2008-09-14 02:04 ACTION is back from (goth club) Sabbat 2008-09-14 02:04 flips: oct 31th is halloween btw 2008-09-14 02:04 you might like to reschedule 2008-09-14 02:04 of course 2008-09-14 02:04 that was the point 2008-09-14 02:05 ACTION chuckles 2008-09-14 02:05 you going trick or treating? 2008-09-14 02:05 ok, now I would like to know your rationale for this 2008-09-14 02:05 probably neither, I don't live in an area that's family friendly 2008-09-14 02:06 be back in a bit 2008-09-14 02:06 I applaud you for your sense of humor, but I'm left a bit hanging as to what you're intending 2008-09-14 02:06 ACTION is freaking drunk now 2008-09-14 02:06 drunk IRCing 2008-09-14 02:06 I'll be sober in about 1/2 hour or so 2008-09-14 02:06 ok 2008-09-14 02:08 ACTION heads to get late night food 2008-09-14 02:08 flips: btw, I live right behind a big goth night/club in San Diego, Sabbat 2008-09-14 02:09 you can get a listing of clubs from socalgoth (southern cal goth) 2008-09-14 02:09 which has unified LA through SD listing 2008-09-14 02:09 for goth/industrial events 2008-09-14 04:13 -!- trymeeeee(~zxcvbnm@123.236.188.107) has joined #tux3 2008-09-14 05:28 -!- Aks(~ankitsriv@123.237.71.198) has joined #tux3 2008-09-14 05:34 -!- Aks(~ankitsriv@123.237.71.198) has left #tux3 2008-09-14 11:05 -!- pgquiles(~pgquiles@50.Red-79-153-248.staticIP.rima-tde.net) has joined #tux3 2008-09-14 11:13 -!- stargazr5(~gauravstt@59.95.30.8) has joined #tux3 2008-09-14 11:51 -!- Kirantpatil(~kiran@122.167.212.171) has joined #tux3 2008-09-14 11:51 -!- Kirantpatil(~kiran@122.167.212.171) has left #tux3 2008-09-14 11:54 -!- MaZe(~MaZe@c-24-6-86-168.hsd1.ca.comcast.net) has joined #tux3 2008-09-14 13:06 -!- RazvanM(~RazvanM@dazzler.isi.jhu.edu) has joined #tux3 2008-09-14 16:04 ACTION starts a new edition of the Tux3 Report 2008-09-14 16:05 "Xattrs and Atoms" 2008-09-14 16:05 nice to have this as a fait accompli 2008-09-14 16:05 almsot 2008-09-14 16:05 just have to set the layout fields for the real filesystem and do some full system testing 2008-09-14 16:38 sk8 oclock 2008-09-14 17:17 hey 2008-09-14 19:46 flips: check my repo 2008-09-14 19:46 fix for inode->xcache leak 2008-09-14 19:59 will do 2008-09-14 20:00 pulls very slowly when there's a simultaneous kernel download in progress 2008-09-14 20:00 got to get me sum more a that bandwidth 2008-09-14 20:02 shapor, what's that (int)strlen(name) for? 2008-09-14 20:13 hey, should I do the tux3 kernel part with git or mercurial? 2008-09-14 20:14 maybe I should as on the mercurial channel 2008-09-14 20:16 hey shapor 2008-09-14 20:22 it'll eventually need to be git in the kernel 2008-09-14 20:27 %.*s expects int not size_t 2008-09-14 20:28 flips: ^ 2008-09-14 20:28 konrad, not really 2008-09-14 20:28 there are some mercurial projects for kernel things, for example btrfs 2008-09-14 20:28 hm, ok 2008-09-14 20:29 it's rather stupid for strlen to return size_t 2008-09-14 20:29 indeed 2008-09-14 20:29 kinda makes you want to reimplement it, doesnt it? 2008-09-14 20:29 like anybody should scan that much ascii text looking for a crappy null byte 2008-09-14 20:29 in asm ;) 2008-09-14 20:30 