Back to Home / #uml / 2007 / 09 / Prev Day | Next Day
#uml IRC Logs for 2007-09-30

---Logopened Sun Sep 30 00:00:53 2007
00:09|-|tasaro [] has quit [Ping timeout: 480 seconds]
00:10|-|psy [] has joined #uml
00:12|-|linbot [] has quit [Ping timeout: 480 seconds]
00:13<psy>I'm trying to make it possible to give a bit of memory that I can allow multiple instances of uml access to but I'm not having much luck, can anyone here give me some pointers? have you done this?
00:15<magotari>An interesting idea. Cannot say that I tried though.
00:18<psy>eh -- seems to be the common response -- so far all we've got to work on is a potential security bug ( Other issues, point 4) which hasn't been of much use yet
00:19<magotari>Sorry, I am just a newbie to here, linux, everything.
00:19<magotari>Actually, I came here for help.
00:19<magotari>I just thought that any response would be better than silence, which is what I got so far.
00:20<psy>hehe, fair enough -- thanks for that =)
00:20<psy>so far my time around here is pretty quiet as well -- you'll probable have best luck if you've got UML specific questions if you can show up when jdike is here
00:20<magotari>Sorry to get your hopes up.
00:21<magotari>He is the one who I am waiting for.
00:21<magotari>I looked at the channel logs.
00:21<magotari>No activity without him, looks like.
00:21<psy>what's your question anyway? I probably can't aswer it but now I'm curious
00:21<magotari>I changed my host kernel.
00:22<magotari>And now all my UML kernels are busted, including the one from the UML website.
00:22<psy>weird -- have you tried building one from source?
00:22<magotari>Kernel panic - not syncing: start_userspace : expected SIGSTOP, got status = 256
00:23<magotari>Yes, the new one is built from source.
00:23<magotari>Gentoo here.
00:23<psy>the new UML, sorry
00:23<magotari>I googled, and this error is a memsplit problem.
00:23<magotari>That too. The new UML is also from source.
00:24<magotari>And sure enough, my new host kernel has a different memsplit.
00:24<magotari>And changing back to the old memsplit solves the problem again.
00:27<psy>and you want it to work with the new memsplit, I'm assuming?
00:28<magotari>Yeah, I need my full 1G of ram.
00:28<magotari>To fit in more UMLs for experiments.
00:29<psy>silly question -- when you build your new UML have you checked to make sure you were building it with the right memsplit? (in make menuconfig : UML specific -> Host memory split)
00:29<magotari>I think I checked, but let me check again.
00:30<magotari># CONFIG_HOST_VMSPLIT_3G is not set
00:30<magotari># CONFIG_HOST_VMSPLIT_2G is not set
00:30<magotari># CONFIG_HOST_VMSPLIT_1G is not set
00:30<magotari># CONFIG_VMSPLIT_3G is not set
00:30<magotari># CONFIG_VMSPLIT_2G is not set
00:30<magotari># CONFIG_VMSPLIT_2G_OPT is not set
00:30<magotari># CONFIG_VMSPLIT_1G is not set
00:30<magotari>The first one is UML, the second one is host.
00:30<magotari>Both 23-rc7
00:31<magotari>Strange, eh?
00:31<psy>yea -- to my untrained eye that looks right
00:32<magotari>Yeah. I am getting quite freaked out without a working virtual machine. I think I am addicted or something.
00:33<magotari>Oh well, modprobe kqemu time.
00:34<psy>best of luck
00:34<magotari>Thank you for listening. :)
00:35|-|VS_ChanLog [] has joined #uml
00:36|-|linbot [] has joined #uml
00:37|-|caker [] has joined #uml
03:04|-|psy [] has quit [Ping timeout: 480 seconds]
04:00|-|tyler29 [] has joined #uml
04:35|-|tyler29 [] has quit [Ping timeout: 480 seconds]
04:35|-|tyler29 [] has joined #uml
06:25|-|tyler29 [] has quit [Ping timeout: 480 seconds]
06:44|-|tyler29 [] has joined #uml
07:04|-|magotari [] has quit [Quit: Lost terminal]
07:40|-|GoGi [] has quit [Ping timeout: 480 seconds]
07:52|-|GoGi [] has joined #uml
08:17|-|tasaro [] has joined #uml
09:14|-|Magotari [] has joined #uml
09:29|-|Blissex [] has quit [Remote host closed the connection]
09:41|-|dang [] has joined #uml
10:44|-|psy [] has joined #uml
11:23|-|psy [] has quit [Ping timeout: 480 seconds]
11:43|-|psy [] has joined #uml
11:45<psy>is the physical memory, as seen by uml, the memory space that your user has access to? in other words: in multiple instances of UML if I were to grab the same physical page would it all map back to the same memory?
