#tyndur · 2010-02-15
132 Nachrichten · Seite 1 von 1 · Original-CSV herunterladen
- kwolfMoin
- pcmattmanHi kwolf
- kwolfHi pcmattman
- pcmattmanPlease excuse the internal issues with Pedigree over the past 24 hours :)
- kwolfYou mean eddyb? I'm not that quick to leave a channel ;-)
- pcmattmanWell, he might come here and complain
- pcmattmanOn a different note, I am now implementing the mem.h stuff into Pedigree's CDI module
- kwolfWell, if he comes here, we can have ops
- kwolfI mean, I'm not going to ban him right away now, but if he continues here, I will do so eventually
- pcmattmanSounds good
- kwolfAre you going to implement the real thing for mem.h?
- pcmattmanWhat do you mean, "the real thing"?
- kwolfI started with a very simplistic thing for tyndur. Only allocate complete pages, no scatter/gather, always copy
- pcmattmanRight
- kwolfShould be compliant with the spec, but rather inefficient
- pcmattmanI'll probably avoid scatter/gather initially, but everything else should be implementable from the beginning
- pcmattmanRealistically though scatter/gather isn't too complicated, it's just easier to ignore it for now
- kwolfscatter/gather is an option that you may or may not implement in cdi_mem_alloc when physical memory is requested, but it needs not be contiguous
- kwolfBut in general I think it will get useful only when we start passing mem_areas from external sources
- pcmattmanThe best thing is that our PhysicalMemoryManager is actually designed in such a way that these calls can directly wrap native kernel calls
- kwolfYour PMM can directly handle requests of less than a page?
- pcmattmanOkay, most features ;)
- kwolfHeh, yes, then you end up with my implementation. ;-)
- kwolfThis one: http://list.tyndur.org/pipermail/tyndur-devel/2010-February/001667.html
- pcmattmanHeh, nice
- pcmattmanThese well-designed kernels are great, aren't they? :P
- kwolfSo shouldn't take you too long ;-)
- pcmattmanAlready doing a test compile to see if it builds
- pcmattmanHeh, failed :P
- pcmattman-1 != null, it seems
- kwolfBut it's almost the same ;-)
- kwolfAnyway, I would be suspicious of code that seems to work at the first attempt
- pcmattmanMaybe I should use e1000 or something to make certain the mem functions work
- pcmattman(in qemu)
- kwolfHm, last time I tried (Friday?) no CDI drivers did work in Pedigree
- kwolfNeither e1000 nor pcnet
- pcmattmanWas that before or after the mem.h stuff went into the CDI master?
- kwolfI reverted the submodule to the last commit before mem.h
- kwolfOtherwise it wouldn't even build
- pcmattmanHmm, e1000 is being executed, but doesn't even pick up a device
- pcmattmanAren't regressions just wonderful? :)
- kwolfThey are what makes OS dev interesting :-)
- pcmattmanOh right, I actually disabled the net stuff because I can't figure out how to get a cdi_net_driver in the CDI NIC wrapper
- kwolfHm, I guess you should fix that ;-)
- kwolfIs this what happened with the driver initialisation patches?
- pcmattmanI'd say so
- pcmattmanAh fantastic, I haven't even properly updated most of the code after those patches
- kwolfLooks like we're losing instead of gaining compliant OSes...
- kwolfAnd I don't think a complete implementation does even exist
- pcmattmanI'll work at it once I commit the mem.h stuff
- pcmattmanAt least Pedigree compiles and boots to bash now
- kwolfIf it compiles, boot up shouldn't be too difficult with disabled drivers ;-)
- pcmattmanOkay, pushed that code finally
- pcmattmanWorking through the backlog of things I needed to do last month but never got around to...
- kwolfYup, at some point you just need to do it. But everything CDI related should be pretty straightforward.
