2008-09-23 01:55 folks 2008-09-23 02:41 -!- tim_dimm_(~mobile@cpe-76-90-98-247.socal.res.rr.com) has joined #tux3 2008-09-23 02:42 Greetings 2008-09-23 03:17 -!- konrad(~konrad@D-128-208-53-208.dhcp4.washington.edu) has joined #tux3 2008-09-23 04:49 -!- kbingham(~kbingham@92.10.168.77) has joined #tux3 2008-09-23 05:43 -!- Bobby(~Bobby@nat-inn.mentorg.com) has joined #tux3 2008-09-23 05:44 hellooo 2008-09-23 05:44 anyone here mind explaining extents to me??? 2008-09-23 05:45 -!- pranith(ca4bcee2@webchat.mibbit.com) has joined #tux3 2008-09-23 06:14 -!- smitht(~chatzilla@ool-182f94db.dyn.optonline.net) has joined #tux3 2008-09-23 06:16 Hello all, I am eager to look over the kernel port of Tux3, however I cannot get git to clone the repo at http://phunq.net/ddtree, has anyone done this successfully? Thanks. Trevor 2008-09-23 09:50 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #tux3 2008-09-23 10:45 -!- tim_dimm(~timothyhu@cpe-76-90-98-247.socal.res.rr.com) has joined #tux3 2008-09-23 10:48 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #tux3 2008-09-23 11:58 "Results 1 - 10 of about 413,000 for tux3" 2008-09-23 12:12 maze, there? 2008-09-23 12:12 yes ;-( 2008-09-23 12:12 your suggestion re putting the xattrs up high in files 2008-09-23 12:13 hmm 2008-09-23 12:13 really a good idea, except for the problem of deepening the radix tree 2008-09-23 12:13 I'd think that would be relativelly insiginificant 2008-09-23 12:13 every file with xattrs would have a 6 level radix tree 2008-09-23 12:13 its significant I think 2008-09-23 12:14 compared to 1 level for most files now 2008-09-23 12:14 however 2008-09-23 12:14 if the radix tree code were to be modified to have one level of btree at the top level... 2008-09-23 12:14 perhaps optionally 2008-09-23 12:14 then it gets practical 2008-09-23 12:14 I'm not quite sure why it would get so deep so quickly? 2008-09-23 12:15 its a radix tree 2008-09-23 12:15 if you map something at the top of the space, the entire tree has to deepen 2008-09-23 12:15 you could use signed 2008-09-23 12:15 ah 2008-09-23 12:15 but 2008-09-23 12:15 well, with an offset 2008-09-23 12:16 so the only cost is to look up and maintain that offset 2008-09-23 12:16 yes, that improves it without much stress 2008-09-23 12:16 good idea 2008-09-23 12:16 I guess, my problem is, I'm not quite sure what a radix tree is 2008-09-23 12:16 well... you can take a run at your first core kernel hack then, after we have tux3's requirements to justify it 2008-09-23 12:17 ah 2008-09-23 12:17 better clear that up 2008-09-23 12:17 I'd assumed you'd be using a normal number of leaves determines depth type of tree here 2008-09-23 12:17 fundamental tool of software engineering 2008-09-23 12:17 is a radix tree what the cpu uses for tlb? 2008-09-23 12:17 radix tree is to btree as bucket sort is to quicksort 2008-09-23 12:17 sorta 2008-09-23 12:17 radix tree is directly indexed at each level instead of binsearched 2008-09-23 12:18 therefore needs no keys 2008-09-23 12:18 index nodes are twice as compact 2008-09-23 12:18 right, so isn't the cpu virt to phys mapping a radix tree? 2008-09-23 12:18 probe is much faster 2008-09-23 12:18 but you still pay a l1 cache pressure penalty for each level of the tree 2008-09-23 12:18 it is, and that is a problem in some cases 2008-09-23 12:18 64 bit machines struggle with it 2008-09-23 12:19 ok, so I just wan't aware of the name 'radix tree' 2008-09-23 12:19 no other name for it that I know 2008-09-23 12:19 like radix sort 2008-09-23 12:19 wasn't aware of any name ;-) 2008-09-23 12:19 you were looking for some reason to split something into two for 32-bit machines 2008-09-23 12:19 can't remember what it was... something about 1EB of space? 