11:50|-|flamingo [] has joined #uml
11:53|-|psy [] has quit [Ping timeout: 480 seconds]
11:54|-|psy [] has joined #uml
12:32|-|psy [] has quit []
12:32|-|psy [] has joined #uml
12:35<Magotari>psy: About your question from an hour ago... Wouldn't grabbing the same memory by more than one process end up in a SIGSEGV?
12:38<psy>Magotari: fair point
13:07|-|psy [] has quit [Ping timeout: 480 seconds]
13:08|-|psy [] has joined #uml
13:30|-|florin [] has joined #uml
13:31<florin>hi... I am trying to build a new linux root_fs (CentOS 5) ... I got it booting, it enters runlevel 3... but after starting syslog it just hangs and opens some xterms... however, these exterms are associated to the host system and not to the uml (guest)... anyone has any idea ?
13:35<Magotari>Hmm... That is strange. Is is accesible by ssh?
13:35<florin>didn't set any networking yet...
13:35<florin>using the kernel from btw
13:36<florin>what centos pkg/application displays the login shell ?
13:36<Magotari>Did you have any serial consoles in inittab? If yes, you can get to your system by adding ssl=port:<port> to the uml commandline, and by doing a "telnet localhost <port>"
13:37<florin>let me see
13:37<Magotari>I think the executable is called getty, agetty or something like that.
13:37<florin>mgetty ?
13:37<florin>I have that one
13:37<Magotari>I use port 9000, usually.
13:37<Magotari>mgetty? I am not sure if that one.
13:38<Magotari>Try just agetty
13:38<Magotari>I think the m in mgetty stands for modem.
13:38<florin>this is from inittab
13:38<Magotari>Ah, mingetty...
13:39<Magotari>Really, try agetty, it should work. I tried mingetty for something and I had a lot of trouble with it.
13:39|-|the_hydra [~mulyadi@] has joined #uml
13:39<Magotari>If agetty works, maybe then try other getty programs.
13:41<florin>let's see
13:42<florin>the fc5 fs works with mingetty well
13:43<florin>I'm almost sure that I forgot to install something but I can't figure what
13:52|-|tyler29 [] has quit [Ping timeout: 480 seconds]
13:52<florin>line_ioctl: tty0: ioctl TIOCLINUX called
13:52<florin>does this have anything to do ? :-ss
13:53<the_hydra>florin: excuse me, I came late ... whar's the problem?
13:53<florin>hi... I am trying to build a new linux root_fs (CentOS 5) ... I got it booting, it enters runlevel 3... but after starting syslog it just hangs and opens some xterms... however, these exterms are associated to the host system and not to the uml (guest)... anyone has any idea ?
13:54<florin>won't give me a serial console either, just tested :(
13:54|-|tyler29 [] has joined #uml
13:54<the_hydra>could you paste command you use to start that UML instance?
13:56<the_hydra>just that?
13:56<florin>shouldn't it go ?
13:56<Magotari>It should.
13:57<Magotari>As long as the root fs is named root_fs
13:57<florin>of course, it loads it
13:57<florin>INIT: Entering runlevel: 3
13:57<florin>Entering non-interactive startup
13:57<florin>Starting system logger: [ OK ]
13:57<florin>Starting kernel logger: [ OK ]
13:57<florin>and it hangs here
13:57<the_hydra>probably something related to SELinux?
13:57<florin>might be
13:57<the_hydra>or NPTL?
