Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!seismo!gatech!bloom-beacon!husc6!harvard!spdcc!dyer From: dyer@spdcc.COM (Steve Dyer) Newsgroups: comp.unix.wizards Subject: Implementing 4.X TCP on non-VAXes Message-ID: <203@spdcc.COM> Date: Wed, 22-Jul-87 12:24:31 EDT Article-I.D.: spdcc.203 Posted: Wed Jul 22 12:24:31 1987 Date-Received: Fri, 24-Jul-87 05:09:30 EDT Organization: S.P. Dyer Computer Consulting, Cambridge MA Lines: 15 Summary: software interrupts on machines without 'em 4.1BSD network code was switchable between having a network process for consuming network input or using the VAX software interrupt system to do this at interrupt level and avoid the overhead of a context switch. The 4.2 and 4.3 network code assumes a software interrupt system. Since not every hardware architecture supports this feature and the code has been ported to a wide variety of different machines, I was wondering what the most common ways of implementing this were. Do you examine the state of the software interrupt flag upon returning from a hardware interrupt? Check it during a timeout() routine? Thanks, -- Steve Dyer dyer@harvard.harvard.edu dyer@spdcc.COM aka {ihnp4,harvard,linus,ima,bbn,m2c}!spdcc!dyer