luca_ | hy everybody ! | 00:08 |
---|---|---|
luca_ | im' intrested in implementing a file-system changes notification tool | 00:09 |
luca_ | i am wondering if is possible to get journaling event from my module | 00:12 |
luca_ | are there particular kernel structures involved with the journaling system ? | 00:13 |
BenC | luca_: fs/jfs/ might give some useful info | 00:17 |
luca_ | thank you BenC | 00:21 |
luca_ | are there anything under /proc or /sys that will let me know if some change in the filesystem occurs? | 00:31 |
mjg59 | No | 00:32 |
luca_ | sorry I got lost in fs/jfs | 00:32 |
mjg59 | If you're purely interested at the filesystem level, inotify will give you notifications | 00:32 |
mjg59 | But I suspect it won't work too well if you try to look at the whole filesystem | 00:33 |
pwnguin | that would be pretty crazy | 00:37 |
pwnguin | and god help you if that notification framework ended up causing writes to the filesystem | 00:38 |
mjg59 | If you're interested in things at the block level rather than the filesystem level, check block/blk-core.c and search for block_dump. fs-writeback.c has similar code that gives notifications if a page is dirtied. | 00:40 |
luca_ | Is it possible (kernel or user space) to read directly that table the journaling fill with envents occurred to file? | 00:40 |
luca_ | read it , and understand what happened? | 00:41 |
mjg59 | Well, sure | 00:42 |
mjg59 | But ext3 only journals metadata by defau;t | 00:42 |
mjg59 | File data changes won't go anywhere near the journal | 00:42 |
luca_ | mmm. right | 00:44 |
luca_ | do you know more or less how does this happens: | 00:46 |
luca_ | i open a directory with a gui as nautilus | 00:47 |
luca_ | i create a file from a shell into that directory | 00:47 |
luca_ | and magically nautilus know that a file was created | 00:48 |
luca_ | and show the file icon almost immedately | 00:48 |
mjg59 | It uses inotify | 00:48 |
luca_ | thanks mjg59 | 00:54 |
=== asac_ is now known as asac | ||
=== CjNr13 is now known as CjNr14 | ||
CjNr14 | does someone know if there is a way to export a memory area from the host to the guest (Video RAM/ROM for exemple) ? | 13:09 |
CjNr14 | Do I need to write a driver like the pci frontend/backend one? | 13:09 |
CjNr14 | sorry, wrong chan lol | 13:10 |
rtg__ | Ng: talk to me. any e1000 joy? | 14:21 |
Ng | rtg__: not as yet, I don't see silbs around atm, but I'll keep my eye open. | 14:27 |
rtg__ | Ng: k | 14:27 |
Ng | I might even keep both eyes open ;) | 14:28 |
Nafallo | hehe | 14:51 |
munckfish | oh depressing. Recompiling the 2.6.25 kernel is taking 2 hrs 20 mins on my PS3 (weeps). | 20:52 |
munckfish | and that's with ccache | 20:53 |
* munckfish eyes G5s on Ebay, wonders what wife would think ... | 20:56 | |
dupondje | https://bugs.launchpad.net/ubuntu/+bug/235889 | 22:23 |
dupondje | somebody can take a look ? :) | 22:23 |
mkrufky | dupondje: Leann asked you to do a bisection test | 22:26 |
mkrufky | if you do that test, you'll be able to find the exact changeset that fixes the problem, so that they can think about backporting it to hardy | 22:26 |
dupondje | I must say I don't really understand it howto :x | 22:26 |
mkrufky | google git-bisect | 22:26 |
dupondje | I must compile kernel 100000 times then until I get the right patch ? ... | 22:27 |
pwnguin | well | 22:27 |
mkrufky | BISECTION TEST | 22:27 |
pwnguin | more like binary search | 22:27 |
mkrufky | bisection lets you find the fix in ~7 builds | 22:27 |
pwnguin | so something like log base 2 of the number of commits | 22:27 |
mkrufky | er, wait... i think i pulled that number 7 out of my arse .... | 22:28 |
mkrufky | anyway, bisection helps to find the fix in the least # of builds possible | 22:28 |
dupondje | any id what part of kernel I should look @ for a possible patch ? | 22:28 |
pwnguin | thats what bisect helps with | 22:29 |
pwnguin | you'll find the changeset that introduced the bug with bisect, and that'll narrow the places down considerably | 22:29 |
mkrufky | im goin home .. .tty all later | 22:30 |
pwnguin | dupondje: i agree, i wish building the ubuntu kernel was simpler | 22:31 |
dupondje | well pwnguin, It crashes with ubuntu kernels ... but not with the latest kernel.org kernel ... | 22:31 |
dupondje | dunno about the older kernel.org versions | 22:31 |
dupondje | so maby its just a bug introduced into the ubuntu patches | 22:32 |
pwnguin | here's an idea | 22:32 |
pwnguin | oh, it still doesnt build | 22:33 |
dupondje | indeed :) | 22:33 |
dupondje | ogasawara also had that id ;) | 22:33 |
pwnguin | if ubuntu's kernel built, there shuold be a fairly small diff | 22:33 |
dupondje | and my damn slow CPU :'( | 22:36 |
dupondje | AMD Athlon64 3800+ :( | 22:37 |
pwnguin | plenty fast | 22:37 |
dupondje | not fast enough imo :D | 22:38 |
|dupondje| | mmm :x | 22:45 |
=== |dupondje| is now known as dupondje | ||
infinity | rtg__: BTW, I have a working hppa kernel patch for you for inclusion in -20- after 8.04.1 is out the door. | 23:11 |
infinity | rtg__: chinstrap:~adconrad/hppa.diff | 23:18 |
infinity | rtg__: Obviously, the change to 2-binary-arch.mk is for our own personal use, but the config changes work smashingly, and that's the precise source we're using in the DC right now. | 23:19 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!