Back to Home / #uml / 2007 / 11 / Prev Day | Next Day
#uml IRC Logs for 2007-11-08

---Logopened Thu Nov 08 00:00:23 2007
02:15|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has joined #uml
02:40|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has quit [Quit: aroscha]
03:26|-|aroscha [~aroscha@77.119.0.179] has joined #uml
04:00|-|aroscha [~aroscha@77.119.0.179] has quit [Quit: aroscha]
04:38|-|silug [~steve@38.104.162.2] has quit [Ping timeout: 480 seconds]
04:47|-|fonz_ [~fonz@gate.tellitec.de] has quit [Ping timeout: 480 seconds]
04:50|-|fonz_ [~fonz@gate.tellitec.de] has joined #uml
05:35|-|fonz_ [~fonz@gate.tellitec.de] has quit [Ping timeout: 480 seconds]
05:36|-|fonz_ [~fonz@pD95371FD.dip.t-dialin.net] has joined #uml
06:02|-|aroscha [~aroscha@213.235.234.114] has joined #uml
06:30|-|silug [~steve@38.104.162.2] has joined #uml
06:56|-|aroscha [~aroscha@213.235.234.114] has quit [Quit: aroscha]
07:40|-|krau [~cktakahas@200.184.118.132] has quit [Ping timeout: 480 seconds]
07:46|-|fonz__ [~fonz@pD9537118.dip.t-dialin.net] has joined #uml
07:46|-|fonz__ [~fonz@pD9537118.dip.t-dialin.net] has left #uml []
07:51|-|fonz_ [~fonz@pD95371FD.dip.t-dialin.net] has quit [Ping timeout: 480 seconds]
08:29|-|dang [~dang@nemesis.fprintf.net] has quit [Quit: Leaving.]
08:39|-|aroscha [~aroscha@77.117.8.145] has joined #uml
08:45|-|besonen_mobile [~besonen_m@71-220-228-70.eugn.qwest.net] has quit [Read error: Connection reset by peer]
08:45|-|besonen_mobile [~besonen_m@71-220-228-70.eugn.qwest.net] has joined #uml
08:49|-|aroscha [~aroscha@77.117.8.145] has quit [Ping timeout: 480 seconds]
09:00|-|dang [~dang@aa-redwall.nexthop.com] has joined #uml
10:17|-|jdike [~jdike@pool-71-248-190-161.bstnma.fios.verizon.net] has joined #uml
10:18<jdike-#uml->>Hi guys
10:20<caker-#uml->>hello
10:22<Magotari-#uml->>jdike: Whatever -bb meant is just not working. The case where end_interrupt will call schedule is really rare, at least the way I test it. On a pure fork benchmark I got negative performance improvement. It did not call arch_switch_to just 8 times during the 15 second benchmark run.
10:22<jdike-#uml->>I'm suprised it happens at all
10:22<jdike-#uml->>surprised
10:22<Magotari-#uml->>It does though.
10:22<Magotari-#uml->>More often on a kernel build.
10:22<jdike-#uml->>but it hurts performance?
10:22<Magotari-#uml->>I will have to test a kernel compile now.
10:23<jdike-#uml->>it should make no difference
10:23<Magotari-#uml->>In my benchmarks it did hurt it a bit. The only way to check what end_interupt is about to do in regard to schedule is to call need_resched.
10:24<Magotari-#uml->>And the special case where it did not need_resched... It did not occur enough to give a benefit.
10:24<jdike-#uml->>what did your change do, exactly?
10:24<Magotari-#uml->>Of course the impact was very minimal, but it was there.
10:24<Magotari-#uml->>Simple. A oneliner patch.
10:24<jdike-#uml->>I would just have moved the arch_switch_to to below interrupt_end()
10:25<jdike-#uml->>that should make no differene
10:25<jdike-#uml->>difference
10:26<Magotari-#uml->>Ohshit.
10:26<Magotari-#uml->>I think I did something wrong in the oneliner.
10:26<jdike-#uml->>hehe
10:27<jdike-#uml->>no matter how small the patch, there's always room for a bug
10:27<jdike-#uml->>sometimes several
10:27<Magotari-#uml->>It is a bloody miracle that the system was not down before it started. It was doing extra work exactly when it was not supposed to do being it, and not doing it when it was supposed to be doing it.
10:31<Magotari-#uml->>Hmm... That is very funny indeed. The code ran just fine either way. No crashes, nothing. Even with my patch wrong.
10:32|-|aroscha [~aroscha@77.117.136.112] has joined #uml
10:32<Magotari-#uml->>Oh, and the inverted case does not impact performance.
