2008-09-24 00:10 -!- shapor kicked bh ("be nice") 2008-09-24 00:25 -!- pranith(ca4bcee2@webchat.mibbit.com) has joined #tux3 2008-09-24 00:25 hello 2008-09-24 00:45 hi pranith 2008-09-24 00:46 sleep time... 2008-09-24 01:17 -!- pgquiles(~pgquiles@42.Red-83-39-60.dynamicIP.rima-tde.net) has joined #tux3 2008-09-24 01:26 -!- hirofumi(~hirofumi@210.171.168.39) has joined #tux3 2008-09-24 01:32 -!- hirofumi(~hirofumi@210.171.168.39) has joined #tux3 2008-09-24 01:52 -!- MaZe(~MaZe@c-24-6-86-168.hsd1.ca.comcast.net) has joined #tux3 2008-09-24 01:58 hello 2008-09-24 01:58 anyone here help me on setting up a fuse tux3? 2008-09-24 01:59 MaZe: ? 2008-09-24 01:59 hmm 2008-09-24 02:00 ? 2008-09-24 02:00 oh, tux3 fuse... I'm probably not the best person, seeing as I haven't tried running it yet 2008-09-24 02:00 hmm, ok 2008-09-24 02:00 I'm mucking around with the kernel and haven't compiled tux3 yet even... still working on (a) learning about the kernel, (b) writing options parsing, and (c) getting a solid debug environment, and most importantly... I have work ;-( 2008-09-24 02:01 oh.. 2008-09-24 02:01 speaking of options parsing... 2008-09-24 02:01 flipz: -o mkfs ofcourse, but can also potentially do -o resize=#blocks 2008-09-24 02:01 hmm 2008-09-24 02:01 pranith: sure 2008-09-24 02:02 i can help 2008-09-24 02:02 hey shapor 2008-09-24 02:02 cool 2008-09-24 02:02 flipz: and support that as a -o remount option as well 2008-09-24 02:02 MaZe: which code are u working on? 2008-09-24 02:02 will have to look see how remount is implemented 2008-09-24 02:03 right now? generic kernel space options parser 2008-09-24 02:03 shapor: what do i need to do? 2008-09-24 02:03 well tux3fs since ls works with it 2008-09-24 02:03 readdir doesn't work in tux3fuse 2008-09-24 02:03 MaZe: ok 2008-09-24 02:03 well, at least not for me, i was going to try a new version of fuse 2008-09-24 02:03 it seems to be returning some data but somethings not right 2008-09-24 02:03 shapor: hmm. where can i get the code? and any install options? 2008-09-24 02:04 tux3fuse is the low level api one 2008-09-24 02:04 its in the repo 2008-09-24 02:04 user/test 2008-09-24 02:04 right next to all the other files 2008-09-24 02:04 ok 2008-09-24 02:05 shapor: can i have the link please... 2008-09-24 02:05 :) 2008-09-24 02:05 its in the mercurial repository at http://tux3.org/tux3 2008-09-24 02:06 hg clone http://tux3.org/tux3 2008-09-24 02:06 then it will be in tux3/user/test 2008-09-24 02:06 just run "make makefs" 2008-09-24 02:07 and "make debug" 2008-09-24 02:07 and it should be mounted on /tmp/test 2008-09-24 02:07 ok 2008-09-24 02:44 anyone mind explaining something about extents to me ? :) 2008-09-24 02:45 i want to know how they help in addressing a larger disk area 2008-09-24 02:45 shapor: ? 2008-09-24 02:45 MaZe: ? 2008-09-24 02:45 hmm? 2008-09-24 02:45 any idea on extents? 2008-09-24 02:45 they don't really... they're really just a performance optimization 2008-09-24 02:45 instead of splitting a file into blocks 2008-09-24 02:46 and then storing the location of every single block 2008-09-24 02:46 ok, u map a chunk of block to a single extent... 