user-mode-linux-user

Re: [uml-user] when running multiple UML instances, seems they are running 1 by 1

[Riccardo M. <ric...@gm...>]

Hi Ning Ji,

On Thu, Jan 26, 2012 at 18:14, ning ji <ni...@ho...> wrote:
> Basically you see the messages 1 by 1 from different instances,
> not that smooth.
>
> Though these are processes running in a ubuntu, should be multi-threaded.


I'm not sure what you mean by "see the messages 1 by 1", but each UML
instance runs as a separate Linux process (several "interwoven"
processes, actually), so multi-threading does not play any role here.

Best regards,
Riccardo

[uml-user] when running multiple UML instances, seems they are running 1 by 1

[ning ji <ni...@ho...>]

Basically you see the messages 1 by 1 from different instances,not that smooth.
Though these are processes running in a ubuntu, should be multi-threaded.
Is this expected to be so or it's just running slow ?
thanks ! 		 	   		  

Re: [uml-user] SKAS host patches and wiki

[richard -r. w. <ric...@gm...>]

On Thu, Jan 19, 2012 at 7:41 PM, Antoine Martin <an...@na...> wrote:
> On 01/19/2012 10:01 PM, richard -rw- weinberger wrote:
>> On Thu, Jan 19, 2012 at 3:23 PM, Nick Cripps <nic...@gm...> wrote:
>>> I saw this thread:
>>> http://sourceforge.net/mailarchive/forum.php?thread_name=CAFLxGvz2wdfRtryzJ1t0G11rxzOmhdNEVi6A%3DXS5uLG0PA7VAQ%40mail.gmail.com&forum_name=user-mode-linux-user
>>> which seems to indicate that SKAS3 isn't being maintained but mentions
>>> SKAS0. Can anyone tell me what SKAS0 is, what the disadvantages
>>> compared to SKAS3 are and, where I can get a patch for it?
>>
>> SKAS0 is in mainline and it's default for UML.
>> (The old tt-mode is gone.)
>>
>> SKAS0 is mostly SKAS3 but completely different implemented.
>>
>> And yeah, it's time to clean up the UML website.
>> I have write-access to http://user-mode-linux.sourceforge.net/.
>> Maybe we can combine it with http://uml.devloop.org.uk/.
>> Antoine, what do you think?
> Sure, it might be worth adding a FAQ, that sort of thing.
> I could setup a trac/wiki if that's any help?
>

I'm not sure what the best (and easiest) solution is.

We could use a hosted wiki from sourceforge.net and redirect
http://user-mode-linux.sourceforge.net/index.html to it,
such that all old links to http://user-mode-linux.sourceforge.net/* don't break.

-- 
Thanks,
//richard

Re: [uml-user] Running 32bit binaries in a 64bit UML instance

[richard -r. w. <ric...@gm...>]

On Sat, Jan 21, 2012 at 12:56 AM, Nick Cripps <nic...@gm...> wrote:
> Hi,
> When running 'make menuconfig ARCH=uml' there is no 'Executable file
> formats / Emulations' menu. I guess this makes sense since I guess
> that only binaries that can be run on the host kernel can be run from
> UML. Does this mean that if a x86_64 host kernel supports IA32
> emulation then 32bit binaries can be run from inside the UML instance
> as well? If so, does anything need to be enabled in the UML kernel
> config to support it?
>

x86_64 UML has no support for IA32 emulation.

-- 
Thanks,
//richard

[uml-user] Running 32bit binaries in a 64bit UML instance

[Nick C. <nic...@gm...>]

Hi,
When running 'make menuconfig ARCH=uml' there is no 'Executable file
formats / Emulations' menu. I guess this makes sense since I guess
that only binaries that can be run on the host kernel can be run from
UML. Does this mean that if a x86_64 host kernel supports IA32
emulation then 32bit binaries can be run from inside the UML instance
as well? If so, does anything need to be enabled in the UML kernel
config to support it?