13:58<the_hydra>I don't remember the status of TLS (sorry, i mean this, not NPTL) support in recent UML
13:58<the_hydra>try to pas "selinux=0"
13:58<the_hydra>in kernel argument
13:59<florin>I renamed /lib/tls
13:59<Magotari>the_hydra: I booted real partitions using UML, and there was no problem. I think TLS is fine.
13:59<Magotari>Don't quote me on that though.
13:59<the_hydra>Magotari: ok
14:00<the_hydra>florin: could you attach gdb to UML and get the stack trace?
14:00<florin>let me try smth and I'll try to gdb then
14:00<the_hydra>can't guarantee if that could lead us to the real least we'll get a lead
14:01<the_hydra>florin: one more thing, is it SKAS0 or 3?
14:01<florin>no skas
14:01<the_hydra>no skas? sure?
14:01<florin>just my fc7 kernel
14:01<the_hydra>TT mode?
14:01<the_hydra>ok, that mean you probably use SKAS0
14:02<the_hydra>that's the default without kernel SKAS support
14:02<florin> .. SMPthis is my host
14:02<the_hydra>i see
14:03<florin>let me try with agetty and see if it has something to do it
14:03<the_hydra>florin: try to boot into runlevel 1
14:03<the_hydra>and see if it helps
14:03<the_hydra>we try "divide and conquer" style here..if you agree
14:04<the_hydra>my finger points to one of services right now
14:05<florin>you're right
14:05<florin>goes well in runlevel 1
14:06<Magotari>Funny that one.
14:06<florin>K89netplugd K89rdisc S10network S12syslog S13mcstrans S25netfs S99local
14:06<florin>stops at syslog
14:06<florin>I guess I'll disable mcstrans and see what it does
14:07<the_hydra>florin: at least you know where to go from now
14:07<the_hydra>florin: sorry, can't help further
14:07<florin>okies, thans a lot
14:09|-|kos_tom [] has joined #uml
14:09<the_hydra>hello thomas
14:10|-|flamingo [] has quit [Ping timeout: 480 seconds]
14:10<Magotari>Since many new people are here... I have a problem with UML relating to memsplit.
14:11<Magotari>Since I changed my memsplit in the host, all my guest kernels are busted.
14:11<Magotari>Even after I recompiled them for the right host memsplit.
14:11<Magotari>Any ideas?
14:11<the_hydra>what memsplit you use?
14:12<the_hydra>probably you reduce user space too much?
14:13<the_hydra>sorry I don't remember, that means user/kernel = 1:3?
14:13<the_hydra>or 3:1?
14:13<Magotari>Erm... Sec.
14:13|-|psy [] has quit [Ping timeout: 480 seconds]
14:13<Magotari>(X) 3G/1G user/kernel split (for full 1G low memory)
14:14<the_hydra>ok that one
14:14<Magotari>It used to be ( ) 3G/1G user/kernel split
14:15<Magotari>I have 1G memory on this computer.
14:15<the_hydra>maybe that makes some variables/structs put into high address "stomped"
14:15|-|psy [] has joined #uml
14:15<the_hydra>since that means, a bit lesser user address space
14:15<the_hydra>note ..." a bit"
14:15<the_hydra>so not much
14:15<the_hydra>kos_tom: pls CMIIW... maybe you know better
14:16<Magotari>free used to give me 900MB, now it gives me my 1G. I am puzzled why it would reduce user adress space.
14:16<kos_tom>the_hydra: hm, sorry, I don't have time now.
14:16<Magotari>I see extra megabytes.
14:16<Magotari>I guess memory is not something I studied, so I don't know.
14:17<the_hydra>imagine this: by doing this, your addres space is shrinked
14:17<the_hydra>by choosing the normal 3G/1G, you have address 0-3G in hand
14:17<the_hydra>now, you have 0 - 3G minus something
14:18<Magotari>Erm. But the box has only 1G. Not a byte more.
14:18<the_hydra>in the other hand (without going deeper into kernel stuffs), 3G/1G (ptimal) makes your 1G RAm capacity all addressable at once
14:20<the_hydra>Magotari: actually it's about the way Linux kernel address RAM, not really a matter of probing RAM capacity
14:20<Magotari>This seems counterintuitive, that I see more memory, but there is less adressable... I guess I must read up on this more, because I really don't understand.
