Lenovo x230 xserver Crash before install

2344 views
Skip to first unread message

Chris

unread,
Aug 12, 2012, 10:26:30 AM8/12/12
to qubes...@googlegroups.com
Hi all,
After booting live DVD rc1 everything seams to bootstrap correctly! But when anaconda starts it only displays a message below the big blue screen that the xserver crashed and then to press some keys to reboot.... this happens in normal and save graphics mode! Is there a way to get additional error information (stack trace/ something else..). The x230 has new ivy bridge graphics could this be the problem?
Greetings stoffl
--
Sent from my Android phone. Please excuse my brevity.

Marek Marczykowski

unread,
Aug 13, 2012, 9:56:52 AM8/13/12
to qubes...@googlegroups.com, Chris
On 12.08.2012 16:26, Chris wrote:
> Hi all,
> After booting live DVD rc1 everything seams to bootstrap correctly! But when anaconda starts it only displays a message below the big blue screen that the xserver crashed and then to press some keys to reboot.... this happens in normal and save graphics mode! Is there a way to get additional error information (stack trace/ something else..). The x230 has new ivy bridge graphics could this be the problem?

You can try switch to other console (Alt+F2) and then look into logs. Most
useful should be Xorg log (/var/log/Xorg.log, can be also in /tmp/) and kernel
messages (dmesg command).

You can also try "nopat" kernel parameter (at boot loader stage press TAB and
add it right before "--- initrd.img".

--
Best Regards / Pozdrawiam,
Marek Marczykowski
Invisible Things Lab

signature.asc

stoffl4ever

unread,
Aug 13, 2012, 1:03:26 PM8/13/12
to Marek Marczykowski, qubes...@googlegroups.com
On 08/13/2012 03:56 PM, Marek Marczykowski wrote:
> On 12.08.2012 16:26, Chris wrote:
>> Hi all,
>> After booting live DVD rc1 everything seams to bootstrap correctly! But when anaconda starts it only displays a message below the big blue screen that the xserver crashed and then to press some keys to reboot.... this happens in normal and save graphics mode! Is there a way to get additional error information (stack trace/ something else..). The x230 has new ivy bridge graphics could this be the problem?
> You can try switch to other console (Alt+F2) and then look into logs. Most
> useful should be Xorg log (/var/log/Xorg.log, can be also in /tmp/) and kernel
> messages (dmesg command).
X.log:
[ 30.432] (EE) VESA(0): V_BIOS address 0xb9040 out of range
[ 30.432] (II) UnloadModule: "vesa"
[ 30.432] (II) UnloadModule: "int10"
[ 30.432] (II) Unloading /usr/lib64/xorg/modules/libint10.so
[ 30.432] (II) UnloadModule: "vbe"
[ 30.432] (II) Unloading /usr/lib64/xorg/modules/libvbe.so
[ 30.432] (EE) Screen(s) found, but none have a usable configuration.
[ 30.432] Fatal server error:
[ 30.432] no screens found
[ 30.432] Please consult the Fedora Project support at
http://bodhi.fedoraproject.org/ for help.
[ 30.432] Please also check the log file at "/tmp/X.log" for
additional information.

anaconda.log:
01:10:37,628 WARNING anaconda.stdout: X startup failed, aborting
installation

> You can also try "nopat" kernel parameter (at boot loader stage press TAB and
> add it right before "--- initrd.img".
unfortunately didn't work, no success, same results!

Above there are a few dumps from the logs. I am still searching for this
xserver error.
If I have a solution I will post it here! If you or somebody else has an
idea what i should test then
just drop me a line! Thank you!

Best regards,
Chris

Marek Marczykowski

unread,
Aug 13, 2012, 4:20:40 PM8/13/12
to stoffl4ever, qubes...@googlegroups.com
On 13.08.2012 19:03, stoffl4ever wrote:
> On 08/13/2012 03:56 PM, Marek Marczykowski wrote:
>> On 12.08.2012 16:26, Chris wrote:
>>> Hi all,
>>> After booting live DVD rc1 everything seams to bootstrap correctly! But when anaconda starts it only displays a message below the big blue screen that the xserver crashed and then to press some keys to reboot.... this happens in normal and save graphics mode! Is there a way to get additional error information (stack trace/ something else..). The x230 has new ivy bridge graphics could this be the problem?
>> You can try switch to other console (Alt+F2) and then look into logs. Most
>> useful should be Xorg log (/var/log/Xorg.log, can be also in /tmp/) and kernel
>> messages (dmesg command).
> X.log:
> [ 30.432] (EE) VESA(0): V_BIOS address 0xb9040 out of range

(...)

Ok, here is fail from VESA driver. But what about the intel one? It should be
tried somehow earlier...
Does it work in some other linux distribution? If yes, what version of kernel
and xorg driver (xorg-x11-drv-intel package, or something like this) do you have?
signature.asc

stoffl4ever

unread,
Aug 14, 2012, 6:53:52 AM8/14/12
to Marek Marczykowski, qubes...@googlegroups.com
Sry for stealing your time, if would have taken a closer look i would
have noticed that the loaded intel driver is version 7.x which has no
support for ivy bridge its even written in the log which chipsets it
supports...

