Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cornell!uw-beaver!tektronix!pogo!orca!tekecs!bruce From: bruce@tigger.wv.tek.com (Bruce Robertson;685-2912;;;tigger) Newsgroups: comp.arch Subject: Re: Unaligned Accesses (was Re: How to use silicon) Message-ID: Date: 31 Mar 89 17:44:46 GMT References: <37196@bbn.COM> <1989Mar16.190043.23227@utzoo.uucp> <24889@amdcad.AMD.COM> <355@bnr-fos.UUCP> <13@microsoft.UUCP> <362@bnr-fos.UUCP> <59@microsoft.UUCP> <343@unicads.UUCP> <25000@amdcad.AMD.COM> <1989Mar29.232103.68 Sender: nobody@tekecs.GWD.TEK.COM Organization: Tektronix ITD, Wilsonville, OR Lines: 20 In article bruce@tigger.wv.tek.com (Bruce Robertson;685-2912;;;tigger) writes: >... nothing says you can't write a utility to reorder your >structure members at the C source level, optimizing for a particular >architecture. Obviously, you would have to avoid using this utility >if the order of structure elements matters (you wouldn't want to do it >to the Unix kernel!) In article <1989Mar29.232103.6835@utzoo.uucp> henry@utzoo.uucp (Henry Spencer) writes: > If you look at the kernel sources, especially the older parts, you'll > find that there is a remarkable tendency for (e.g.) "char" members of > structs to be grouped together so they will pack densely. What I was thinking of is things like structure definitions of stack frames, device registers, etc. -- Bruce Robertson Domain: bruce@tigger.wv.tek.com UUCP: tektronix!tekecs!tigger!bruce