14:20<the_hydra>Magotari: if you have some basic kernel understanding, I can continue the explanation
14:21<Magotari>the_hydra: Try it. I am not too good, but I want to improve, so I want to listen.
14:21<the_hydra>ok, I'll try to be as simple as possible
14:21<the_hydra>normally, we go with 3G/1G (not the "optimal" one)
14:21<Magotari>Yeah, the default.
14:21<the_hydra>that means, 3G space for user space, 1G for kernel
14:21<the_hydra>in this 1G room, not all are usable
14:21<Magotari>Question here.
14:22<Magotari>I have 1G. So who gets the memory that I do have? Userspace or kernel?
14:22<the_hydra>kernel needs to map the RAM first
14:23<Magotari>Ok. Gotcha.
14:23<Magotari>So not all is usable... And?
14:23[~]Magotari is very curious now.
14:23<the_hydra>so in other word, before your RAM can be used by user application, in needs to be mapped inside kernel address space
14:23<the_hydra>1G seems fit, right?
14:23<the_hydra>but in reality is not
14:24<Magotari>A trick of some kind?
14:24<the_hydra>since the upper 128 address space is reserved for special needs
14:24<the_hydra>so you are left with 1G - 128 MB
14:24<Magotari>Right. The 100 megabytes that I did not see before.
14:24<the_hydra>= 896 MB
14:24<Magotari>Close enough, yeah.
14:24<the_hydra>that's why you see that 900 MB
14:24<Magotari>Can you give a few examples of special needs?
14:24<the_hydra>with "optimal" option, it moves "the border" a bit
14:24<the_hydra>Magotari: that includes mapping for so called "high memory"
14:25<the_hydra>a temporary mapping so kernel can address >1G RAM
14:25<florin>hmm... I attached gdb and all looks good I think... maybe I just need to install something and I'm too dumb
14:25<the_hydra>florin: no interesting stack trace?
14:25<Magotari>the_hydra: CONFIG_NOHIGHMEM=y
14:26<the_hydra>Magotari: i see
14:26<the_hydra>Magotari: now, because of this adjustment, you don't have the full 3G space
14:26<the_hydra>but instead 3G minus something
14:26<Magotari>Yes, because we moved the border. So far I get it.
14:27<the_hydra>so if something is mapped near 3G in user address space, it is no longer applicable
14:27<the_hydra>i can't recall the exact "shifting"
14:27<the_hydra>Magotari: ok at least you get the idea
14:28<Magotari>I can see why UML compiled for the not optimal memsplit would have a problem.
14:28<Magotari>But I did compile it for the optimal host split.
14:28<Magotari>Shouldn't that fix it?
14:28<florin>can I paste the stack here ?
14:28<the_hydra>not idea Magotari, so far I always use normal mem split
14:29<the_hydra>florin: better use pastebin
14:29<florin>one sec
14:29<the_hydra>Magotari: try to compile your UML kernel for 2/2 mem split
14:29<the_hydra>Magotari: see if it helps
14:30<the_hydra>Magotari: check jdike's note about dealing with non default mem split
14:30<Magotari>Where is it?
14:30<Magotari>I looked for such things, could not find them.
14:31|-|flamingo [] has joined #uml
14:31<florin>would it be that new_thread *( ?
14:31<Magotari>Compiling. Thanks for help. We shall see if that helps.
14:31<the_hydra>Magotari: old uml website
14:32<the_hydra>florin: checking...
14:33<the_hydra>Magotari: good luck
14:33<Magotari>"The fix for this is to rebuild UML from source after enabling CONFIG_HOST_2G_2G"
14:33<Magotari>I have my 2/2 kernel ready, testing...
14:34<Magotari>Hey, it booted.
14:35<Magotari>So far it seems, yes.
14:35<the_hydra>phew :)
14:35<the_hydra>florin: sec
14:35<Magotari>I think it is a bug that it does not work with the 1G-optimal host split.
