Re: Antw: RE: [lwip-users] SemSignal gets called on aalreadysignaledsemaphore

[Jonathan Larmour]

Goldschmidt Simon wrote:
> I think a better solution would be to redesign tcp_sent() to signal
> the semaphore only if someone is really waiting.
> Otherwise some OSes could throw errors if a binary semaphore is
> signaled more than once.

I think it would be hard to deal with race conditions between testing that 
someone is waiting and doing the signal in one place, and setting that 
someone is waiting, and doing the wait in the other place.

Instead I think it would have to be down to the port to deal with it if 
the OS does that - they should be able to deal with it in an OS-specific 
way, whereas abstracting that in lwIP would be harder. So instead we can 
just expect the semaphore to be a pure binary one, with no problems 
associated with "extra" signals.

Or maybe it's better to change the abstraction to mutexes and condition 
variables, as has been discussed elsewhere :).

Jifl
--
eCosCentric Limited      http://www.eCosCentric.com/     The eCos experts
Barnwell House, Barnwell Drive, Cambridge, UK.       Tel: +44 1223 245571
Registered in England and Wales: Reg No 4422071.
------["The best things in life aren't things."]------      Opinions==mine