Cheers,
Nick

Re: [uml-user] SKAS host patches and wiki

[Antoine M. <an...@na...>]

On 01/19/2012 10:01 PM, richard -rw- weinberger wrote:
> On Thu, Jan 19, 2012 at 3:23 PM, Nick Cripps <nic...@gm...> wrote:
>> I saw this thread:
>> http://sourceforge.net/mailarchive/forum.php?thread_name=CAFLxGvz2wdfRtryzJ1t0G11rxzOmhdNEVi6A%3DXS5uLG0PA7VAQ%40mail.gmail.com&forum_name=user-mode-linux-user
>> which seems to indicate that SKAS3 isn't being maintained but mentions
>> SKAS0. Can anyone tell me what SKAS0 is, what the disadvantages
>> compared to SKAS3 are and, where I can get a patch for it?
> 
> SKAS0 is in mainline and it's default for UML.
> (The old tt-mode is gone.)
> 
> SKAS0 is mostly SKAS3 but completely different implemented.
> 
> And yeah, it's time to clean up the UML website.
> I have write-access to http://user-mode-linux.sourceforge.net/.
> Maybe we can combine it with http://uml.devloop.org.uk/.
> Antoine, what do you think?
Sure, it might be worth adding a FAQ, that sort of thing.
I could setup a trac/wiki if that's any help?

Antoine

Re: [uml-user] SKAS host patches and wiki

[richard -r. w. <ric...@gm...>]

On Thu, Jan 19, 2012 at 5:40 PM, Nick Cripps <nic...@gm...> wrote:
> Thanks for the quick reply. What options do I need to have enabled in
> the host kernel to support SKAS0?
>

There is no need to enable SKAS0 in the host kernel.

-- 
Thanks,
//richard

P.s: Please don't top-post.

Re: [uml-user] SKAS host patches and wiki

[Nick C. <nic...@gm...>]

Thanks for the quick reply. What options do I need to have enabled in
the host kernel to support SKAS0?

Cheers,
Nick

On 19 January 2012 15:01, richard -rw- weinberger
<ric...@gm...> wrote:
> On Thu, Jan 19, 2012 at 3:23 PM, Nick Cripps <nic...@gm...> wrote:
>> I saw this thread:
>> http://sourceforge.net/mailarchive/forum.php?thread_name=CAFLxGvz2wdfRtryzJ1t0G11rxzOmhdNEVi6A%3DXS5uLG0PA7VAQ%40mail.gmail.com&forum_name=user-mode-linux-user
>> which seems to indicate that SKAS3 isn't being maintained but mentions
>> SKAS0. Can anyone tell me what SKAS0 is, what the disadvantages
>> compared to SKAS3 are and, where I can get a patch for it?
>
> SKAS0 is in mainline and it's default for UML.
> (The old tt-mode is gone.)
>
> SKAS0 is mostly SKAS3 but completely different implemented.
>
> And yeah, it's time to clean up the UML website.
> I have write-access to http://user-mode-linux.sourceforge.net/.
> Maybe we can combine it with http://uml.devloop.org.uk/.
> Antoine, what do you think?
>
> --
> Thanks,
> //richard

Re: [uml-user] SKAS host patches and wiki

[richard -r. w. <ric...@gm...>]

On Thu, Jan 19, 2012 at 3:23 PM, Nick Cripps <nic...@gm...> wrote:
> I saw this thread:
> http://sourceforge.net/mailarchive/forum.php?thread_name=CAFLxGvz2wdfRtryzJ1t0G11rxzOmhdNEVi6A%3DXS5uLG0PA7VAQ%40mail.gmail.com&forum_name=user-mode-linux-user
> which seems to indicate that SKAS3 isn't being maintained but mentions
> SKAS0. Can anyone tell me what SKAS0 is, what the disadvantages
> compared to SKAS3 are and, where I can get a patch for it?

SKAS0 is in mainline and it's default for UML.
(The old tt-mode is gone.)

SKAS0 is mostly SKAS3 but completely different implemented.

And yeah, it's time to clean up the UML website.
I have write-access to http://user-mode-linux.sourceforge.net/.
Maybe we can combine it with http://uml.devloop.org.uk/.
Antoine, what do you think?

-- 
Thanks,
//richard

[uml-user] SKAS host patches and wiki

[Nick C. <nic...@gm...>]

Hi,
If people weren't aware, the wiki linked from the UML website seems to be down.

I'm trying to build a kernel with the host/SKAS patch. There is a link
on the website to SKAS3 patches for 2.6.23 or SKAS4 patches for 2.6.24
but, ideally, I'd like to use a newer kernel to host UML. Does anyone
know if patches exist for 2.6.39 or later?