2008-09-24 02:46 you split the file into linear sequence of blocks (linear in the sense they are ordered sequentially one after the other on disk) 2008-09-24 02:46 this way you only need to store a mapping (file blocks N..M) -> (disk block X..Y) 2008-09-24 02:47 hmm, ok 2008-09-24 02:47 or indeed just a map of [N] -> [X] is enough (since M and Y are -1 of the next set) 2008-09-24 02:47 thus you have a file as a a set of extents (linear set of blocks), instead of as a set of blocks 2008-09-24 02:47 hmm, nice 2008-09-24 02:48 since you want files to be linear as much as possible (and thus contain few extents) [hence running defragmentors, etc in windows] 2008-09-24 02:48 you will usually end up with relatively few extents, and thus it takes less space to store and can have better performance (especially if well implemented) than just a naive block list 2008-09-24 02:50 basically you get both space savings on disk (and in memory), and better performance, due to having/needing to read in fewer disk blocks (which have a tendency to get pretty randomly distributed) than in a block based fs 2008-09-24 02:51 hmm, thats the main advantage then 2008-09-24 02:51 not addressing a large disk area 2008-09-24 02:56 shapor: problem with fuse :( 2008-09-24 02:56 pranith: whats the problem 2008-09-24 02:57 permission denied 2008-09-24 02:57 doing what 2008-09-24 02:57 cd /tmp/test 2008-09-24 02:58 there are no permissions on the test directory 2008-09-24 02:58 it just say 'd?????' on a ls -l 2008-09-24 02:58 btw, make debug did not return 2008-09-24 02:59 to the command prompt 2008-09-24 03:02 oh i've seen that before 2008-09-24 03:02 hmm 2008-09-24 03:02 are you running ls as root? 2008-09-24 03:02 what do i do? 2008-09-24 03:02 nope 2008-09-24 03:02 try that 2008-09-24 03:02 as a normal user 2008-09-24 03:02 ok 2008-09-24 03:02 try root 2008-09-24 03:03 ohk, got the permissions now as root 2008-09-24 03:03 but... why? 2008-09-24 03:03 because the fuse implementation is *very* rough around the edges 2008-09-24 03:03 shouldn't fuse be accessible as a normal user 2008-09-24 03:04 yes, patches welcome :) 2008-09-24 03:04 :) 2008-09-24 03:04 if you hit ctrl-c 2008-09-24 03:04 and re run it as a normal user 2008-09-24 03:04 ./tux3fs /tmp/testdev /tmp/test -f 2008-09-24 03:04 instead of make debug 2008-09-24 03:04 i think it will work 2008-09-24 03:04 hmm 2008-09-24 03:05 i dint run make debug as root... 2008-09-24 03:05 but it has a sudo command in it 2008-09-24 03:05 oh 2008-09-24 03:05 :) 2008-09-24 03:05 the fuse implementation is really just meant as a test harness 2008-09-24 03:05 we know there are a lot of bugs in it 2008-09-24 03:05 ok 2008-09-24 03:06 i was trying to figure out why readdir isn't working right in tux3fuse (the low level one) 2008-09-24 03:06 oh 2008-09-24 03:06 since i think porting to that will make the kernel port a little easier 2008-09-24 03:06 since the api is more vfs-ish 2008-09-24 03:06 remains to be seen, i havne't had a lot of time to work on it recently 2008-09-24 03:07 you part time work on tux3?? 2008-09-24 03:08 wow! tux3 is gplv3!! 2008-09-24 03:08 how do u get it into the kernel? 2008-09-24 03:15 i think it will be v2 in kernel 2008-09-24 03:15 i work nights and weekends when i have time 2008-09-24 03:16 its not my day job ;) 2008-09-24 03:16 its not anyones day job afaik 2008-09-24 03:17 hmm, what does flips do? 2008-09-24 03:18 i mean what does he do for a day job :D 2008-09-24 03:18 i know he develops tux3... 2008-09-24 03:21 i think hes been mostly working on tux3 recently 2008-09-24 03:33 pranith: regarding gplv2 vs 3. This is solved by the remark 2008-09-24 03:33 * By contributing changes to this file you grant the original copyright holder 2008-09-24 03:33 * the right to distribute those changes under any license. 2008-09-24 03:41 :) 2008-09-24 03:41 -!- MaZe(~MaZe@c-24-6-86-168.hsd1.ca.comcast.net) has joined #tux3 2008-09-24 03:41 its just what reiser did :D 2008-09-24 03:41 hope flips doesnt flip out like reiser :P 2008-09-24 03:52 -!