14:36<the_hydra>Magotari: very likely
14:36<the_hydra>Magotari: report to jeff immediately
14:36<Magotari>I did.
14:36<the_hydra>Magotari: with as much detail as possible
14:36<Magotari>I sent an email to the UML-devel list.
14:36<Magotari>About 4 days ago. Or was it three?
14:36<the_hydra>florin: what version of guestkernel you use?
14:36<Magotari>It got ignored.
14:37<Magotari>I was not too clear, but still...
14:37<the_hydra>Magotari: to jdike directly
14:37<Magotari>Ok, I will do it, tomorrow. I am quite sleep deprived.
14:37<the_hydra>Magotari: and pls visit this channel again the other day...he always visits here when not too busy
14:37<Magotari>I have plenty of problems with UML still, so I can keep him busy. :)
14:38<Magotari>For example: When something goes wrong I have to "killall linux"
14:38<Magotari>Because three threads stay on, and keep the resources tied.
14:38<the_hydra>i see
14:39<the_hydra>i hope that's not TT mode
14:39<florin> this one
14:39<the_hydra>florin: hmm
14:39<Magotari>I am using the rc7 from 2.6.23
14:39<Magotari>I don't think it even supports TT.
14:39<the_hydra>florin: i am kinda scared with that -rc2 suffix
14:40<Magotari>Both for host and guests.
14:40<the_hydra>florin: could you use the stable versioninstead? or -mm?
14:40<florin>-mm ?
14:40<the_hydra>-mm patchset
14:40<the_hydra>andrew morton's
14:41<the_hydra>a transit place for patches before goin' into mainline
14:41<the_hydra>UML patches usually go there 1st
14:41<Magotari>The kernel patchset for hardcore people who don't mind their system blowing up.
14:41<the_hydra>especially the recent fixes
14:41<the_hydra>Magotari: thank you for the straighforward explanation :D
14:42<florin>I'll try :)
14:42<Magotari>Yeah, I have been watching the kernel devs for a while now. One day I hope to be one of them. After I learn C. *sigh*
14:42<the_hydra>florin: sounds better IMHO
14:43<the_hydra>Magotari: if you're consistent, you can learn it
14:43<the_hydra>Magotari: i started from early 2003
14:43<florin>Magotari, same here, I know C but the kernel is a little to high for me :P
14:43<florin>well, maybe I just don't have enough time :P
14:43<florin>and experience
14:43<Magotari>We shall see.
14:44<the_hydra>florin: one recipe...don't bring 100% user space understanding into kernel space
14:44<the_hydra>or you'll completely confused
14:44<florin>no, no
14:44<the_hydra>and vice versa
14:44<florin>I know what programming means :P
14:44<florin>trust me
14:49<the_hydra>stop when starting syslog, right?
14:50<florin>gdb looks exactly the same as well
14:50<florin>let me disable syslog, lol
14:50<Magotari>the_hydra: Can syslog access the log directory?
14:50<the_hydra>Magotari: you mean, accessing /var/log?
14:50<Magotari>I found a bug in syslog-ng where things went crazy if it could not. I hear it is a common problem too.
14:50<the_hydra>ehm...not sure, but logically it can
14:50<Magotari>Ergh. I meant florin.
14:51<Magotari>Sorry, wrong person. I am seriously sleep deprived here.
14:51<the_hydra>Magotari: better take a rest
14:51<Magotari>I wish I could.
14:51<Magotari>I just got UML running, and I am to go to sleep?
14:51<florin>it's not syslog-ng and it should be fine, let me check anyway
14:51|-|psy [] has quit [Ping timeout: 480 seconds]
14:52<Magotari>If I manage to boot my other partition using it, I can start compiling right away.
14:52<Magotari>This is why UML is my favourite, using a physical partition for rootfs is easy. :)
14:52|-|psy [] has joined #uml
14:54<the_hydra>Magotari: hm i see
14:54<florin>disabled all services
14:54<florin>let's do it the bad way, lol
14:55<florin>takes a little long time to start udev as well :-ss
14:55<florin>Remounting root filesystem in read-write mode: [ OK ]
14:55<florin>Mounting local filesystems: [ OK ]
14:55<florin>Enabling /etc/fstab swaps: [ OK ]
14:55<florin>INIT: Entering runlevel: 3
14:55<florin>Entering non-interactive startup
14:56<the_hydra>eh wait, you said "enabling swap"?
