Go to the FreeLists Home Page Home Signup Help Login
 



[openbeosnetteam] || [Date Prev] [05-2002 Date Index] [Date Next] || [Thread Prev] [05-2002 Thread Index] [Thread Next]

[openbeosnetteam] Re: Tonight...

  • From: "David Reid" <dreid@xxxxxxxxxxxx>
  • To: <openbeosnetteam@xxxxxxxxxxxxx>
  • Date: Wed, 1 May 2002 23:25:53 +0100
I asked him and he just said it was part of their design...

FWIW, I'm running with the timer ops as threads and so far haven't seen the
problem. I think therefore we'll go back to using your timer threads for
kernel land as well... Once the download finishes I'll look at making the
changes.

If you want to do the userland stack thing that'd be very cool :)

Next is to find out why we have memory issues...

david

----- Original Message -----
From: "Axel Dörfler" <axeld@xxxxxxxxxxxxxxxx>
To: <openbeosnetteam@xxxxxxxxxxxxx>
Sent: Thursday, May 02, 2002 1:04 AM
Subject: [openbeosnetteam] Re: Tonight...


> > Hmm, well there's a new problem in town and he looks like this...
> >
> > KERNEL : area_resize : cannot grow area at 0x80157000 (heap) in
> > /boot/home/net_t
> > est/NetPositive (000fb000 -> 01240000)
> > resize_area error
>
> It looks we're allocating too much memory somewhere, so perhaps just
> try to track those.
> BTW for testing a userland version of the stack would be very
> beneficial. You can just load applications like "ping" from the stack
> as an add-on, and have a libnet.so that's directly linked against the
> stack - this way you can start almost every application from a userland
> stack without having to implement IPC.
>
> > KERNEL PANIC: add_timer(001a6a60, 3000, 00129750, 2) already in timer
> > queue
> > (idl
> > Now, the timers are interrupt driven and so I wonder if we're doing
> > something in them that we shouldn't be or are simply taking too long
> > and so
> > the idle thread is inserting 2 interrupts... Thoughts?
>
> Since BONE also comes with its own timer functions, I just would always
> use my version, since that one is really non-critical and thread-safe.
> If we really want to investigate in it again, we can do so later, but I
> don't think it's worth the effort, and I don't feel good to use
> interrupt driven kernel timer functions for the stack.
> Just ask Howard why he implemented his own timers in BONE, and I guess
> you should have a good answer to our problems :-)
>
> Adios...
>    Axel.
>
>
>
>






[ Home | Signup | Help | Login | Archives | Lists ]

All trademarks and copyrights within the FreeLists archives are owned by their respective owners.
Everything else ©2007 Avenir Technologies, LLC.