[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: NetBSD drivers on HURD?

From: Matthieu Lemerre
Subject: Re: NetBSD drivers on HURD?
Date: Sat, 30 Jul 2005 15:34:48 +0200
User-agent: Gnus/5.11 (Gnus v5.11) Emacs/22.0.50 (gnu/linux)

Daniel Wagner <address@hidden> writes:

> Hi Mike,
> On Friday 29 July 2005 21:00, address@hidden wrote:
>> So far, my plan for the device driver framework is to finish up anything
>> deva/fabrica may have left unfinished or unclean.  
> The fabrica code is only a mess because I could never get near anything what 
> we needed.  It was the chicken-egg problem. I needed some infrastructure from 
> hurd-l4 but  hurd-l4 is depending on the drivers. Anyway, the code was a 
> first try but I don't think it very useful at all at this point. You might be 
> better of starting from scratch.
> cheers,
> daniel

Still I think there is interesting things, like:

-In the TODO, there are notes about drivers should add entropy to a
 random spool. I wouldn't have thought about that, and it's something
 that Mike would have to take into consideration.

-I don't know if libfthread is working, but I guess a lightweight
 thread library can be needed (maybe not for device drivers, but for
 things like the plm maybe?). Not a priority though; I'v read some
 device drivers and they didn't appear to be multithreaded.

-The create_archive script will be useful too.

I've been looking a bit about the FreeBSD drivers.  There is indeed a
 really clean interface between the modules, but that does not mean
 that we could tranlate a kobj method call into a RPC: for instance
 interrupts handler functions pointer are passed as a parameter in
 some method , like in 

 if ((error = bus_setup_intr(dev, hsc->hsc_ires, INTR_TYPE_NET, hme_intr,
             sc, &hsc->hsc_ih)) != 0)

There is also structure pointers passed as parameters for networking
spool and stuff like that.

What this mean is that reusing freeBSD drivers will mean having them
running in the same address space, or at least some subsystems running
together in the same address space.

But we could maybe gradually change some RPCs to run them in separate
address spaces.


> _______________________________________________
> L4-hurd mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/l4-hurd

reply via email to

[Prev in Thread] Current Thread [Next in Thread]