10:47<Magotari-#uml->>Hell, after I got frustrated and deleted the whole switch_to bit from the fork_handler, the system worked right as rain. With worst performance I tested to date. Although those extra 50ms don't hurt too much over 15seconds, I guess.
10:47<Magotari-#uml->>Erm, I meant 500.
10:48<jdike-#uml->>after deleting code, you got worse performance?
10:49<Magotari-#uml->>Yes.
10:49<Magotari-#uml->>I am as surprised as you are.
10:49<Magotari-#uml->>I am also surprised that it booted and worked at all.
10:50<jdike-#uml->>me too
10:50<jdike-#uml->>that should break anything that uses TLS
10:50<Magotari-#uml->>And this benchmark is the first one which seems consistent in the results. C for the win.
10:50<Magotari-#uml->>The kilogram booted fine. Now compiling the kernel.
10:51<Magotari-#uml->>So... Yeah. Wow.
10:53<Magotari-#uml->>You know, the switch_to is nearly a nop if you don't have tls enabled, so maybe I don't have it through some crazy twist of fate. Gonna check it out.
10:53|-|aroscha [~aroscha@77.117.136.112] has quit [Quit: aroscha]
10:54<jdike-#uml->>In UML, it's basically just a setjmp/longjmp
10:54<jdike-#uml->>on the host, it's a context switch
10:54<jdike-#uml->>once you reach userspace
10:57|-|krau [~cktakahas@200.184.118.132] has joined #uml
11:30|-|tchan [~tchan@c-24-13-84-219.hsd1.il.comcast.net] has quit [Quit: WeeChat 0.2.7-dev]
11:31|-|tchan [~tchan@c-24-13-84-219.hsd1.il.comcast.net] has joined #uml
12:03|-|anderiv [~anderiv@207-67-87-34.static.twtelecom.net] has quit [Quit: leaving]
12:03|-|anderiv [~anderiv@207-67-87-34.static.twtelecom.net] has joined #uml
12:31|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has joined #uml
12:50|-|dang [~dang@aa-redwall.nexthop.com] has quit [Ping timeout: 480 seconds]
13:03|-|dang [~dang@aa-redwall.nexthop.com] has joined #uml
13:14|-|ram [~ram@pool-72-90-125-50.ptldor.fios.verizon.net] has joined #uml
13:14|-|hfb [~hfb@pool-71-106-219-180.lsanca.dsl-w.verizon.net] has joined #uml
13:21|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has quit [Quit: aroscha]
13:43<fo0bar-#uml->>jdike: hey, do you know what the status is of skas3? I was able to get it working with 2.6.22 using the 2.6.20 patch and some tweaks, but is blaisorblade still working on it?
13:43|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has joined #uml
13:47|-|aroscha [~aroscha@chello213047053193.30.11.tuwien.teleweb.at] has quit []
13:49<jdike-#uml->>dunno what bb is up to
13:49<jdike-#uml->>he's been quiet recentluy
14:21<Magotari-#uml->>Oh. So it was blaisorblade who left that comment.
14:22<jdike-#uml->>yup
14:25|-|mgross [~mgross@jffwprtest.jf.intel.com] has joined #uml
14:31|-|ram [~ram@pool-72-90-125-50.ptldor.fios.verizon.net] has quit [Ping timeout: 480 seconds]
14:32|-|dsoul_ [darksoul@vice.ii.uj.edu.pl] has joined #uml
14:33|-|krau [~cktakahas@200.184.118.132] has quit [Ping timeout: 480 seconds]
14:36|-|Netsplit synthon.oftc.net <-> solenoid.oftc.net quits: dsoul, kokoko1, peterz, apic, Magotari
14:37|-|Netsplit over, joins: Magotari
14:37|-|Netsplit over, joins: peterz
14:40|-|krau [~cktakahas@200.184.118.132] has joined #uml
14:41|-|apic [~andi@paranerd-geekophren.name] has joined #uml
14:49|-|kokoko1 [~Slacker@203.148.65.19] has joined #uml
14:51|-|tyler29 [~tyler@ARennes-257-1-94-158.w81-48.abo.wanadoo.fr] has joined #uml
15:02|-|ram [~ram@bi01p1.co.us.ibm.com] has joined #uml
15:10|-|Baltam [~WIKIMOKI@tor-irc.dnsbl.oftc.net] has quit [Ping timeout: 480 seconds]
15:23|-|waldner [~waldner@host-84-220-175-96.cust-adsl.tiscali.it] has joined #uml
15:25<waldner-#uml->>jdike: if you want, I'm here to continue helping in debugging the port-helper issue
15:25<jdike-#uml->>OK
15:26<jdike-#uml->>did you build your UML yourself?