2008-09-23 12:20 was that for total fs? 2008-09-23 12:20 hmm, must have been 2008-09-23 12:20 I wasn't either until "wind" showed up on #kernelnewbies with his plan of changing the page cache from a hash to something better 2008-09-23 12:20 I recall he considered half a dozen types of trees I'd never heard of 2008-09-23 12:20 he? 2008-09-23 12:20 the one thing he was sure of, all of them would be better than a hash 2008-09-23 12:20 he was right 2008-09-23 12:20 John Levon 2008-09-23 12:20 h 2008-09-23 12:21 ah 2008-09-23 12:21 after that, showed very little interest in Linux 2008-09-23 12:21 not sure why 2008-09-23 12:21 heading out with team for Lunch. Will be back in 30 minutes. 2008-09-23 12:21 bye 2008-09-23 12:21 ;-) 2008-09-23 12:48 -!- data(~data@echo489.server4you.de) has joined #tux3 2008-09-23 12:51 -!- data(~data@echo489.server4you.de) has joined #tux3 2008-09-23 12:55 -!- pgquiles(~pgquiles@42.Red-83-39-60.dynamicIP.rima-tde.net) has joined #tux3 2008-09-23 13:12 back 2008-09-23 13:23 crawl-8.cuill.com ;-) 2008-09-23 13:23 that was fast 2008-09-23 13:52 folks 2008-09-23 14:10 flipz: talking about implementing xattrs ? 2008-09-23 14:10 already ahve 2008-09-23 14:10 you should check out the code 2008-09-23 14:19 yeah, been looking at it 2008-09-23 14:19 and thinking about the allocation map problem a bit 2008-09-23 14:20 not sure what kind of tree structure to use to represent areas on the disk that might have a certain amount of free blocks 2008-09-23 14:27 it's a bitmap 2008-09-23 14:27 not a tree 2008-09-23 14:27 there will be accelerator bits in the pointers to bitmap blocks eventually 2008-09-23 14:27 to know which blocks have how much space free 2008-09-23 14:27 for now it's just a big linear block map 2008-09-23 14:34 -!- RazvanM(~RazvanM@dazzler.isi.jhu.edu) has joined #tux3 2008-09-23 15:22 big linear scan then ? 2008-09-23 15:37 for now 2008-09-23 15:37 it certainly won't stay that way 2008-09-23 15:37 even now 2008-09-23 15:37 the scan is directed 2008-09-23 15:37 to a preferred target area 2008-09-23 18:11 -!- tim_dimm(~timothyhu@cpe-76-90-98-247.socal.res.rr.com) has joined #tux3 2008-09-23 18:35 -!- ajonat(~ajonat@190.48.115.242) has joined #tux3 2008-09-23 19:16 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #tux3 2008-09-23 19:52 -!- RalucaM(~ral@scout-9.cnds.jhu.edu) has joined #tux3 2008-09-23 19:56 ACTION is working for a deadline for tomorrow so it will not be present at today's lesson :-( 2008-09-23 20:00 :-( 2008-09-23 20:08 -!- ajonat(~ajonat@190.48.119.175) has joined #tux3 2008-09-23 20:22 -!- Kirantpatil(~kiran@122.167.199.68) has joined #tux3 2008-09-23 20:22 -!- Kirantpatil(~kiran@122.167.199.68) has left #tux3 2008-09-23 20:29 no session at all today? 2008-09-23 20:29 teacher skipped class 2008-09-23 20:30 looks like it 2008-09-23 20:37 -!- macan(~chatzilla@159.226.41.129) has joined #tux3 2008-09-23 22:40 -!- konrad(~konrad@D-128-208-53-208.dhcp4.washington.edu) has joined #tux3 2008-09-23 22:47 -!- tim_dimm_(~mobile@166.134.66.229) has joined #tux3 2008-09-23 23:15 -!- RazvanM(~RazvanM@pool-151-196-118-156.balt.east.verizon.net) has joined #tux3 2008-09-23 23:26 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #tux3 2008-09-23 23:35 -!- nataliep(~nataliep@207.47.98.129.static.nextweb.net) has joined #tux3 2008-09-23 23:38 -!- ajonat(~ajonat@190.48.127.55) has joined #tux3 2008-09-23 23:46 flake 2008-09-23 23:46 flipz: = flake 2008-09-23 23:59 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #tux3