lwip-devel
[Top][All Lists]
Advanced

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

Re: [lwip-devel] core/timers.c with NO_SYS?


From: Leon Woestenberg
Subject: Re: [lwip-devel] core/timers.c with NO_SYS?
Date: Mon, 13 Jun 2011 22:36:56 +0200

Hello Bill,

On Mon, Jun 13, 2011 at 10:00 PM, Leon Woestenberg
<address@hidden> wrote:
> On Mon, Jun 13, 2011 at 8:25 PM, Bill Auerbach
> <address@hidden> wrote:
>>
>> I wouldn't change it.  IMO if lwIP were going to work in a fully preemptive
>> RTOS environment with shared PCB and sockets, it needs to be designed to do
>> so, not shoehorned in.
>>
> Uhm, I'm exactly using lwIP in such systems and with shared PCB's and
> sockets. However, not using sys_arch of course.
>

Giving this some thoughts, lwIP indeed does not seem fit for the kind
of system Bill describes (and I want), at least not with what we
currently have available above the core layer.

I've implemented these systems out of contrib, but now that there is a
very wide and open space of fully pre-emptive RTOSses, my goal is to
have a new version of lwIP that matches these systems and have one or
two RTOS ports.

With fully pre-emptive, I also mean minimal interrupts disabled time.

However, that means disabling anything beyond the core API, and doing
the RTOS integration in a proper manner, not the current stuff which
doesn't scale to such systems.

Agreed with Bill, that in some cases does require finer grained
locking, but these are not interrupt disables, rather mutexes or
condition variables.

Regards,
-- 
Leon



reply via email to

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