DragonFly commits List (threaded) for 2005-05
Re: cvs commit: src/sys/i386/i386 nexus.c src/sys/i386/include atomic.h src/sys/kern kern_poll.c lwkt_serialize.c src/sys/net if.c if_var.h rtsock.c src/sbin/ifconfig ifconfig.c src/sys/dev/netif/dc if_dc.c src/sys/dev/netif/em if_em.c if_em.h ...
:> Not sure what that is, it isn't the bug I was talking about. The
:> bug I introduced was a case where the EM interface simply stops
:Just a question before my installing the new kernel - does this code work
:correctly for non-IFF_POLLING case (after em_disable_intr() is called,
:it never calls lwkt_serialize_handler_enable())?
Not for EM. It will mess up with or without polling. I think I
found the trouble, I screwed up some code that was handling a particular
race condition. I'm testing the fix now and expect to commit tonight.
The code you quoted is correct. The default initialization of the
serializer is 'enabled'. For polling operation the interrupt is
disabled. When polling is turned off the interrupt is re-enabled.