Back to Home / #uml / 2007 / 05 / Prev Day | Next Day
#uml IRC Logs for 2007-05-02

---Logopened Wed May 02 00:00:06 2007
01:40|-|rasix [~jeruk@] has quit [Ping timeout: 480 seconds]
01:57|-|Ancalagon [] has joined #uml
02:29|-|flatronf700B [~flatronf7@] has joined #uml
02:51|-|motp [~motp@] has joined #uml
03:57|-|Internat-bed [] has joined #uml
03:57|-|Internat [] has quit [Read error: Connection reset by peer]
04:08|-|polyonymous [] has quit [Ping timeout: 480 seconds]
04:19|-|polyonymous [] has joined #uml
05:48|-|fbenites [~chatzilla@] has joined #uml
07:23|-|motp [~motp@] has quit [Quit: Leaving]
07:54|-|fbenites [~chatzilla@] has quit [Quit: ChatZilla 0.9.78 [Firefox]]
09:36|-|kokoko1 [~Slacker@] has joined #uml
10:02|-|jdike [] has joined #uml
10:02<jdike>Hi guys
10:04<kokoko1>hiya jdike
10:05<jdike>kokoko1, don't you have the day off?
10:05<kokoko1>I am from home :)
10:06<kokoko1>imean not working just wondering here and there :)
10:07[~]kokoko1 kinda impressed with jdike memory :)
10:07<jdike>I remember yesterday
10:08|-|Ancalagon [] has quit [Quit: ChatZilla [Firefox]]
10:16|-|hfb [] has joined #uml
10:16|-|hfb [] has left #uml []
11:15|-|ram [] has joined #uml
12:11|-|the_hydra [~mulyadi@] has joined #uml
12:22<the_hydra>hi all
12:23[~]kokoko1 pokes the_hydra
12:23<the_hydra>poking is a nice way to greet somebody these days
12:23<the_hydra>kids, don't try this at home
12:23<the_hydra>very2x dangerous
12:30<the_hydra>jdike: hi
12:30<jdike>kokoko1 was missing you yesterday
12:31<the_hydra>jdike: with the risk considered as getting dumb, could you give a comment about one of my work ?
12:31<the_hydra>jdike: yeah, he pokes me
12:31<the_hydra>sorry via pm
12:33<jdike>OK, I'll take a look later
12:33<the_hydra>jdike: thanks in advance
12:41<kokoko1>thanks jdike yep i talk to him few minutes ago via private chat coz it was off topic
12:42<jdike>just making sure you're not so lonely today
12:42<jdike>we couldn't have that
12:42<kokoko1>heheh i have Internet :D
12:43<kokoko1>working on cisco switches snmp atm
12:50|-|Netsplit <-> quits: Internat-bed
13:08|-|kokoko1 [~Slacker@] has quit [Ping timeout: 480 seconds]
13:37|-|mjf [] has joined #uml
13:45|-|aroscha [] has joined #uml
13:48<mjf>I got this:
13:48<mjf>Does anyone know why?
13:48<mjf>Is it a bug?
13:51<mjf>jdike: can Iget rid of it? there was nothing similar with Sarge. Now I use Etch. What has changed? What does it actually mean?
13:51<jdike>comment out the printk
13:52<mjf>Sorry, but I did not understood :(
13:52<jdike>well, without source, you're not going to comment out anything
13:52<jdike>but it's harmless
13:53<mjf>Aha. OK. But I have still the same kernel as I used with Sarge ;)
13:53<jdike>same UML kernel?
13:54<jdike>but booting an upgrade distro inside it
13:54<mjf>debian-40r0-i386:~# uname -r
13:55<mjf>It was compiled under the Etch, I used it with Sarge virtual images and now I started to use Etch images and got these printks. Isn't it strange?
13:56<jdike>it's using ioctls on the console that it didn't use to
13:56<mjf>Hm. Hrm...
13:57|-|aroscha [] has quit [Ping timeout: 480 seconds]
13:59<mjf>There is another strange thing. Networking stopped working with Etch. I can not see eth0 interface as with Sarge. I use the same setting: eth0=tuntap,uml0eth0 where uml0eth0 is TAP interface.
14:00<mjf>I am a bit confused of this one problem. Should't be "distro" independent and just work?
14:01<mjf>The same vmlinux, the same settings and no eth0 inside :(((
14:01<mjf>Nothing about interface in dmesg...
14:01|-|aroscha [] has joined #uml
14:01<mjf>Simply said, nothing about the interface in logs at all...
14:02<jdike>what does ifconfig -a say?
14:03<mjf>I know just ip l show
14:03<mjf>show just lo
14:03<mjf>no eth0 at all
14:04<mjf>I think it must be something about udev...
14:04<mjf>Wait a moment, please...
14:05<mjf>Yeah. Works now.
14:06<mjf>I erased the damned generated files in /etc/udev/rules.d (persistent-net-blahblah and rebooted the UML and now I got eth0)
14:06<mjf>I promise I will soon switch to gentoo :)))
14:08<mjf>Do you think it is normal that ifconfig is still the main tool to setup interfaces while obsoleted since 1992 or so? Why the hell! Why I was so much arguing everywhere for use of iproute when nobody listens? This makes me so sad. Is the situation the same in gentoo? I hope not.
14:11|-|aroscha [] has quit [Ping timeout: 480 seconds]
14:14<mjf>jdike: Thanks a lot helping me.
14:41|-|aroscha [] has joined #uml
14:45|-|Internat [] has joined #uml
14:51<the_hydra>jdike: sorry to bring #kernelnewbies talks here
14:51<the_hydra>jdike: do you really think, kernel should offer all those HZ possibiliies?