I saw this thread:
http://sourceforge.net/mailarchive/forum.php?thread_name=CAFLxGvz2wdfRtryzJ1t0G11rxzOmhdNEVi6A%3DXS5uLG0PA7VAQ%40mail.gmail.com&forum_name=user-mode-linux-user
which seems to indicate that SKAS3 isn't being maintained but mentions
SKAS0. Can anyone tell me what SKAS0 is, what the disadvantages
compared to SKAS3 are and, where I can get a patch for it?

Cheers,
Nick Cripps

Re: [uml-user] the filesystem (fedoracore5) is OK for Ubuntu 1004 (2.6.32)?

[Antoine M. <an...@na...>]

On 01/16/2012 03:47 PM, Yiming wrote:
> Hi, all
> 
> I am trying to using uml on Ubuntu 1004 (2.6.32). I was wondering
> whether I can use the filesystem (fedoracore5) on official website as
> the rot fs for my uml. Thank you!
I assume you followed the link from the completely out-of-date uml
sourceforge page. I have removed the download this points to because
running a distribution which has not had any security updates for over 5
years is dangerous and unhelpful. You can find newer root_fs here:
http://fs.devloop.org.uk/
Note that they are no longer built specifically for UML, so some may
have issues when running under UML (bootsplash/init related problems)
The scripts are included if you want to re-generate them. For Fedora,
you may want to look at febootstrap, and debootstrap for Debian based
distros.

Cheers
Antoine
> 
> Best,
> 
> Yiming
> 
> 
> 
> ------------------------------------------------------------------------------
> RSA(R) Conference 2012
> Mar 27 - Feb 2
> Save $400 by Jan. 27
> Register now!
> http://p.sf.net/sfu/rsa-sfdev2dev2
> 
> 
> 
> _______________________________________________
> User-mode-linux-user mailing list
> Use...@li...
> https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user

[uml-user] the filesystem (fedoracore5) is OK for Ubuntu 1004 (2.6.32)?

[Yiming <sd...@gm...>]

Hi, all

 

I am trying to using uml on Ubuntu 1004 (2.6.32). I was wondering whether I
can use the filesystem (fedoracore5) on official website as the rot fs for
my uml. Thank you!

 

Best,

Yiming

Re: [uml-user] xterm got kill when i run application in uml, how come ?

[ning ji <ni...@ho...>]

Hi Richard,

so i run 20 of them again,
at the moment i can find the 1st kernel panic,
i dumped the memory stats, they look fine to me.
(The numbers look pretty much the same after all umls are loaded.)

Is there anything else i need to check ?
thanks !

root@VQEUbuntu32:/home/shadmin/10uml# free -m
             total       used       free     shared    buffers     cached
Mem:          3024       1071       1952          0          4        837
-/+ buffers/cache:        229       2794
Swap:         8855          0       8855
root@VQEUbuntu32:/home/shadmin/10uml# df  /dev/shm
Filesystem           1K-blocks      Used Available Use% Mounted on
none                   1548436    442072   1106364  29% /dev/shm



> Date: Tue, 3 Jan 2012 19:06:35 +0100
> Subject: Re: [uml-user] xterm got kill when i run application in uml, how come ?
> From: ric...@gm...
> To: ni...@ho...
> CC: lar...@er...; use...@li...
> 
> On Tue, Jan 3, 2012 at 6:01 PM, ning ji <ni...@ho...> wrote:
> > Starting system logger: [  OK  ]
> > Retrigger failed udev events[  OK  ]
> > [  156.810000] helper_wait : waitpid process 13447 failed, errno = 10
> > [  156.810000] * modprobe tun
> > [  156.810000] * ifconfig tap4 10.38.145.169 netmask 255.255.255.255 up
> > [  156.810000] * bash -c echo 1 > /proc/sys/net/ipv4/ip_forward
> 
> Maybe you're running out of host resources.
> It looks like a helper program is dying.
> 
> -- 
> Thanks,
> //richard
 		 	   		  

Re: [uml-user] xterm got kill when i run application in uml, how come ?

