[ECOS] Flag query

Gary Thomas gary@mlbassoc.com
Fri Oct 27 15:08:00 GMT 2006


Bell, Andrew [Allen & Heath UK] wrote:
> Hello All
> 
> I'm still trying to diagnose a netstack stall, and have traced the
> packet delivery through my fec driver, into the generic Ethernet driver
> and into if_ethersubr.c which demuxes the different protocols and then
> calls software interrupts using schednetisr. In eCos this function seams
> to be implemented using flags, the network thread sleeps on
> cyg_flag_wait, and when schednetisr or's in a bit signaling a particular
> protocol has arrived and is ready for a push into the protocol stack
> cyg_netint in support.c is suppose to wake and call the appropriate
> netisr handler. Fine. 
> 
> My trace shows the call to schednetisr, and inside flags.cxx I can see
> setbits is called, but cyg_netint (network thread) fails to wake. My
> question .... Is there a good way of debugging the scheduling loop, from
> what I can see the bottom of setbits in flags.cxx should call
> Cyg_Scheduler::unlock, shouldn't this cause a reschedule ? My network
> thread priority in higher than any user thread (lower numerically), im
> kind of hoping for a context switch. Eventually my connection gets
> dropped as the starvation of my tcp/ip stack means no acks are being
> tx'd.
> 
> Any help as always would be greatly appreciated.

If you connect via GDB and break when this happens, then you should
be able to get information via:
   (gdb) info threads

This should tell you information about all of the threads in the system.

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss



More information about the Ecos-discuss mailing list