14:51<jdike>did you?
14:51<jdike>I really don't know
14:52<the_hydra>I also can't really claim, as far as I can tell, for example on VCD playing with various HZ (perhaps since the fps isn't so high), xine reports almost zero dropped/skipped frame
14:58<the_hydra>jdike: thanks for the comment.. I am reading it
15:00<the_hydra>jdike: one thing I can say right now, you're really fast when scanning my article...
15:00<jdike>yeah, it was a quick read
15:11|-|aroscha [] has quit [Ping timeout: 480 seconds]
15:12<the_hydra>jdike: great criticism, about mregparm, correct, i think it just matters only if we compile core library with such unusual mregparm
15:13<jdike>and what's with the mismatch between the assembly and source?
15:13<jdike>you fiddled the test in the middle of writing?
15:14<the_hydra>I believe I already check the consistency
15:14<the_hydra>however, I will do my best to recheck again...
15:18<the_hydra>jdike: about the loop, actually the number of the accumulator once you hit the end of the loop
15:18<the_hydra>the inner one AFAIK is expanded to 6 (1+2+3)
15:18<jdike>and there are 20,000,000 iterations
15:18<the_hydra>but I don't claim i am the best when seeing asm codes
15:19<jdike>so, that makes acc == 120,000,000
15:20<jdike>I'm still mildly curious about why gcc didn't unroll the whole thing and just calculate the result
15:22<the_hydra>jdike: probably because gcc can not really "guess" that
15:25<the_hydra>jdike: oh and right, more line in gcc -E, actually that was some part of complete code...I showed the reader one of the interesting spot..namely the ripped ifdef
15:40<jdike>except you closed up the ripped ifdef
15:40<jdike>I guess that gcc will try to unroll small loops
15:40<jdike>and 0 ... 1000 isn't small
15:41<jdike>but it unrolls 0 ... 4 and discovers that it's basicaly a constant
15:47<the_hydra>jdike: pardon me, I don't understand, "closed up"?
15:48<jdike>gcc -E puts empty lines in place of the #if TEST/#endif
15:48<jdike>they didn't appear in your listing
15:49<the_hydra>correct, I made that up
15:49<the_hydra>since I think I can save some spaces to tighten up
15:50<the_hydra>jdike: technically, do you think I should mention it on erratas?
15:50<jdike>depends on what you're trying to show, I guess
15:51<jdike>if it affects the conclusions, I would say so, but I don't see that this does
15:55<the_hydra>well, as you might guess, i just intended to show the ifdef is gone... so I guess it doesn't really matter
15:55<the_hydra>however, I take that loop issue seriously, maybe I am doing wrong...although I hope not
15:55<jdike>you weren't trying to count lines exactly
15:57<the_hydra>jdike: I find this article as learning journey mostly, however, as gentleman, it's my responsilbility to correct any mistake
15:59<the_hydra>jdike: technically, perhaps you have any suggestion so I can write better in the future?
16:00<the_hydra>jdike: and right, my grammar/vocab still sucks, i'm working on it
16:00<jdike>it's not bad
16:00<jdike>occassional wrong word, but not too many
16:08<the_hydra>hm , 10000 * 5000 * 6
16:08<jdike>yeah, that's what the asm says
16:08<the_hydra>and yes, the source is correct too
16:09<jdike>the source says 10000 * 5000 * 4
16:09<the_hydra>let me think...
16:09<the_hydra>the inner loop is expanded into 4 "+="
16:09<jdike>wait a minute
16:09<the_hydra>the first is actually adding zero
16:11<the_hydra>please don't count this as the way to debate you jeff, it's actually me checking my own calculation
16:11<jdike>we're confusing the acc value with the loop count
16:11<jdike>I may have done that too
16:11<jdike>("for(j=0;j<5000;j++) ... for(k=0;k<4;k++)") because that is literally a request to do 30,000 loops.
16:12<the_hydra>s**t, right
16:12<jdike>but there, you're explicitly talking about a loop count, when you meant the final acc value
16:12<the_hydra>s**t, right
16:12<the_hydra>I admit that
16:12<the_hydra>I make the reader confused
16:12<jdike>now, I don't know whether my criticism of the asm mismatch is right
16:13<jdike>if you never had a 0 ... 6, then I was wrong
16:13<the_hydra>actually you point me bigger problem, I said "loop"
16:13<the_hydra>while I point to a loop that barely ask 5000*4
16:14<jdike>Instead of looping three times in the innermost loop
16:14<jdike>four times
16:14<jdike>it makes your program do only 100,000,000 loops instead of 300,000,000
16:14<jdike>50,000,000 instead of 200,000,000
16:16<the_hydra>great :( I missed that simple wording
16:21<the_hydra>ok gtg jeff, I stay a bit late because there is homework to do
16:24|-|the_hydra [~mulyadi@] has left #uml []
16:37|-|mjf [] has quit [Quit: leaving]
17:00|-|aroscha [] has joined #uml
17:03|-|jdike [] has quit [Quit: Leaving]
17:11|-|ram [] has quit [Ping timeout: 480 seconds]
17:18|-|fghj [~tkxue@agp.Stanford.EDU] has joined #uml
17:22|-|ram [] has joined #uml
19:11|-|fghj [~tkxue@agp.Stanford.EDU] has quit [Quit: Leaving]
19:16|-|ram [] has quit [Ping timeout: 480 seconds]
19:27|-|ram [] has joined #uml
20:23|-|ram [] has quit [Ping timeout: 480 seconds]
21:23|-|rasix [~jeruk@] has joined #uml
22:59|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:59|-|VS_ChanLog [] has joined #uml
---Logclosed Thu May 03 00:00:44 2007