[richard -r. w. <ric...@gm...>]

On Tue, Jan 3, 2012 at 6:01 PM, ning ji <ni...@ho...> wrote:
> Starting system logger: [  OK  ]
> Retrigger failed udev events[  OK  ]
> [  156.810000] helper_wait : waitpid process 13447 failed, errno = 10
> [  156.810000] * modprobe tun
> [  156.810000] * ifconfig tap4 10.38.145.169 netmask 255.255.255.255 up
> [  156.810000] * bash -c echo 1 > /proc/sys/net/ipv4/ip_forward

Maybe you're running out of host resources.
It looks like a helper program is dying.

-- 
Thanks,
//richard

Re: [uml-user] xterm got kill when i run application in uml, how come ?

[ning ji <ni...@ho...>]

This is what i get when running 20 instances of uml in a vmware,
kernel is 2.6.39.4.
Thanks !


Starting system logger: [  OK  ]
Retrigger failed udev events[  OK  ]
[  156.810000] helper_wait : waitpid process 13447 failed, errno = 10
[  156.810000] * modprobe tun
[  156.810000] * ifconfig tap4 10.38.145.169 netmask 255.255.255.255 up
[  156.810000] * bash -c echo 1 > /proc/sys/net/ipv4/ip_forward
[  156.810000] Kernel panic - not syncing: Kernel mode signal 4
[  156.810000] 0861381c:  [<084d18bc>] dump_stack+0x1c/0x20
[  156.810000] 08613834:  [<084d190d>] panic+0x4d/0x13f
[  156.810000] 08613868:  [<08065b66>] relay_signal+0x30/0x72
[  156.810000] 08613888:  [<08071916>] sig_handler_common+0x61/0x70
[  156.810000] 08613900:  [<08071a3e>] sig_handler+0x31/0x3d
[  156.810000] 0861390c:  [<08071be2>] handle_signal+0x4d/0x7b
[  156.810000] 0861392c:  [<080732f7>] hard_handler+0xf/0x14
[  156.810000] 0861393c:  [<0068f400>] 0x68f400
[  156.810000] 08613c34:  [<080639e9>] do_IRQ+0x26/0x3d
[  156.810000] 08613c50:  [<08063b83>] sigio_handler+0x46/0x5f
[  156.810000] 08613c68:  [<08071916>] sig_handler_common+0x61/0x70
[  156.810000] 08613ce0:  [<08071a3e>] sig_handler+0x31/0x3d
[  156.810000] 08613cec:  [<08071be2>] handle_signal+0x4d/0x7b
[  156.810000] 08613d0c:  [<080732f7>] hard_handler+0xf/0x14
[  156.810000] 08613d1c:  [<0068f400>] 0x68f400
"bug3" 50 lines, 2697 characters



> Date: Sat, 24 Dec 2011 11:19:07 +0100
> Subject: Re: [uml-user] xterm got kill when i run application in uml, how come ?
> From: ric...@gm...
> To: ni...@ho...
> CC: lar...@er...; use...@li...
> 
> On Sat, Dec 24, 2011 at 4:42 AM, ning ji <ni...@ho...> wrote:
> > I left the error message in the company, but briefly the TAP devices have
> > issues.
> 
> Without the exact message nobody can help you.
> 
> > So far this project turns out to be a mess and i'm praying they won't fire
> > me.
> >
> 
> If UML is the root cause of your problems, there are lots of alternatives.
> Most UML users are using KVM today...
> 
> -- 
> Thanks,
> //richard
 		 	   		  

Re: [uml-user] xterm got kill when i run application in uml, how come ?

[richard -r. w. <ric...@gm...>]

On Sat, Dec 24, 2011 at 4:42 AM, ning ji <ni...@ho...> wrote:
> I left the error message in the company, but briefly the TAP devices have
> issues.

Without the exact message nobody can help you.

> So far this project turns out to be a mess and i'm praying they won't fire
> me.
>

If UML is the root cause of your problems, there are lots of alternatives.
Most UML users are using KVM today...

-- 
Thanks,
//richard

Re: [uml-user] xterm got kill when i run application in uml, how come ?

[ning ji <ni...@ho...>]

Thanks Lars.
Actually i did more testing, without my app, i just start 20 UMLs all together, some xterms got killed.So the problem is UML, not my app.
I left the error message in the company, but briefly the TAP devices have issues.In my case, each UML has a static ip. When all 20 starts, you can easily see the error with stack trace.
Now if i run all these in a vm in a vmWare server, it's even easier to reproduce, everything seems to be slower.
So far this project turns out to be a mess and i'm praying they won't fire me.