14:56<the_hydra>and where's the swap anyway?
14:56<florin>you're right
14:56<florin>well, it should skip it anyway, I have no swaps in fstab
14:56<the_hydra>made one...with dd or any thing
14:56<florin>let me add one
14:57<the_hydra>sure none are declared in fstab?
14:57<the_hydra>if yes, then simple ignore my above warning
14:58<florin># cat /etc/fstab
14:58<florin>/dev/ubda / ext3 defaults 0 1
14:58<florin>proc /proc proc defaults 0 0
14:58<florin>none /dev/pts devpts gid=5,mode=620 0 0
14:58<the_hydra>and mem?
14:59<the_hydra>i am just afraid it's taken up already
14:59<florin>as I said, I'm dumb
14:59<the_hydra>there are many factor here... safely passing runlevel 1 could mean:
14:59<the_hydra>1. less service, less memory consumption
14:59<the_hydra>2. faulty services
15:00<the_hydra>3. accessing certain device files which aren't present
15:00<the_hydra>and so on
15:00<florin>probably it's a device
15:00<florin>how much memory it gets by default ?
15:00<florin># ./kernel32- mem=1024M and got the same
15:01<the_hydra>i also forgot, try to run uml kernel with -h option
15:01<the_hydra>i see
15:01<the_hydra>this time, where it ended ?
15:01<the_hydra>still @ syslog?
15:01<florin>I disabled all services that were installed
15:01<florin>INIT: Entering runlevel: 3
15:01<florin>Entering non-interactive startup
15:02<florin>here it stops
15:02<florin>let's see devices
15:02<the_hydra>i smell console devices problem
15:03<the_hydra>no /sys , ?
15:03<the_hydra>hey, proc isn't declared like you did
15:03<the_hydra>none /proc proc defaults 0 0
15:03<florin> .. /sys is not required for uml afaik
15:04<the_hydra>check /proc again
15:04<florin>set to none
15:05|-|tyler29 [] has quit [Ping timeout: 480 seconds]
15:06<florin>[42949373.420000] line_ioctl: tty0: ioctl KDSIGACCEPT called
15:06<florin>that means it loaded tty0 I think...
15:06<the_hydra>i wish that's not ioctl related bug
15:07<the_hydra>IIRC tty starts on 1?
15:08<florin>/dev/zero /dev/random /dev/urandom ? are these required ?
15:08<the_hydra>check your inittab
15:08<the_hydra>in guest, you mean?
15:08<the_hydra>imo no
15:08<the_hydra>but who knows...
15:09<florin>1:2345:respawn:/sbin/mingetty tty1
15:09<the_hydra>what happened?
15:09<florin>you're right, I don't have tty0
15:09<florin>in inittab
15:09<florin>starts at 1
15:09<the_hydra>you don't need to IMHO
15:09<the_hydra>that's the confusion starts, who connect to tty0?
15:10<florin>no idea, that's what it says after mounted root
15:13<florin>here's the funny thing... until now I was able to type in the terminal, while it was hanging there... now I'm not... I don't know when this changed :-s
15:14<the_hydra>I can't offer another solution right now...pls report to jeff ASAP
15:14<florin>jeff*@* ? :)
15:15<florin>I'm quite new around here
15:15<the_hydra>florin: i am fairly sure his e-mail address is mentioned on UML website
15:15|-|tyler29 [] has joined #uml
15:15<florin>I'll go and sleep a little now, see ya tomorrow :P
15:15|-|tyler29 [] has quit [Remote host closed the connection]
15:15|-|florin changed nick to florin-not-her
15:16|-|florin-not-her changed nick to florin-not-here
15:16<the_hydra>Magotari: seems like I have to go offline too soon
15:17<Magotari>the_hydra: Sure.
15:17<Magotari>Thanks for the warning.
