Back to Home / #uml / 2007 / 09 / Prev Day | Next Day
#uml IRC Logs for 2007-09-28

---Logopened Fri Sep 28 00:00:44 2007
00:35|-|flamingo [] has left #uml [leaving]
01:29|-|rlb [] has joined #uml
01:31<rlb>I've still had no luck with uml and uml_switch. Any idea why I might see this during boot:
01:31<rlb>Choosing a random ethernet address for device eth0
01:31<rlb>Netdevice 0 ... : daemon backend (uml_switch version 3) - unix:/var/run/uml-utilities/uml_switch.ctl
01:31<rlb>but still get an error when running "ifconfig eth0"?
01:31<rlb>"eth0: error fetching interface information: Device not found"
03:58|-|tyler29 [] has joined #uml
04:41|-|tyler29 [] has quit [Read error: Connection reset by peer]
04:59|-|tyler29 [] has joined #uml
05:24|-|tyler29 [] has quit [Ping timeout: 480 seconds]
05:41|-|tyler29 [] has joined #uml
07:21|-|tyler29 [] has quit [Ping timeout: 480 seconds]
07:28|-|psy [] has joined #uml
07:44|-|tyler29 [] has joined #uml
07:44|-|ram [] has joined #uml
08:28|-|dang [] has joined #uml
09:16|-|psy [] has quit [Ping timeout: 480 seconds]
09:16|-|jdike [] has joined #uml
09:16<jdike>Hi guys
10:18|-|psy [] has joined #uml
10:22|-|hfb [] has joined #uml
10:22<dgraves>WHO'S YOUR DADDY!!!!!
10:24<jdike>Oh man
10:25<jdike>who has his medication
10:35|-|psy [] has quit [Ping timeout: 480 seconds]
10:39|-|psy [] has joined #uml
10:54|-|psy [] has quit [Read error: Operation timed out]
11:00|-|flamingo [] has joined #uml
11:03<flamingo>I'm trying to compile a driver which I wrote. This driver includes <sys/time.h> and <signal.h>. The compiler is not able to find the headers whereas another driver cow_user.c which happens to be in the same directory includes those headers and compiles just fine. Is there any change that I have to do in the Makefile to tell the compiler about the location of the headers?
11:05<jdike>you have to decide whether that's a kernel file or a libc file
11:06<flamingo>my driver would go as a part of the kernel
11:06<jdike>um no
11:06<jdike>UML is a mix of kernel stuff and libc stuff
11:06<jdike>you need to decide whether this file is part of the kernel side of UML or the libc side
11:07<flamingo>I'm trying to use the timer in the driver. So that would mean, inside the UML, this would be the libc header
11:07<jdike>OK, so you're making system calls to the host from here, and you need the libc headers
11:08<flamingo>from here == from UML, yes
11:08<jdike>from this file
11:08<flamingo>I want the timer to be controlled by the UML, basically a setitimer()
11:09<flamingo>and nt the host timer
11:09<jdike>OK, that's kernel functionality
11:09<jdike>why do you want sys/time.h and signal.h?
11:09<flamingo>ah I see
11:10<flamingo>I'm trying to implement polling in the driver
11:10<jdike>doesn't answer the question
11:10<flamingo>by using the SIGALRM interrupts
11:10<jdike>you don't need anything from the host in order to implement polling
11:11<flamingo>yes, right and I'm not using anything from the host. I'm just trying to compile it outside of the UML (since I don't have libc etc. in UML yet) and pulling the .ko into the UML
11:12<jdike>you don't need sys/time.h and signal.h then
11:12<flamingo>um, what do I use to use UML kernel's timers?
11:13<jdike>look at the *timer.h files in include/linx
11:13<flamingo>these include/linux/* are for the UML functionality and <sys/time.h> is for the host functionality you mean?
