[lwip-users] Re: [lwip] Major critical region issues in pbuf.c

[Kieran Mansley]

Hi David,

Related issues to this have been brought up before and the
discussion should still be in the archives I think.  See (for example)

http://www.sics.se/mailing-lists/lwip.html/msg00717.html

(and the rest of the thread).

If I remember rightly Adam concluded by pointing out that as these are
high frequency operations adding proper locks around them would be a
massive overhead.  They are also only called from within the core of the
stack itself, and this is not designed (in the normal case) to be
multi-threaded anyway.  As it happens, there are a few problems with the
pbuf code, and it is on my list of things to take a look at.  I've also
added a lightweight lock to sys_arch, so I might try seeing if this helps
out in this case.

Does anyone else have any particular thoughts on this?

Kieran

[This message was sent through the lwip discussion list.]