============================================================ -----Original Message----- From: "Craddock, Chris" <[EMAIL PROTECTED]> Sent: 7/6/2006 12:57 PM To: "[email protected]" <[email protected]> Subject: Re: Help with TCB status
Ed (and Jeff Smith) asked > Are you sure PLO can be used this way? Will the "compare" portion work > against arbitrary values not serialized/updated using PLO? Well now that you mention it, I'm NOT sure. The compare and load functions are supposed to be for following chains that can change while you're looking at them. I had a conversation about PLO with Bob Rogers a long time ago and came away with the impression that the operands were fetched (in the normal way) -before- the comparison was made, so if you got CC=0 then, at least in the compare and load case, you had a point-in-time consistent view which is all that I wanted. I haven't actually used this "trick" anywhere and since it might not actually work the way I thought, I won't use it. I can't blame Bob because I was probably reading between the lines of what he said anyway. CC ============================================================ Greetings, It's quite clear from reading the PoPs that the COMPARE is separated in time from the LOAD. The only thing that makes the whole operation look atomic is holding the hardware lock. So, all CPU must agree on the PLO lock token (R1) and only use PLO for reading or writing (the standard COMPARE-SWAP instructions won't interoperate with PLO). Jeffrey D. Smith Farsight Systems Corporation 24 BURLINGTON DR LONGMONT, CO 80501 303-774-9381 direct 303-709-8153 cell 303-484-6170 fax ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
