Return to BSD News archive
Newsgroups: comp.os.386bsd.bugs Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!haven.umd.edu!cs.umd.edu!nmrdc1!rdc30 From: rdc30@nmrdc1.nmrdc.nnmc.navy.mil (LCDR Michael E. Dobson) Subject: Re: [ANSWER for FAQ] IRQ2 vs IRQ9 [was: 3c503 probe] Message-ID: <CA7M1M.ApE@nmrdc1.nmrdc.nnmc.navy.mil> Organization: Naval Medical Research & Development Command References: <221hoo$n4d@terminator.rs.itd.umich.edu> <dvs.163.742733797@ze8.rz.uni-duesseldorf.de> Date: Thu, 15 Jul 1993 14:10:32 GMT Lines: 25 In article <dvs.163.742733797@ze8.rz.uni-duesseldorf.de> dvs@ze8.rz.uni-duesseldorf.de (Wolfgang R. Mueller) writes: >In article <221hoo$n4d@terminator.rs.itd.umich.edu> pauls@terminator.rs.itd.umich.edu (Paul Southworth) writes: >>Anyone know why the stock 3c503 driver thinks that the card is at IRQ 9 >>(and it even works) when IRQ 9 is not even a possible IRQ for that card? >>The GENERICISA conf file says IRQ 2 ..... > >Questions of this type reappear every now and then. >When IBM switched from XT to AT they included a second interrupt controller. >This had to be cascaded, and IRQ 2 on the first one was chosen for this. >From the new IRQs 8 to 15 then IRQ 9 was chosen to be connected to the >temporarily orphaned IRQ 2 line on the bus. So the hardware IRQ 2 line on >the bus activates the software service routine for IRQ 9 ( i.e. interrupt >vector 0x71 ). The standard bios service for example for vector 0x71 is an >int instruction for vector 0x0a ( the one formerly in charge for IRQ 2 ). > >So as a conclusion, in most contexts you need not worry about any >discrepancies between IRQ 2 and IRQ 9. The only context where this is a problem is where there are devices on both IRQ2 and IRQ9. I ran into this with a floppy tape controller (Everex) and my ethernet card (TRW PC2001). I solved it by setting the ethernet card to IRQ11. Now if there was a 386bsd driver for the TRW card .... Mike