OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

kmip message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: Re: [kmip] Groups - Sun 32-bit Binary Alignment Proposal,v2 (Sun 32-bit Binary Alignment Proposal.pdf) uploaded

Glen Jaquette wrote:
OF08F261E1.8089946B-ON072575A8.0015C5E6-072575A8.00160B97@us.ibm.com" type="cite">
        Your proposed changes do make the TTLV protocol 4 byte aligned.  I would just observe that with one additional constraint, i.e. if you forced the Value field to be 4 + N*8 bytes long, where N is a whole number, then it would also be 8-byte alighted for 64-bit processing.

Hi Glen,

That's true.  However, if we want to go for full 64-bit support, I'd prefer that we rearrange the layout so that the Tag, Type, and Length fields are within one 64-bit word.  That way the Value field could just be a simple multiple of 8 bytes, instead of 4 plus a multiple of 8.  Combining Tag and Type into one field would probably be more efficient because generally a particular Tag only has one supported Type, so the whole thing just becomes a 32-bit 'Super Tag' that can be quickly referenced in a switch statement or lookup table.

If we're interested in doing 64-bits 'right' instead of 'fast', I'd recommend we follow the sketch proposal I sent to Jon Callas yesterday.  That said, the Sun requirement is simply for optimal processing on 32-bit ARM processors, so we would approve any of these approaches.


fn:Matt Ball
org:Sun Microsystems, Inc.;Key Management
adr:;;500 El Dorado Blvd, Bldg 5;Broomfield;CO;80021;U.S.A.
title:Staff Engineer

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]