Path: utzoo!utgpu!jarvis.csri.toronto.edu!clyde.concordia.ca!uunet!bfmny0!tneff From: tneff@bfmny0.UU.NET (Tom Neff) Newsgroups: comp.unix.i386 Subject: Re: YASPP (Yet another serial port problem) Message-ID: <15040@bfmny0.UU.NET> Date: 28 Dec 89 21:28:02 GMT References: <10740@attctc.Dallas.TX.US> <12653@cbnews.ATT.COM> Reply-To: tneff@bfmny0.UU.NET (Tom Neff) Lines: 24 In article <12653@cbnews.ATT.COM> mjs@cbnews.ATT.COM (martin.j.shannon,59112,lc,4nr10,201 580 5757) writes: >Yeah, well, DOS doesn't use interrupts for the COM devices, so the fact that >a serial port runs under DOS *DOES* *NOT* mean you have it configured properly >for UNIX (ISC or any other brand of System V Release 3+ on the '386). The above quoted poster may be confusing the fact that the standard IBM PC compatible BIOS serial port service (interrupt 0x14) is polled rather than interrupt driven, so if you use the BIOS (or the default DOS named COMn: devices, whose drivers use the BIOS) for your async I/O, you will not be using interrupts. But the majority of commercial, public domain and shareware DOS hosted communications packages DO NOT use the polled BIOS async service; instead, they substitute their own buffered, interrupt-driven handlers for the sake of improved performance. (The necessity of reinventing this particular wheel in every package ranks as one of the major PC development headaches, but as Ted Kennedy said, that's all water under the bridge.) Therefore it's very possible to test a board's interrupt strapping etc. from DOS. That's not saying that the UNIX V/386 drivers mightn't be pickier than a given DOS comm package. -- "UNIX should be used :: Tom Neff or as an adjective." -- AT&T :: ...uunet!bfmny0!tneff (UUCP only)