Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- (04:38:12 PM) dvchench@gmail.com/EBBEE500: hello. do you have a minute for a quick OS Proj3 question?
- (04:38:19 PM) Yu Li: yes
- (04:43:44 PM) dvchench@gmail.com/EBBEE500: im not sure that this is 100% in the spirit of the assignment, but right now, I process all of the memory requests sequentially in the order that they appear on the input list. is this acceptable or does there need to be concurrency and processes waiting as needed?
- if it is OK to just process them all sequentially in order, that makes the use of semaphores by the disk driver a pointless because my pipes are asynchronous. I am putting in teh semaphores, but the program is built so the disk queue never really has more than one request in it. I know this is not in the spirit of the original assignment, but have you since allowed this kind of simplification? Looking at all the e-mail it seems like by now everyone just does whatever they feel like lol.
- (04:44:31 PM) dvchench@gmail.com/EBBEE500: *i mean my pipes are synchronous ^, that is they block.
- (04:47:08 PM) Yu Li: You cannot get full points if no concurrency
- (04:48:13 PM) Yu Li: I mean concurrency is a grading rule, and deserves some points
- (04:51:51 PM) Yu Li has signed off.
- (04:51:56 PM) Yu Li has signed on.
- (04:52:08 PM) dvchench@gmail.com/EBBEE500: OK. that's fair. so basically we cannot send a request from p1 until its previous request has been acknowleged and processed, but we can queue up reqs for p2,p3?
- (04:52:30 PM) Yu Li: yes
- (04:53:51 PM) dvchench@gmail.com/EBBEE500: so if p1 has a disk access, and p2 has like 5 accesses with things already un frames, then u can have p2 do like 5 ops, while p1 is waiting for the disk to come back?
- (04:54:06 PM) Yu Li: exactly
- (04:54:14 PM) dvchench@gmail.com/EBBEE500: maening that the disk actually needs direct access to the frames[] array?
- (04:54:46 PM) Yu Li: What disk do you mean?
- (04:54:54 PM) Yu Li: paging disk?
- (04:54:56 PM) dvchench@gmail.com/EBBEE500: yes
- (04:55:17 PM) dvchench@gmail.com/EBBEE500: or at least the disk driver...
- (04:55:47 PM) Yu Li: frames[] is maintained by replacement process only
- (04:56:35 PM) dvchench@gmail.com/EBBEE500: then whats the point of putting the mem address inside the disk read requests?
- (04:56:49 PM) dvchench@gmail.com/EBBEE500: instead of jist the disk address
- (04:57:28 PM) Yu Li: ok, disk can load data into main memory, but Frames[] is not main memery
- (04:57:50 PM) Yu Li: Frames[] just record info for main memory usage
- (04:58:34 PM) dvchench@gmail.com/EBBEE500: but since we never deal with actual data payload, that's just extraneous info that is never used?
- (04:59:00 PM) dvchench@gmail.com/EBBEE500: it's jsut to make us aware of a DMA-type thing happening?
- (04:59:03 PM) Yu Li: yes. we have no real hardware. we just simulate
- (04:59:09 PM) Yu Li: yes
- (05:00:57 PM) dvchench@gmail.com/EBBEE500: OK. i think i'm just gonna scrap a lot of what I did before and hopefully make a neat logical project that gets late points off of it but makes perfect sense (hopefully) :-/ . Late policy of -5% per day still applies as before? no cutoff date where you won't accept it?
- (05:01:21 PM) Yu Li: yes
- (05:01:28 PM) dvchench@gmail.com/EBBEE500: thank you
- (05:01:33 PM) Yu Li: welcome
- (05:03:24 PM) Yu Li has signed off.
Add Comment
Please, Sign In to add comment