Well, heres my 3 cents worth on LPTCOM port addresses and IRQs: The +quot;standard+quot; a

Master Index Current Directory Index Go to SkepticTank Go to Human Rights activist Keith Henson Go to Scientology cult

Skeptic Tank!

Well, here's my 3 cents worth on LPT/COM port addresses and IRQ's: The "standard" addresses for the LPT and COm ports are: COM1 3F8 IRQ4 COM2 2F8 IRQ3 COM3 2E8 IRQ4 COM4 2E0 IRQ3 LPT1 3BC IRQ7 LPT2 378 IRQ7 LPT3 278 IRQ5 However,.... I have also seen LPT2 use IRQ5 and LPT3 use IRQ7!!! COM3 and COM4 addresses are not really that "standard", and COM4 especially may vary depending on your board. I believe that the using IRQ7 for all LPT ports in a system works fine, LPT ports seem to share IRQ's just fine. This is not nearly as true for COM ports, although in most cases the pairs COM1/3 {IRQ4) and COM2/4 (IRQ3) work fine. Some COM boards/modems are just not happy sharing an IRQ with another COM port/modem!! If you have trouble with a COm port in a machine with more than two ports, try disabling/removing the other member of the pair and see if the problem goes away. I have also seen some software that is hardcoded to expect the above address/IRQ "standard" settings, even though most CO{ boards allow you to m{x and match (say COM2 using IRQ 2,4 or 5)!! Unless you are setup for non-standard IRQ #'s to avoid some other IRQ conflict, it's best to stick to the "standards". For an explaination of when COM1 isn't COM1 see the message I will post titled COM/LPT ports and their addresses. It explains how the BIOS determines how to assign the COM1/2/3/4 & LPT1/2/3 device names to a hardware port at a particular address. ------ One of the most confusing things about how PC hardware is configured has to do with the way the BIOS assigns device names (COM1, COM2, etc) to a hardware port. Most users think that by addressing their serial board at address 2F8, that they have set that port up as COM2. This is not always true!!! If this poor user didn't have a serial port at address 3F8 (COM1's "standard" address), then their port at 2F8 will be given the device name COM1!!!!!! How can this be??? Well, let me explain. When your machine boots and the BIOS does its initialization magic, it goes out to the hardware of your system, and checks certain addresses (in a fixed order) for serial ports and parallel ports. The first serial port it finds is given the DOS device name "COM1". The second is given the name "COM2", and so on for up to 4 serial ports. Parallel ports are given device names in the same way. The first port found gets the device name "LPT1" and so on. Serial ports are searched for at addresses (in this order!): 3F8 2F8 2E8 2E0 Parallel ports are searched for at (in this order): 3BC 378 278 As I said before, the first port of each type found gets the first device name (COM1 or LPT1), the second gets the second (COM2 or LPT2), and so on. Thus in the simple example above, since the first address in the serial port list that the BIOS finds an actual serial port at is 2F8 (the "standard" COM2 address), that port gets the device name COM1 !!! The only reason I have figured out for this, is so that if a user only has one serial port on their machine, they can just call it COM1, they don't have to know hosw it's addressed. This causes problems for some dumb software that assumes that serial/parallel ports use the "standard" IRQ (interrupt request) assignments. In the example, the user has a serial port that looks to DOS like it's COM1, but the IRQ is probably comfigured to COM2's "Standard" IRQ (IRQ. If a dumb piece of software says, "Gee, I'm using COM1, so it must be using IRQ4. I'll setup~r the IRQ4 interrupt vector so I'll get the interrupts for that po!!" The problem is that the software will NEVER see the ports interrupt (it is actually using IRQ3)!!! No interrupts = No data transfer!!!!! Most programs (TELIX, PROCOMM, etc.) allow the user to change the port address/IRQ information (for that program only!), so that they can correctly access the serial ports. NOTE: Parallel ports don't seem to have these sorts of problems.


E-Mail Fredric L. Rice / The Skeptic Tank