rob@xyzoom.info.com (Rob Lingelbach) (02/18/91)
I've been trying to compile smail3 for my system running ISC 2.2. I get through building the dependencies, but about 10 minutes into building the binaries, the kernel panics. It never happens at exactly the same place--just generally during heavy disk I/O. I also get some random errors in component files, that don't appear to exist when I examine them. The processor is a 25 MHz 386 w/64k cache, 8 MB RAM, the controller is an Adaptec 1542B with a Maxtor 207 MB drive. I have replaced the controller and the drive, with no change. I've tried adding BIOS wait states to the Adaptec, with no effect. Is anyone else having this problem? Thanks. Rob Lingelbach rob@xyzoom.info.com -or- {well-connected}!uunet!xyzoom!rob KB6CUN Compuserve: 71101,176 2641 Rinconia Dr L.A. CA 90068 voice: 213 464-6266 --------"'Tis pride that pulls the country down" (Shakespeare)-------
larry@nstar.rn.com (Larry Snyder) (02/19/91)
rob@xyzoom.info.com (Rob Lingelbach) writes: >I've been trying to compile smail3 for my system running ISC 2.2. >I get through building the dependencies, but about 10 minutes into >building the binaries, the kernel panics. It never happens at exactly >the same place--just generally during heavy disk I/O. I also get some >random errors in component files, that don't appear to exist when I >examine them. >The processor is a 25 MHz 386 w/64k cache, 8 MB RAM, the controller is >an Adaptec 1542B with a Maxtor 207 MB drive. >I have replaced the controller and the drive, with no change. I've >tried adding BIOS wait states to the Adaptec, with no effect. >Is anyone else having this problem? Thanks. chances are your motherboard isn't compatible with the DMA on the Adaptec 1542B -- Did you run the DMA test under DEBUG (using DOS) on the 1542B? another issue is the DMASPEED and BUSSON/BUSSOFF paramaters in the kernel -- -- Larry Snyder, NSTAR Public Access Unix 219-289-0287 (HST/PEP/V.32/v.42bis) regional UUCP mapping coordinator {larry@nstar.rn.com, ..!uunet!nstar.rn.com!larry}
aland@informix.com (Colonel Panic) (03/01/91)
In article <1991Feb19.134450.6095@nstar.rn.com> larry@nstar.rn.com (Larry Snyder) writes: >... >chances are your motherboard isn't compatible with the DMA on the >Adaptec 1542B -- Did you run the DMA test under DEBUG (using DOS) >on the 1542B? > >another issue is the DMASPEED and BUSSON/BUSSOFF paramaters in >the kernel -- > Larry Snyder, NSTAR Public Access Unix 219-289-0287 (HST/PEP/V.32/v.42bis) Am I *blind*? I've got TFM right in front of me, and the only DMA- or bus-related parm I can find is DMAEXCL... where did you find out about DMASPEED and BUSS*? (or were these pre-2.2?) -- Alan Denney aland@informix.com {pyramid|uunet}!infmx!aland I'm Pro-Anarchy... and I Vote!
larry@nstar.rn.com (Larry Snyder) (03/03/91)
aland@informix.com (Colonel Panic) writes: >Am I *blind*? I've got TFM right in front of me, and the only DMA- or >bus-related parm I can find is DMAEXCL... where did you find out about >DMASPEED and BUSS*? (or were these pre-2.2?) no, your not blind - none if this is covered in the manual - here is part of my file: #define AHA_BUSON 9 /* 5&9 defaults */ #define AHA_BUSOFF 5 /* 10 & 5 were working until recently */ #define AHA_DMASPEED 4 /* 1 for 6.7 mhz, 2 for 8 mhz, 3 for 10 mhz */ /* 4 for 5.7 mhz */ ok, the first two defs are 5 and 9 by default - as you can see I have them reversed which increased disk IO by around 31% here on nstar raising these paramaters too much will cause the floppy drive to overrun next is DMA Speed - I have mine at 4 (5.7mhz) - my other motherboard (a SOYO) supported up to 8 megs (2) without problems but my current 386/33 can't handle anything higher than 5.7 without cause problems when the systems gets very loaded by tweaking these defs and rebuilding the kernel it is possible to make a noticable increase in diskio -- but it is also possible to screw up your disk - - use caution - and make backups -- -- Larry Snyder, NSTAR Public Access Unix 219-289-0287 (HST/PEP/V.32/v.42bis) regional UUCP mapping coordinator {larry@nstar.rn.com, ..!uunet!nstar.rn.com!larry}