15:28<waldner-#uml->>yes
15:28<waldner-#uml->>I can paste the config somewhere if you want
15:29<jdike-#uml->>with debugging info?
15:29|-|silug [~steve@38.104.162.2] has quit [Read error: Connection reset by peer]
15:30<waldner-#uml->>I mean the kernel config
15:31<waldner-#uml->>the kernel is ~5MB, so I guess it has no debug info
15:31<jdike-#uml->>I guess not
15:31<jdike-#uml->>linux --showconfig | grep DEBUG_INFO
15:32<waldner-#uml->>unfortunately, I cannot rebuild it as I'm under gentoo and there are problems with the headers
15:32<waldner-#uml->>ok, grep returns nothing
15:34<jdike-#uml->>hmm
15:34<jdike-#uml->>looking at data would be helpful
15:35<waldner-#uml->>I'm sorry
15:35<waldner-#uml->>do you know a workaround for the gentoo linux-headers issue?
15:36<jdike-#uml->>nope
15:36<jdike-#uml->>what's the problem?
15:37<jdike-#uml->>asm/page.h or asm/user.h not existing?
15:37<waldner-#uml->>http://rafb.net/p/MrU1PU26.html
15:39<jdike-#uml->>not in git yet, hold on
15:40<waldner-#uml->>if it's something I can fix by hand or with a patch, I can do it
15:43<jdike-#uml->>http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff_plain;h=7a82b63f19b0a05a76271aee1eb7905cd1c7d5ad
15:43<waldner-#uml->>ok, hold on while I try to apply the patch and rebuild the kernel with debug info
15:44<waldner-#uml->>thanks
15:45<waldner-#uml->>Makfile-i386? This is an amd64
15:45<waldner-#uml->>is it correct?
15:46<jdike-#uml->>I was wondering the same thing
15:47<waldner-#uml->>lol
15:47<jdike-#uml->>it fixes that error on some platform
15:48<jdike-#uml->>waitaminute
15:48<jdike-#uml->>wrong error
15:49<waldner-#uml->>meanwhile I download 2.6.24-rc2
15:52<jdike-#uml->>I'm gonna have to disappear on you again
15:55<waldner-#uml->>oh well
15:55<waldner-#uml->>no prob
15:55<waldner-#uml->>the port-helper thing is not a big issue
15:56<waldner-#uml->>we'll try to do something another time
15:56<waldner-#uml->>thanks for your helpfulness
15:58|-|tyler29 [~tyler@ARennes-257-1-94-158.w81-48.abo.wanadoo.fr] has quit [Ping timeout: 480 seconds]
16:08|-|tyler29 [~tyler@ARennes-257-1-125-168.w86-210.abo.wanadoo.fr] has joined #uml
17:06|-|dang [~dang@aa-redwall.nexthop.com] has quit [Quit: Leaving.]
17:13|-|waldner [~waldner@host-84-220-175-96.cust-adsl.tiscali.it] has quit [Quit: using sirc version 2.211+KSIRC/1.3.12]
17:25|-|silug [~steve@38.104.162.2] has joined #uml
17:47|-|tyler29 [~tyler@ARennes-257-1-125-168.w86-210.abo.wanadoo.fr] has quit [Ping timeout: 480 seconds]
17:58|-|karrde_ changed nick to da-x_
18:46|-|dang [~dang@nemesis.fprintf.net] has joined #uml
19:09|-|mgross [~mgross@jffwprtest.jf.intel.com] has quit [Quit: Leaving]
19:33|-|hfb [~hfb@pool-71-106-219-180.lsanca.dsl-w.verizon.net] has quit [Quit: Leaving]
20:51|-|ram [~ram@bi01p1.co.us.ibm.com] has quit [Ping timeout: 480 seconds]
21:23|-|dang [~dang@nemesis.fprintf.net] has quit [Ping timeout: 480 seconds]
22:07|-|Nem^ [~Nem@dslb-084-056-232-209.pools.arcor-ip.net] has quit [Ping timeout: 480 seconds]
22:17|-|Nem^ [~Nem@dslb-084-056-254-226.pools.arcor-ip.net] has joined #uml
22:33|-|dang [~dang@nemesis.fprintf.net] has joined #uml
22:36|-|jdike [~jdike@pool-71-248-190-161.bstnma.fios.verizon.net] has quit [Quit: Leaving]
23:59|-|VS_ChanLog [~stats@ns.theshore.net] has left #uml [Rotating Logs]
23:59|-|VS_ChanLog [~stats@ns.theshore.net] has joined #uml
---Logclosed Fri Nov 09 00:00:56 2007