Return to BSD News archive
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!spool.mu.edu!nigel.msen.com!math.fu-berlin.de!irz401!uriah!not-for-mail From: j@bonnie.tcd-dresden.de (J Wunsch) Newsgroups: comp.os.386bsd.bugs Subject: Re: NMI port 61 a0, Port 70 ff Error! Date: 9 Aug 1993 15:27:23 +0200 Organization: Textil Computer Design GmbH Dresden, Germany Lines: 41 Message-ID: <245jbrINNrb1@bonnie.tcd-dresden.de> References: <242iro$n9i@jadzia.CSOS.ORST.EDU> <1993Aug8.162909.16863@sophia.smith.edu> NNTP-Posting-Host: bonnie.tcd-dresden.de In <1993Aug8.162909.16863@sophia.smith.edu> jfieber@sophia.smith.edu (J Fieber) writes: >In article <242iro$n9i@jadzia.CSOS.ORST.EDU> cntrline@CSOS.ORST.EDU (Centerline Computers) writes: >> >>Usually, I get that "NMI port 61 a0, port 70 ff" message... Sometimes >>I noticed I got a "Bus error". Note that in 386BSD, NMI handling is a broken thing. Obviously, Bill had a bad idea (and a wrong description) of the PC hardware. Ports 0x61 and 0x70 are not much informational for the reason of an NMI. Instead, port 0x62 (PPI port C) should be queried, with 0x80 bit set => (motherboard) RAM parity error detected, 0x40 => I/O channel check signal (NMI requested by ISA device). The handling is very inconsistent. For my experience, IOCHCK signals always cause the kernel to panic and dump core (this is a nice thing to debug a hung kernel). RAM parity errors sometimes dump core, but are sometimes (when occuring in user mode?) only syslogged. However, in the latter case, the affected pages aren't marked invalid. That's why the program gets a bus error signalled. (Due to the broken byte, the program's flow becomes corrupted.) But the page is still valid, so if the program is started again, its pages are found in-core and re-used. Causing the same trouble... >Bad memory. [the guy requested his vendor to replace the RAM] If you can't get your vendor to replace them: . Try other RAM timings (wait states) . Always use identical SIMM's within one bank. . Avoid intermixing 9-chip and 3-chip SIMM's. . Avoid page interleave between SIMM's of different vendors. > (NMI can pop up under >MS-DOG as well so tech support might even believe you) But sometimes it would occur only in protected mode (its requirements appear to be harder). You can reference protected mode under DOG via a RAM disk. -- in real life: J"org Wunsch | ) o o | primary: joerg_wunsch@tcd-dresden.de above 1.8 MHz: DL 8 DTL | ) | | private: joerg_wunsch@uriah.sax.de | . * ) == | ``An elephant is a mouse with an operating system.''