your basic 5 byte assembly program 2008-09-14 20:30 12 if you can a really fancy fast one 2008-09-14 20:30 scasb makes it easy doesnt it? 2008-09-14 20:30 scasb is slow on a lot of procs 2008-09-14 20:30 have'nt been keeping up with the latest 2008-09-14 20:30 hmm 2008-09-14 20:31 but a simple look using basic register instructions is fastest today 2008-09-14 20:31 let the superscaler logic do its thing 2008-09-14 20:31 and the shadow registers 2008-09-14 20:31 simple loop 2008-09-14 20:31 anyway, we're git 2008-09-14 20:31 I just checked it in 2008-09-14 20:31 tux3 stub kernel fs is landing tonight 2008-09-14 20:32 shapor, anyway we have %t 2008-09-14 20:32 that's for this braindamage I think 2008-09-14 20:32 wll 2008-09-14 20:32 doesn't work for %.*s 2008-09-14 20:32 yuck 2008-09-14 20:33 stupid ancient unix gods 2008-09-14 20:34 heh 2008-09-14 20:38 there we go 2008-09-14 20:41 typical linux: CONFIG_MMU means "CONFIG_NOMMU" 2008-09-14 20:42 tux3 will not support nommu for now 2008-09-14 20:42 if somebody wants that they can pay for it 2008-09-14 20:43 ah, ramfs has an actual application 2008-09-14 20:43 it implements rootfs 2008-09-14 20:43 that's why it got a little bloaty 2008-09-14 20:43 lately 2008-09-14 20:47 * Tux3 Versioning Filesystem 2008-09-14 20:47 * 2008-09-14 20:47 * Portions Copyright (C) 2000 Linus Torvalds, 2000 Transmeta Corp. 2008-09-14 20:47 * Licensed under the GPL v2 2008-09-14 20:47 */ 2008-09-14 20:47 well 2008-09-14 20:47 what about the other (c) 2008-09-14 20:48 * Tux3 Versioning Filesystem 2008-09-14 20:48 * 2008-09-14 20:48 * Copyright (c) 2008, Daniel Phillips 2008-09-14 20:48 * Portions Copyright (C) 2000 Linus Torvalds, 2000 Transmeta Corp. 2008-09-14 20:48 * Licensed under the GPL v2 2008-09-14 20:48 */ 2008-09-14 20:48 there we go 2008-09-14 20:50 one little c one C 2008-09-14 20:52 hrm there is a still one leak in the inode test 2008-09-14 20:52 ==15560== 8,160 (8,040 direct, 120 indirect) bytes in 1 blocks are definitely lost in loss record 4 of 7 2008-09-14 20:52 ==15560== at 0x4A1B858: malloc (vg_replace_malloc.c:149) 2008-09-14 20:52 ==15560== by 0x401DC2: new_map (buffer.c:442) 2008-09-14 20:52 ==15560== by 0x40988C: new_inode (inode.c:111) 2008-09-14 20:52 ==15560== by 0x40AD46: make_tux3 (inode.c:476) 2008-09-14 20:52 linus wrote the big C 2008-09-14 20:52 ==15560== by 0x40B17B: main (inode.c:530) 2008-09-14 20:53 I'm not correcting his typos 2008-09-14 20:53 I treat his copyright notice as (c) linus 2008-09-14 20:53 well 2008-09-14 20:53 it does look stupid 2008-09-14 20:53 there we go, changed to (c), I'm a flagrant copyright scofflaw 2008-09-14 20:53 arrest me 2008-09-14 20:54 i do not expect that leak to last long 2008-09-14 20:59 config TUX3_FS 2008-09-14 20:59 tristate "Tux3 Versioning Filesystem" 2008-09-14 20:59 help 2008-09-14 20:59 To compile this file system support as a module, choose M here: the 2008-09-14 20:59 module will be called tux3. 2008-09-14 20:59 If unsure, say Maybe. 