- pgquiles(~pgquiles@42.Red-83-39-60.dynamicIP.rima-tde.net) has joined #tux3 2008-09-24 04:37 what is tux3fs.c and tux3fuse.c? 2008-09-24 04:37 any difference between the two? 2008-09-24 04:41 tux3fuse uses the low level fuse api 2008-09-24 04:41 and readdir is currently broken 2008-09-24 04:49 yeah 2008-09-24 04:49 ok 2008-09-24 04:56 any idea on how i can debug this code?? 2008-09-24 04:56 printf's are nice.. but gdb rocks 2008-09-24 04:56 :D 2008-09-24 05:03 what does printf("'%.*s'", namelen, name) do? 2008-09-24 05:03 im not sure of this printf specifier :( 2008-09-24 05:58 hello 2008-09-24 05:59 anybody here? 2008-09-24 07:22 pranith: A field width or precision, or both, may be indicated by an asterisk `*' or an asterisk followed by one or more decimal digits and a `$' instead of a digit string. In this case, an int argument supplies the field width or precision. A negative field width is treated as a left adjustment flag followed by a positive field width; a negative precision is treated as though it were missing. If a single format directi 2008-09-24 07:22 (from man 3 printf) 2008-09-24 07:23 (on mac :P) 2008-09-24 07:23 RzM|Away: thanks :) 2008-09-24 07:23 RzM|Away: mac is lame 2008-09-24 07:23 :P 2008-09-24 07:23 so the namelen will tell how much of the name to show :D 2008-09-24 07:23 hmm, roger that 2008-09-24 07:24 The field width 2008-09-24 07:24 An optional decimal digit string (with nonzero first digit) specifying a minimum field width. If the converted value has fewer characters than the field width, it will be padded with spaces on the left (or right, if the left-adjustment flag has been given). Instead of a decimal digit string one may write `*' or `*m$' (for some decimal integer m) to specify that the field width is given in the next argument, or in the 2008-09-24 07:24 from linux 2008-09-24 07:24 I use both ;-) 2008-09-24 07:25 and like both :P 2008-09-24 07:25 hmm 2008-09-24 07:25 i use ubuntu dressed up as mac 2008-09-24 07:25 so i have the best of both worlds 2008-09-24 07:25 :D 2008-09-24 07:25 ;-) 2008-09-24 07:26 any idea why readdir fails in tux3fuse? 2008-09-24 07:26 check this out http://xkcd.com/424/ 2008-09-24 07:26 I didn't have a chance to try that :( 2008-09-24 07:27 got to go 2008-09-24 07:27 hmm, xkcd.. lol 2008-09-24 07:27 have fun 2008-09-24 07:27 u too 2008-09-24 07:27 bbye 2008-09-24 07:57 -!- pgquiles_(~pgquiles@62.43.226.52.static.user.ono.com) has joined #tux3 2008-09-24 08:10 -!- kbingham(~kbingham@92.22.1.228) has joined #tux3 2008-09-24 08:10 flipz: -o mkfs ofcourse, but can also potentially do -o resize=#blocks <- ah yes 2008-09-24 08:20 MaZe, and for that matter, -o remount,resize=#blocks 2008-09-24 08:21 maze, see super_operations->remount_fs 2008-09-24 08:21 there we go, that will have to do for a tux3 U session this time 2008-09-24 08:45 -!- flips(~phillips@phunq.net) has joined #tux3 2008-09-24 08:48 tim_dimm, dcc chat doesn't work with my connection 2008-09-24 08:48 probably have to configure my router or something 2008-09-24 08:49 I'm timing out on the other connection 2008-09-24 08:50 that's because speakeasy went down 2008-09-24 08:50 just reconnect 2008-09-24 08:50 can u do a 11am call friday ? 2008-09-24 08:50 ACTION points at the query chat 2008-09-24 09:45 -!- pgquiles__(~pgquiles@42.Red-83-39-60.dynamicIP.rima-tde.net) has joined #tux3 2008-09-24 10:08 -!- RazvanM(~RazvanM@dazzler.isi.jhu.edu) has joined #tux3 2008-09-24 10:51 -!- ceatinge(~ceatinge@veryclever.net) has left #tux3 2008-09-24 11:08 -!- ceatinge(~ceatinge@72.232.13.50) has joined #tux3 2008-09-24 11:32 -!- pgquiles_(~pgquiles@62.43.226.52.static.user.ono.com) has joined #tux3 2008-09-24 11:40 -!- pranith(7aa040b1@webchat.mibbit.com) has joined #tux3 2008-09-24 11:40 heya 2008-09-24 11:40 hi 2008-09-24 11:40 anyone here? 2008-09-24 11:40 hey flips 2008-09-24 11:40 nope 2008-09-24 11:41 was going through the fuse code today... 