11:14<jdike>include/linux/* is for kernel stuff
11:14<jdike>/usr/include/* is for libc stuff
11:14<flamingo>ah ok, got it
11:14|-|kokoko1 [~Slacker@] has quit [Quit: Leaving]
11:14<jdike>when you want to talk to the host, as you might need to do with your pipe, you'll need libc
11:15<jdike>and that will need to be in a different file
11:15<jdike>which will be compiled differently
11:16<flamingo>ok, I just looked at the include/linux/time.h, it has do_setitimer() is that what I have to call?
11:16<jdike>I doubt it
11:16<jdike>that's the actual system call
11:16<flamingo>there is no setitimer() function in include/linux/*
11:17<flamingo>(I just grepped for it)
11:19|-|kokoko1 [~Slacker@] has joined #uml
11:25|-|ftumch [] has quit [Quit: Goodbye.]
12:03<flamingo>When I thought I had to use setitimer(), I was thinking in a different (wrong) direction altogether. LDD3 helped me understand that kernel timers are implemented by using add_timer() method
12:07|-|tyler29 [] has quit [Ping timeout: 480 seconds]
12:26|-|flamingo [] has quit [Ping timeout: 480 seconds]
12:27|-|tyler29 [] has joined #uml
12:29|-|horst [] has joined #uml
12:30|-|horst [] has quit [Remote host closed the connection]
12:39<peterz>jdike: so, I'm having trouble with net_kern.c in -mm. uml_net_interrupt() is being called, but no data is received. Would any of these patches you posted fix these symptoms?
12:57<jdike>so basically, no network?
13:03<peterz>it usually happens during heavy NFS traffic
13:03<peterz>at which point the box wedges because it has all these requests outstanding
13:04<peterz>I've stared at it a little, but I keep loosing my way in that lower part of the network stack
13:04<peterz>I think its the recent NAPI work
13:04<peterz>I know its not any of the uml patches in -mm
13:05<peterz>(since I reverted them all)
13:05<peterz>a well, I'll just continue poking at it
13:18|-|ram [] has quit [Ping timeout: 480 seconds]
13:18|-|netdrake [netdrake@2001:5c0:904b:ffff::1] has quit [Read error: Connection reset by peer]
13:19|-|netdrake [netdrake@2001:5c0:904b:ffff::1] has joined #uml
13:29|-|ram [] has joined #uml
13:54|-|netdrake [netdrake@2001:5c0:904b:ffff::1] has quit [Ping timeout: 480 seconds]
13:55<jdike>peterz, but the interrupt routine is called?
13:56<jdike>and it reads no data?
13:56<peterz>it does, it seems to allocate an skb and stick it in the network thing, I traced it as far as netif_rx
13:57<peterz>what I did was: printk("."); in uml_net_interrupt()
13:57<peterz>and run: # ping host
13:57<peterz>then start the NFS traffic that kills it
13:58|-|netdrake [] has joined #uml
13:58<peterz>the '.' keep comming
13:58<peterz>ping replies stop
13:58<peterz>nfs freezes
13:59|-|tyler29 [] has quit [Ping timeout: 480 seconds]
14:00<peterz>anyway, 9pm on a friday,.. gtg, will get back on this later
14:15|-|Urgleflogue [~plamen@] has joined #uml
14:15|-|tyler29 [] has joined #uml
14:22|-|flamingo [] has joined #uml
15:11|-|tyler29 [] has quit [Remote host closed the connection]
15:15|-|tyler29 [] has joined #uml
15:25|-|dang [] has quit [Quit: Leaving.]
15:43|-|AllenJB [] has joined #uml
15:44|-|AllenJB [] has left #uml []
15:57|-|tyler29 [] has quit [Ping timeout: 480 seconds]
16:01<rlb>I've still had no luck with eth0=daemon and uml_switch. Any idea why I might see "Choosing a random ethernet address for device eth0... daemon backend..." during boot, but see "eth0: error fetching interface information: Device not found" when I run "ifconfig eth0"?
16:02<rlb>The unix:... location appears correct.
