine-users mailing list
> > wine-users@winehq.com
> > http://www.winehq.com/mailman/listinfo/wine-users
> 
> 
> ====> ---------------------------------
> Kevin DeKorte
> kdekorte@yahoo.com
> 
> 
> __________________________________________________
> Yahoo! - We Remember
> 9-11: A tribute to the more than 3,000 lives lost
> http://dir.remember.yahoo.com/tribute
> _______________________________________________
> wine-users mailing list
> wine-users@winehq.com
> http://www.winehq.com/mailman/listinfo/wine-users
>