This is the mail archive of the
ecos-patches@sources.redhat.com
mailing list for the eCos project.
Re: Net stack serialization fix
Andrew Lunn <andrew.lunn@ascom.ch> writes:
> What happens when the stack sleeps at the well known sleep
> points. Surely it should unlock the mutex there so that other threads
> can use the stack?
>
It will, the sleep() routine already does this. All the sleeps are
already inside splxxx() regions. All I am actually doing in the patch
is to claim the lock a little earlier and release it a little
later. The internal calls to splsoftnet() etc. in the stack
essentially become null operations.
--
Nick Garnett - eCos Kernel Architect
http://www.eCosCentric.com/