2008-09-24 11:41 whats wrong with the readdir function? 2008-09-24 11:41 I haven't looked at it 2008-09-24 11:41 shapor started to look at it 2008-09-24 11:41 getting it working in tux3fs was tricky 2008-09-24 11:41 ohk 2008-09-24 11:41 hmm 2008-09-24 11:42 the readdir internal interface is super crappy 2008-09-24 11:42 tux3fuse.c and tux3fs.c are different... 2008-09-24 11:42 on of the worst interfaces anywhere, for anything 2008-09-24 11:42 that's right 2008-09-24 11:42 is it because fuse uses the fuse api? 2008-09-24 11:42 or something like that? 2008-09-24 11:42 no idea 2008-09-24 11:42 hmm, ok 2008-09-24 11:43 you might try emailing tero 2008-09-24 11:43 who checked in the original tux3fuse patch 2008-09-24 11:43 hmm, ok 2008-09-24 11:43 cc tux3 list if you do please 2008-09-24 11:43 yeah, sure 2008-09-24 11:43 :) 2008-09-24 11:43 you can also bug shapor 2008-09-24 11:43 if you like 2008-09-24 11:43 shapor is fun to bug 2008-09-24 11:44 hehe 2008-09-24 11:44 shapor is busy with his day job 2008-09-24 11:44 he told me 2008-09-24 11:44 so better to try tero for that :) 2008-09-24 11:44 "busy" 2008-09-24 11:44 it's all relative 2008-09-24 11:45 hmm 2008-09-24 11:45 can i have the encrypted email id of tero? 2008-09-24 11:47 ok, got it 2008-09-24 12:08 -!- pranith(7aa040b1@webchat.mibbit.com) has joined #tux3 2008-09-24 12:17 debugging of filemap extents finally begins 2008-09-24 12:17 was hard code to write 2008-09-24 12:18 I found it hard 2008-09-24 12:18 prolly easy for shapor though ;) 2008-09-24 12:36 irob's thoughtful initializing of buffers to "deadly data" has the unintended side effect of preventing valgrind from detecting access to unitialized buffer data 2008-09-24 12:37 ACTION removes 2008-09-24 12:37 ah, there we go, lots of valgrind complaints 2008-09-24 12:38 -!- ceatinge(~ceatinge@72.232.13.50) has joined #tux3 2008-09-24 13:42 -!- pgquiles__(~pgquiles@62.43.226.52.static.user.ono.com) has joined #tux3 2008-09-24 14:26 -!- bh(~billh@ip68-107-26-122.sd.sd.cox.net) has joined #tux3 2008-09-24 14:26 folks :) 2008-09-24 14:57 -!- pgquiles_(~pgquiles@42.Red-83-39-60.dynamicIP.rima-tde.net) has joined #tux3 2008-09-24 15:46 hmm, linux-fsdevel server is slower than molasses in January 2008-09-24 15:47 last time I post to it without ccing lkml, I think 2008-09-24 15:47 flips: sk8 30 for new dads 2008-09-24 15:47 still there? 2008-09-24 15:47 rolling out now 2008-09-24 15:47 I'll head out in about 15 2008-09-24 15:47 meet at the pier? 2008-09-24 15:48 got to home by 5 2008-09-24 15:48 pier-ish 2008-09-24 15:48 should work 2008-09-24 15:48 k 2008-09-24 15:48 rollin' 2008-09-24 15:48 cu 2008-09-24 15:48 no crashes 2008-09-24 15:48 certainly no blue screen of death 2008-09-24 15:49 ;-) 2008-09-24 15:49 don't skate in the dark 2008-09-24 15:49 u haven;'t lived until you bomb latigo in the moonlight 2008-09-24 15:50 hmm, sounds like "haven't died" 2008-09-24 15:51 hey flips 2008-09-24 15:51 hey 2008-09-24 15:51 got to go 2008-09-24 15:51 folks were complaining last night about you being delinquent 2008-09-24 15:51 and now you're abandoning us 2008-09-24 15:55 bh, complaining about what you get for free is not normally considered good style 2008-09-24 15:55 besides, I haven't seen you at one of the sessions 2008-09-24 15:55 I'm also joking if you haven't figured that out by now 2008-09-24 15:56 the internet doesn't see the smile 2008-09-24 15:56 never does 2008-09-24 15:56 you say it like you're never going to see the sun rise again or something like that 2008-09-24 15:57 nothing in santa monica is *that* awful 2008-09-24 16:18 -!