16:03<jdike>ifconfig eth0 all by itself?
16:03<rlb>jdike: right
16:03<jdike>or ifconfig eth blah up?
16:03<rlb>I was just trying to see if it was there.
16:03<jdike>what does ifconfig -a say?
16:05<rlb>Hmm, interesting -- it does show up, but as eth25.
16:05<rlb>(hopefully, that will be consistent)
16:06<jdike>no, it won't
16:06<jdike>do you think eth0 ... eth24 were consistent?
16:06<rlb>jdike: I never saw eth0 ... eth24.
16:06<jdike>they were there
16:06<rlb>(and I have no idea why it's 25 now)
16:06<rlb>Though it sounds like it's just incrementing.
16:07<rlb>Or at least it sounds like that's what you're suggesting.
16:07<jdike>what's happening is your distro is keeping track of MACs and making sure that the name of the interface with a given MAC doesn't change
16:07<rlb>jdike: ahh, udev.
16:07<rlb>jdike: I can fix that.
16:07<jdike>but, it's seeing a new MAC every boot, so it gets a new name
16:08<rlb>jdike: right -- I'll just fix that.
16:08<rlb>jdike: thanks.
16:08<rlb>I suppose I can either adjust udev's rules or specify the MAC.
16:09<rlb>Anyway, for now I'm just happy to know what's going on. Thanks again.
16:09<rlb>(I should have thought of "ifconfig -a"...)
16:12|-|tyler29 [] has joined #uml
16:16|-|besonen_mobile_ [] has quit [Quit: Leaving]
16:40|-|besonen_mobile [] has joined #uml
16:52|-|flamingo [] has quit [Read error: Operation timed out]
16:59|-|tyler29 [] has quit [Remote host closed the connection]
17:03|-|tyler29 [] has joined #uml
17:26|-|krau [~cktakahas@] has quit [Quit: Varei!!!]
18:06|-|tyler29 [] has quit [Ping timeout: 480 seconds]
18:16|-|jdike [] has quit [Quit: Leaving]
18:23|-|newbie [] has joined #uml
18:24<newbie>Hi, I have a question that I've been trying to work out all day. I would like to run uml with all network configuration specified as a boot option so I can use it to boot into a kickstart iso and bootstrap a locally customized version of fc3
18:25|-|hfb [] has quit [Quit: Leaving]
18:26<newbie>Any suggestions - I seems clear to me with tuntap I need static ips for both sides and unclear what all the empty commas in the eth0=tuntap,,,ipaddr param is for
18:26<newbie>where can I find more docs on that
18:27<newbie>every page I can find on and off the wiki seem to recite the magic incantation above as a rote recipe to get networking going
18:27<newbie>if someone can even point me at the right source file in the kernel source code, I can poke around and suspect it's documented in the comments
18:44<ds2>not sure but look in arch/um/drivers
18:44<ds2>it looks like they are handled around there
18:46<ds2>and arch/um/os-Linux/drivers
18:48<newbie>hmm ok
18:48<newbie>let me look
19:11|-|hfb [] has joined #uml
19:12<newbie>stumped for now I guess ...try again next week
19:37|-|hfb [] has quit [Quit: Leaving]
20:46|-|flamingo [] has joined #uml
20:47|-|flamingo [] has quit []
21:30|-|flamingo [] has joined #uml
21:33<flamingo>I have a file on the host which is iomem'd. Inside the UML, I do q_buf = find_iomem(name, size) of that file. Apparently, my writes to q_buf is not getting reflected in the file. Do I have to do anything extra to get it reflected on the host?
21:37<flamingo>I'm just doing a copy_from_user() into q_buf - this is what some of the sample code which I've seen (mmapper_kern.c for example) does.
22:01|-|hfb [~hfb@] has joined #uml
22:58|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:58|-|VS_ChanLog [] has joined #uml
---Logclosed Sat Sep 29 00:00:19 2007