2008-09-14 20:59 hrm its only the map in the sb inode 2008-09-14 21:00 in make_tux3 2008-09-14 21:00 seems odd since free_inode does indeed free the map unless its null 2008-09-14 21:00 the way those initializers work is dodgy 2008-09-14 21:00 structure assignments 2008-09-14 21:01 combined with desginated init = brainmuck 2008-09-14 21:01 probably should do it all with mallocs 2008-09-14 21:01 the fs init that is 2008-09-14 21:01 the reason for the cute little minimal struct decs is getting old 2008-09-14 21:38 -!- amey(~amey@116.73.35.180) has joined #tux3 2008-09-14 21:39 hi 2008-09-14 21:40 -!- amey(~amey@116.73.35.180) has left #tux3 2008-09-14 21:41 cat /proc/filesystems 2008-09-14 21:41 nodev sysfs 2008-09-14 21:41 nodev rootfs 2008-09-14 21:41 nodev bdev 2008-09-14 21:41 nodev proc 2008-09-14 21:41 nodev sockfs 2008-09-14 21:41 nodev pipefs 2008-09-14 21:41 nodev anon_inodefs 2008-09-14 21:41 nodev tmpfs 2008-09-14 21:41 nodev inotifyfs 2008-09-14 21:41 nodev devpts 2008-09-14 21:41 reiserfs 2008-09-14 21:41 ext3 2008-09-14 21:41 ext2 2008-09-14 21:41 nodev tux3 2008-09-14 21:41 nodev ramfs 2008-09-14 21:41 nodev hostfs 2008-09-14 21:41 nodev mqueu 2008-09-14 21:41 let's get rid of some useless ones 2008-09-14 21:42 anon_inodefs <- :p 2008-09-14 21:42 what is that? 2008-09-14 21:42 crap 2008-09-14 21:42 haven't looked at it 2008-09-14 21:42 but I can tell from the name 2008-09-14 21:42 few other dodgy looking ones 2008-09-14 21:43 now, are job is to get rid of the nodev on tux3 2008-09-14 21:43 let's try to mount 2008-09-14 21:45 root@deep:~# mount -t tux3 tux3 /mnt 2008-09-14 21:45 root@deep:~# echo hello >/mnt/foo 2008-09-14 21:45 root@deep:~# cat /mnt/foo 2008-09-14 21:45 hello 2008-09-14 21:46 root@deep:~# mount 2008-09-14 21:46 /dev/ubda on / type ext2 (rw) 2008-09-14 21:46 proc on /proc type proc (rw) 2008-09-14 21:46 devpts on /dev/pts type devpts (rw,gid=5,mode=620) 2008-09-14 21:46 tux3 on /mnt type tux3 (rw) 2008-09-14 21:46 ok, time to check it in 2008-09-14 22:00 http://phunq.net/ddtree 2008-09-14 22:00 http://phunq.net/ddtree?p=tux3fs;a=summary 2008-09-14 22:01 just for now 2008-09-14 22:06 git... it's actually pretty bad 2008-09-14 22:06 compared to mercurial 2008-09-14 22:06 user unfriendly 2008-09-14 22:07 does not do what you expect 2008-09-14 22:18 you need to do commit -a 2008-09-14 22:18 to get what mercurial does for just commit 2008-09-14 22:19 and what any rational person would want 2008-09-14 22:22 -!- RazvanM(~RazvanM@pool-151-196-118-156.balt.east.verizon.net) has joined #tux3 2008-09-14 22:34 nice 2008-09-14 22:42 http://shapor.com/tux3/ updated 2008-09-14 22:50 :) 2008-09-14 22:51 shapor, when's the next round of updates on the design doc? 2008-09-14 22:56 i was just thinking about that 2008-09-14 23:39 -!- nataliep_(~nataliep@207.47.98.129.static.nextweb.net) has joined #tux3 2008-09-14 23:45 so, fuse's pkg-config wants -D_FILE_OFFSET_BITS and so does everything else, or else off_t will be wrong (current just diskio has it) 2008-09-14 23:45 the only problem is, if I put -D_FILE_OFFSET_BITS on everything, then it shoes up twice in the fuse compile 2008-09-14 23:45 esthetically irritating 2008-09-14 23:46 well, it's just going to be that way 2008-09-14 23:46 and our build will start to suck, just like every build