home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.arch
- Path: sparky!uunet!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!ira.uka.de!Germany.EU.net!news.uni-bielefeld.de!rz.ruhr-uni-bochum.de!jan
- From: jan@vesta.neuroinformatik.ruhr-uni-bochum.de (Jan Vorbrueggen)
- Subject: Re: Known hardware bugs.
- Message-ID: <JAN.93Jan21143827@vesta.neuroinformatik.ruhr-uni-bochum.de>
- In-reply-to: ferriere@gr.osf.org's message of 20 Jan 93 15:09:08 GMT
- Organization: Inst. f. Neuroinformatik, Ruhr-Universitaet Bochum, FRG
- References: <93-01-147@comp.compilers>
- Date: 21 Jan 93 14:38:27
- Lines: 20
-
- In article <9301147@comp.compilers> ferriere@gr.osf.org
- (Francois de Ferriere) writes:
-
- I'm looking for known hardware bugs.
-
- Do microcode bugs count as well?
- If so, then there was (and I expect there still is) a nice one in the
- implementation of the REI instruction on the VAX11/780. Briefly, when
- returning from an AST, which is (among other things) done by REI, it is
- confused when just at that time another AST is queued, and restores wrong
- stack pointers into some registers (disclaimer all from ageold memory...).
- This bug lay hidden for at least five years, until a lot of people suddenly
- had unexplained and irreproducible crashed in standalone BACKUP, which works
- almost entirely with ASTs and exercises the system...The fix was to add a
- sanity check to the AST delivery code that would readjust one of pointers
- if it contains an "impossible" value to something valid. Maybe somebody
- with access to the VMS source CD could check whether that fix is still in?
- How about OpenVMS on AXP :)??
-
- Jan
-