15:17<Magotari>I will report the 1G thing to jeff tomorrow.
15:18|-|tyler29 [] has joined #uml
15:19<flamingo>has anyone used iomem before? I'm not seeing the writes from the host immediately available in the guest
15:21<flamingo>from the docs available about UML's iomem facility, it seems to be that the writes from both the host and the guest are immediately available to each other
15:21<flamingo>anyone has any ideas?
15:22<the_hydra>probably that's done in async style
15:25<flamingo>the_hydra, is there a way to specifically set a sync option?
15:27<flamingo>I did a linux --help, there does not seem to be any option to this effect
15:27<the_hydra>i don't have a clue either, sorry
15:28<the_hydra>i just recall that UML support async operations now
15:28<flamingo>the_hydra, as I see
15:29<the_hydra>what do you do @ host?
15:29<flamingo>the_hydra, I'm writing to the file
15:29<flamingo>and in the UML. I try to read it... sort of simulating a device (a classic example)
15:30<the_hydra>how do you open the file?
15:30<the_hydra>have you tried using O_SYNC?
15:31<flamingo>inside the UML, I have my code in the read() of the device driver
15:31<flamingo>so, the file has been opened for me already
15:33<flamingo>in any case, there does not seem to be a way to make the host writes immediately visible to the UML since the UML has the file contents in its buffer cache
15:34<the_hydra>sounds so
15:34<the_hydra>you may try O_DIRECT
15:35<flamingo>the_hydra, where? in open() call?
15:37<flamingo>again, since I'm not opening the file, I can't specify any open flags. I can try using them in my test code, will try now
15:49|-|the_hydra [~mulyadi@] has quit [Quit: using sirc version 2.211+KSIRC/1.3.10]
15:49|-|psy [] has quit [Ping timeout: 480 seconds]
15:51|-|psy [] has joined #uml
15:57|-|tyler29 [] has quit [Ping timeout: 480 seconds]
16:20|-|psy [] has quit [Ping timeout: 480 seconds]
16:20|-|psy [] has joined #uml
16:21|-|netdrake_ [] has joined #uml
16:22|-|netdrake [] has quit [Remote host closed the connection]
16:29|-|GoGi [] has quit [Remote host closed the connection]
16:31|-|kos_tom [] has quit [Remote host closed the connection]
16:40|-|alb [] has joined #uml
16:40|-|Magotari_ [] has joined #uml
16:41|-|Netsplit <-> quits: VS_ChanLog, Baltam, newbie, Magotari, albertito
16:42|-|Baltam [] has joined #uml
16:50|-|psy [] has quit [Ping timeout: 480 seconds]
16:50|-|psy [] has joined #uml
16:58|-|psy [] has quit [Ping timeout: 480 seconds]
17:00|-|psy [] has joined #uml
17:00|-|psy [] has quit []
17:09|-|psy [] has joined #uml
17:09<psy>are these pages still live somewhere? (which is a cache of )?
17:44|-|psy [] has quit [Ping timeout: 480 seconds]
17:44|-|psy [] has joined #uml
18:09|-|Guest2 changed nick to ElectricElf
18:10|-|ElectricElf changed nick to Guest419
18:10|-|Guest419 [] has quit [Quit: Reconnecting]
18:10|-|Electric1lf [] has joined #uml
18:10|-|Electric1lf changed nick to electricElf
18:10|-|electricElf changed nick to ElectricElf
18:37|-|psy [] has quit [Ping timeout: 480 seconds]
18:37|-|psy [] has joined #uml
18:45|-|psy [] has quit [Ping timeout: 480 seconds]
18:46|-|psy [] has joined #uml
18:53|-|flamingo [] has quit [Ping timeout: 480 seconds]
19:02|-|psy [] has quit [Ping timeout: 480 seconds]
19:03|-|psy [] has joined #uml
19:20|-|psy [] has quit [Ping timeout: 480 seconds]
19:29|-|VS_ChanLog [] has joined #uml
20:06|-|flamingo [] has joined #uml
22:58|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:58|-|VS_ChanLog [] has joined #uml
---Logclosed Mon Oct 01 00:00:50 2007