- mingming_(~mingming@bi01p1.co.us.ibm.com) has joined #tux3 2008-09-24 16:24 ACTION waves to flips 2008-09-24 16:33 flips: ping me when you get back 2008-09-24 16:34 shapor: the core code is in dleaf.c right ? 2008-09-24 17:36 aw, missed mingming 2008-09-24 17:40 bh, ping 2008-09-24 17:41 the core code is indeed in dleaf.c 2008-09-24 17:56 hey shapor, I'm forced to link my ugly tux3 page and cr*ppy design doc from the lkml post because your version of the design doesn't have headings in most of it 2008-09-24 17:57 flips: yeah, looking over it now 2008-09-24 17:57 er was about an hour ago 2008-09-24 17:58 it's starting to get interesting now 2008-09-24 17:58 with the dwalk stuff 2008-09-24 17:58 yeah, it's going to get more and more complicated as well 2008-09-24 17:58 this step is making it less complicated 2008-09-24 17:58 general resizing isn't implemented yet which includes truncation 2008-09-24 17:59 a couple of big ugly functions will disappear in a week or so 2008-09-24 17:59 all of that needs to be integrated into atomic logging as well, not really that easy 2008-09-24 17:59 roughly zero impact on dleaf 2008-09-24 17:59 talking about the refactoring ? 2008-09-24 18:00 logging impact 2008-09-24 18:00 logging just needs to be done once for all forms of btree, its generic 2008-09-24 18:00 president bush is giving a speech btw 2008-09-24 18:00 hope he chokes 2008-09-24 18:01 he isn't your favorite president of all time ? 2008-09-24 18:01 least in fact 2008-09-24 18:01 can't think of a worse one 2008-09-24 18:01 well, you should start putting in blank functions for atomic logging 2008-09-24 18:01 anyway 2008-09-24 18:01 #offtopic 2008-09-24 18:02 real functions for atomic logging will go in around the time of the kernel port 2008-09-24 18:02 so no need to fire blanks 2008-09-24 18:03 the blanks are helpful for other folks 2008-09-24 18:03 got to get back to my post 2008-09-24 18:03 nearly running out of time 2008-09-24 18:03 and you can start training people to think in terms of it 2008-09-24 18:03 if somebody steps up to implement it I'll put in some stubs 2008-09-24 18:03 otherwise... 2008-09-24 18:04 got other things to do 2008-09-24 18:04 like train up some hackers for the port 2008-09-24 18:05 you should mark it so that folks understand your thinking regarding it 2008-09-24 18:05 I'm tell'n you that it's going to be useful for me and will indicate a direction with your implementation 2008-09-24 18:07 it's been described in a number of posts 2008-09-24 18:08 we can get links up on the page 2008-09-24 18:27 that would be good 2008-09-24 18:54 -!- tim_dimm_(~mobile@166.135.68.85) has joined #tux3 2008-09-24 19:22 "Tux3 gets a high speed atom smasher" -- just posted to lkml 2008-09-24 20:10 -!- Kirantpatil(~kiran@122.167.206.163) has joined #tux3 2008-09-24 20:21 -!- amey(~amey@116.73.35.180) has joined #tux3 2008-09-24 20:21 -!- amey(~amey@116.73.35.180) has left #tux3 2008-09-24 21:37 -!- amey(~amey@116.73.35.180) has joined #tux3 2008-09-24 21:38 -!- amey(~amey@116.73.35.180) has left #tux3 2008-09-24 22:16 -!- Kirantpatil(~kiran@122.167.197.205) has joined #tux3 2008-09-24 22:16 hello list.. 2008-09-24 22:16 i tried to get junkfs 2008-09-24 22:16 but link http://m.a.z.e.pl/junkfs.tar.gz is not working.. 2008-09-24 22:17 any one point me to the right location.. 2008-09-24 22:30 -!- tim_dimm(~timothyhu@cpe-76-90-98-247.socal.res.rr.com) has joined #tux3 2008-09-24 22:37 hello.. 2008-09-24 22:38 howdy 2008-09-24 22:45 could you please get me the link from where i can download junkfs ? 2008-09-24 22:45 as http://m.a.z.e.pl/junkfs.tar.gz is not working..