> From: lar...@er...
> To: ni...@ho...
> CC: use...@li...
> Date: Fri, 23 Dec 2011 09:51:15 +0100
> Subject: RE: [uml-user] xterm got kill when i run application in uml, how come  ?
> 
> 
> Hi,
> 
> Use "-hold" option to xterm. The terminal window will stay open
> even when the executed program has terminated. But you will have
> to close all windows manually, which you might find very annoying
> considering the number of vm's.
> 
> Regards,
> Lars Ekman
>  
> 
> -----Original Message-----
> From: ning ji [mailto:ni...@ho...] 
> Sent: den 23 december 2011 03:15
> Cc: use...@li...
> Subject: [uml-user] xterm got kill when i run application in uml, how come ?
> 
> it's a little bit complicated,
> 1. i have a script starts 20 xterms like this,
>    xterm -e "uml ..."
>    so basically i run 20 umls in my pc, each in a xterm so i can check 
> the progress.
> 
> 2. each of uml will automatically start my application,
>    which is multithreaded, has pipes, sockets etc.
> 
> when i do this, every once a while some xterms disappear.
> 
> if the UMLs don't run my app, i know it's fine.
> but if my app has bug, how can it kills the xterm ?
> 
> During this procedure, i keep on checking pc memory by "free -m",
> and seems it has plenty of free memory.
> Total is ~3GB, but i never see the used memory exceed 2GB.
> 
> Shared memory usage is only ~20%.
> 
> thanks !
> 
 		 	   		  

Re: [uml-user] xterm got kill when i run application in uml, how come ?

[Lars E. G <lar...@er...>]

Hi,

Use "-hold" option to xterm. The terminal window will stay open
even when the executed program has terminated. But you will have
to close all windows manually, which you might find very annoying
considering the number of vm's.

Regards,
Lars Ekman
 

-----Original Message-----
From: ning ji [mailto:ni...@ho...] 
Sent: den 23 december 2011 03:15
Cc: use...@li...
Subject: [uml-user] xterm got kill when i run application in uml, how come ?

it's a little bit complicated,
1. i have a script starts 20 xterms like this,
   xterm -e "uml ..."
   so basically i run 20 umls in my pc, each in a xterm so i can check 
the progress.

2. each of uml will automatically start my application,
   which is multithreaded, has pipes, sockets etc.

when i do this, every once a while some xterms disappear.

if the UMLs don't run my app, i know it's fine.
but if my app has bug, how can it kills the xterm ?

During this procedure, i keep on checking pc memory by "free -m",
and seems it has plenty of free memory.
Total is ~3GB, but i never see the used memory exceed 2GB.

Shared memory usage is only ~20%.

thanks !

[uml-user] xterm got kill when i run application in uml, how come ?

[ning ji <ni...@ho...>]

it's a little bit complicated,1. i have a script starts 20 xterms like this,   xterm -e "uml ..."   so basically i run 20 umls in my pc, each in a xterm so i can check the progress.2. each of uml will automatically start my application,   which is multithreaded, has pipes, sockets etc.when i do this, every once a while some xterms disappear.if the UMLs don't run my app, i know it's fine.but if my app has bug, how can it kills the xterm ?During this procedure, i keep on checking pc memory by "free -m",and seems it has plenty of free memory.Total is ~3GB, but i never see the used memory exceed 2GB.
Shared memory usage is only ~20%.thanks ! 		 	   		  

Re: [uml-user] stack smashing detected when running apps in the hostfs

[ning ji <ni...@ho...>]

This problem solved, it's a bug in our app. But i'll come with another question.