- pcmattmanIt is so far
- pcmattmanNow to get these NIC drivers working with the new device initialisation code somehow
- kwolf*lol*
- kwolfI like your code
- kwolf /// \todo Fundamentally wrong
- kwolfCould be mine ;-)
- pcmattmanWell, look at how I implemented it ;)
- pcmattmanIt'll be problematic if someone calls cdi_mem_free on the return value, unless cdi_mem_require_flags makes its argument invalid
- kwolfIt doesn't, so the code is fundamentally wrong in two ways ;-)
- pcmattmanMaybe we should update the documentation of cdi_mem_require_flags
- pcmattman"Can cause its argument to become invalid"
- kwolfNo, you should update your code ;-)
- kwolfDrivers will still need the old buffer
- kwolfBecause the old buffer is where the caller expects the read data in the end
- kwolfFor the function is unused anyway, but I hope to change the interfaces some time this week
- kwolf*For now
- pcmattmanTrue
- pcmattmanIt'll be implemented in due time
- kwolfAnd you always can stop me from doing it by posting your own patches to the interface, so that I'm busy with changing tyndur ;-)
- pcmattmanHahaha
- pcmattmanAlso, you used double quotes in mem.h rather than angle brackets, tsk tsk!
- pcmattman:P
- kwolfYou acked it :-P
- pcmattmanI missed that small detail ;)
- kwolfWill prepare a patch to change the remaining quotes. While at it, we can do it for the drivers, too.
- quokmoin
- kwolf__Moin quok
- XanClicMoin,
- XanClic+quok
- quokI had an interesting morning so far
- quokI have to take sleeping pills at night because I have restless leg syndrome
- quokand in the morning I take pills to keep my blood pressure in check.. it's slightly elevated because I'm a bit overweight (and don't really care to try to correct that situation)
- quokwell, this morning I grabbed the wrong pills.. took the sleeping pills instead of the blood pressure pills XD
- XanClicI guessed that *g*
- kwolf__And you're still awake? ;-)
- quokFor now :/
- quokCalled off work, my boss laughed at me and told me to get some rest
- XanClicxD
- quokinstead I'm going to try to code on quokos (and CDI)
- XanClicHm, I doubt if you enjoy reading that code tomorrow ;-)
- kwolf__Ok, I will make sure that any CDI patches you send today are given a thorough review ;-)
- quokMostly I plan on getting CDI stuff started in QuokOS :P
- kwolf__When I did it, a was acutally disappointed because it was so easy ;-)
- kwolf__-a +I
- kwolf__Getting the standard libc function copied into the kernel was most of the work
- kwolf__Ok, my implementation is far from complete, but ata works
- eddybmoin :)
- XanClicYay :D
- DickHuuhnSo und nu
- DickHuuhnEnglish or German :O
- XanClicBetter English
- DickHuuhnOkay
- DickHuuhnSo, who has won the 512 byte contest now?
- DickHuuhnAnd by the way quok: Creative-Commons-Zero is closer to Public Domain than ISC :)
- DickHuuhnOkay XanClic was now so kind to tell me that this discussion has already been made :P
- XanClicNice, eventually someone here who is capable of speaking good English ;-)
- kwolf__Hm, should quok take it personally? ;-)
- DickHuuhnI learned it with my language-buddies Westerwave and Oettinger so stfu
- XanClickwolf__: Meh ;-)
- XanClicI mean someone who is also in #niwohlos and can thus act as my babelfish
- quokAh, I knew I shouldn't have joined #lost :)
- XanClicBecause it's too fast to use Google? ;-)
- quokI'm trying to not use google :P
- quokbut barely following :)
- kwolf__As long as you're not forced to participate... But DickHuuhn did it right. ;-)
- DickHuuhnWhat did I do? D:
- quokDickHuuhn: nobody has won yet as one of my testing machines died. I'm still trying to find a replacement for it... one of the entries was tested on it but not the other, so I don't think it's fair at this point to declare a winner
- DickHuuhnAh, okay
- quokalthough I could just drop that test, but I'd really rather not :)
- kwolf__DickHuuhn: You asked him a question in German. I think this is exactly what he was afraid of. ;-)
- DickHuuhnOh, sorry, quok :D
- quokyes, I understood the german
- quokjust wasn't at all sure how to phrase my response :)
- kwolf__But answering in English was easier, right ;-)
- quokit was!
- DickHuuhnyay
- quokI just didn't want to answer in English in #lost :)