From: ni...@ho...
CC: use...@li...
Date: Tue, 20 Dec 2011 02:46:25 +0000
Subject: Re: [uml-user] stack smashing detected when running apps in the	hostfs








just to be clear, if we run the application in the stand alone linux pc, never seen such errors.
From: ni...@ho...
To: ni...@ho...
CC: use...@li...
Subject: stack smashing detected when running apps in the hostfs
Date: Tue, 20 Dec 2011 01:13:09 +0000







hi everyone,
i'm running 20 UMLs in linux,each of them can access a public folder in the host linux.(
http://user-mode-linux.sourceforge.net/hostfs.html )So i don't need to copy these apps/libs to each UML.so i start all these UMLs at the same time, they will automatically run the same appin the public fold (/mnt/host).But i run intostack smashing detected errorfor 1 of our lib.so.My questions is,is these a bug in the application/lib itself,or i'm running out of resources as i have 20 VMs ?Thanks !--


 		 	   		   		 	   		  

------------------------------------------------------------------------------
Write once. Port to many.
Get the SDK and tools to simplify cross-platform app development. Create 
new or port existing apps to sell to consumers worldwide. Explore the 
Intel AppUpSM program developer opportunity. appdeveloper.intel.com/join
http://p.sf.net/sfu/intel-appdev
_______________________________________________
User-mode-linux-user mailing list
Use...@li...
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user 		 	   		  

Re: [uml-user] stack smashing detected when running apps in the hostfs

[ning ji <ni...@ho...>]

just to be clear, if we run the application in the stand alone linux pc, never seen such errors.
From: ni...@ho...
To: ni...@ho...
CC: use...@li...
Subject: stack smashing detected when running apps in the hostfs
Date: Tue, 20 Dec 2011 01:13:09 +0000







hi everyone,
i'm running 20 UMLs in linux,each of them can access a public folder in the host linux.(
http://user-mode-linux.sourceforge.net/hostfs.html )So i don't need to copy these apps/libs to each UML.so i start all these UMLs at the same time, they will automatically run the same appin the public fold (/mnt/host).But i run intostack smashing detected errorfor 1 of our lib.so.My questions is,is these a bug in the application/lib itself,or i'm running out of resources as i have 20 VMs ?Thanks !--


 		 	   		   		 	   		  

[uml-user] stack smashing detected when running apps in the hostfs

[ning ji <ni...@ho...>]

hi everyone,
i'm running 20 UMLs in linux,each of them can access a public folder in the host linux.(
http://user-mode-linux.sourceforge.net/hostfs.html )So i don't need to copy these apps/libs to each UML.so i start all these UMLs at the same time, they will automatically run the same appin the public fold (/mnt/host).But i run intostack smashing detected errorfor 1 of our lib.so.My questions is,is these a bug in the application/lib itself,or i'm running out of resources as i have 20 VMs ?Thanks !--


 		 	   		  

[uml-user] Setting up NAT: uml can ping remote PC, but not the other way around

[ning ji <ni...@ho...>]

Hi all,
1. I run uml in linux1, ./umlinux ubda=my_rootfs mem=128M eth0=tuntap,,,192.117.0.254
2. after it boot up, i set the uml ip to be 192.117.0.7    route add default gw 192.117.0.254

3. in the linux1, i addiptables -t nat -A POSTROUTING -o eth0 -j MASQUERADE
4. linux 1 and linux 2 can ping each other5. uml and linux1 can ping each other6. uml can ping linux 27. but linux 2 cannot ping uml
pls advice what to do, thanks !

 		 	   		  

Re: [uml-user] impact on erlang?

[richard -r. w. <ric...@gm...>]

On Tue, Dec 13, 2011 at 8:21 PM, Matthias Rieber <sup...@su...> wrote:
> Hi,
>
> Am 13.12.2011 16:13, schrieb richard -rw- weinberger:
>
>> I'll try that. What should I grep for in the strace log?
>>
>> For anything which can cause your problem.
>> E.g. Failing system calls which should not fail...
>>
>> We've had problems with java and glibc where glibc used x86 specific
>> futex operations which are not supported on UML...
>>
> I traced a larger part and it seems that erlang causes many futex calls on
> the hostsystem and none on the uml kernel.
>

That's good. UML has only limited futex support...

-- 
Thanks,
//richard

Re: [uml-user] impact on erlang?

[Matthias R. <sup...@su...>]

Hi,

Am 13.12.2011 20:44, schrieb Riccardo Murri:
>> I traced a larger part and it seems that erlang causes many futex calls
>> on the hostsystem and none on the uml kernel.
>>
> I think you should run the strace *in* the UML, not on the host.

I run it on the uml kernel and host kernel to find the difference why 
it's not running properly on the uml kernel.

Matthias