[ 30.302] (II) Module fbdev: vendor="X.Org Foundation"
[ 30.302] compiled for 1.7.99.3, module version = 0.4.1
[ 30.302] ABI class: X.Org Video Driver, version 7.0
[ 30.302] (II) intel: Driver for Intel Integrated Graphics Chipsets:
i810,
i810-dc100, i810e, i815, i830M, 845G, 854, 852GM/855GM, 865G, 915G,
E7221 (i915), 915GM, 945G, 945GM, 945GME, Pineview GM, Pineview G,
965G, G35, 965Q, 946GZ, 965GM, 965GME/GLE, G33, Q35, Q33, GM45,
4 Series, G45/G43, Q45/Q43, G41, B43, B43, Clarkdale, Arrandale,
Sandybridge, Sandybridge, Sandybridge, Sandybridge, Sandybridge,
Sandybridge, Sandybridge
[ 30.302] (II) VESA: driver for VESA chipsets: vesa
[ 30.302] (II) FBDEV: driver for framebuffer: fbdev

It doesn't fail but when loading the framebuffer drivers it tries VESA
and then it crashes..
What i have tried so far were Ubuntu 12.04 and Fedora 17 both using
intel driver 8.x which has support
for ivy bridge but the best support you get with the fresh builds via
git... some thing like 8.1, but generally
it boils down to kernel update to at least 3.4.x and the drivers which
could be considered as testing/unstable from git repositories,
so this also on standard distributions still a little bit unstable.

I should have bought a x220 not a x230... would have been cheaper...damn;-)
If you have any further things i should try or if you want me to test
something on my hardware for you,
i am eager to help, just post or pm me!

with best regards,
chris

Marek Marczykowski

unread,
Aug 14, 2012, 7:35:03 AM8/14/12
to stoffl4ever, qubes...@googlegroups.com
You can prepare you own installation DVD with newer software, especially
kernel and xorg driver. Basic instruction is here:
http://wiki.qubes-os.org/trac/wiki/QubesBuilder
I have kernel-3.4.4 in my development repo (and currently using it - work
quite well, at least on my hardware). Regarding xorg intel driver, we
currently use one from Fedora 15, you can easily upgrade it to fc17 or even
rawhide (and hope it will work with our ancient xorg server...). But if you
want newer, you will need to find (or prepare yourself) src.rpm package with it.

We haven't pushed 1.0-rc1 sources to "mainstream" repository yet, so you
should switch it so my or Joanna's repo (in builder.conf):
GIT_SUBDIR=marmarek

"make qubes" is equivalent to chain of individual targets:
$ grep ^qubes: Makefile
qubes: get-sources xen core kernel gui addons docs template kde-dom0 installer
qubes-manager dom0-updates sign-all

To make use it, instead of "make qubes", you should:
1. Download all sources
make get-sources

2. Change the kernel
cd qubes-src/kernel
git fetch git://git.qubes-os.org/marmarek/kernel.git devel-3.4:devel-3.4
git co devel-3.4

3. Change the xorg driver
cd qubes-src/dom0-updates
rm xorg-x11-drv-intel*fc15*rpm
yumdownloader --releasever=17 --source xorg-x11-drv-intel


4. Build the qubes (everything except "get-sources"):
make xen core kernel gui addons docs template kde-dom0 installer qubes-manager
dom0-updates sign-all

5. Build installation iso:
make iso
signature.asc

Marek Marczykowski

unread,
Aug 22, 2012, 6:36:24 AM8/22/12
to qubes...@googlegroups.com
To summarize the thread:
Unfortunately ivy bridge based systems are not supported by Qubes 1.0-rc1 (our
GPU driver is to old). Now it is too late to test required (complicated)
changes properly, so final 1.0 release will also have no support for it.

The good news: there are separate branches in our source repository with
development version, which should work on such hardware - totally untested and
without any warranty. If you want a try, building instructions are here:
http://wiki.qubes-os.org/trac/wiki/QubesBuilder

Instruction on wiki covers above case as example and is more complete than in
my previous message (previous lack of some important steps - adding
dependencies for xorg driver).
signature.asc

Marek Marczykowski

unread,
Aug 22, 2012, 11:57:31 AM8/22/12
to marv, qubes...@googlegroups.com
[sending to qubes-devel]

On 22.08.2012 16:10, marv wrote:
>
> aha the whole partition is encrypted. i think i will try it with pmagic ;) if
> not, what do you think, is it necessary to do an new install with unused space
> partition or is there no difference?

I think this won't worth system reinstall.

> boot time is quite a little i think about
> 2 minutes, starting a vm needs about 20-30 seconds
>
> ram used for dom0 is about 4-5 GB ;) ????

Dom0 have all memory not assigned to any other VM. So this can mean that no
other VM needs much memory.

> maybe i found a bug, i have created a new vm named "test", it was under
> applications, then renamed to "gmail test" but it was not more under
> applications, only when the name has no space -> "gmailtest" the vm is visible
> under applications, but no problem...

Space isn't allowed in VM name... If some tool allow you to set such name, it
is bug there.

> something i don't understand is the line of vms wich build the network... if i
> edit the "firewallvm" i get a message that the "firewallvm" is not connected
> to a "Firewall VM"....???

Indeed this can be confusing. "firewallvm" is the name of VM. "Firewall VM" is
VM role - VM with firewall. Actually every "Proxy VM" (VM type) is "Firewall
VM" (VM role).

Some more details about networking here:
http://theinvisiblethings.blogspot.com/2011/09/playing-with-qubes-networking-for-fun.html

> the templatevm has the option for updates, in dom0
> are settings for each vm, there is set
>
> 1. UpdateVM: firewallvm (current) is this correct? why here is not
> fedora-17-x64 for updates?

This is VM used to download updates (or more general: any packages) for dom0.

> 2. ClockVM: netvm (current) means this, that clockvm is not behind the firewall?
>
> 3. Default netVM: firewallvm (current) -> clear, standard setting for new vm's
> right?

Right.

> 4. Default template: fedora-17-x64 -> this is clear
>
> can you explain, is this wrong, how does it work, like this:
>
> netvm owns the network device -> firewallvm is a proxvm & has netvm as ntvm ->
> myownvm is connected to netvm through the firewallvm
>
> outgoing---> myvm > firewall > netvm
>
> incoming--> myvm < firewall < netvm

This is right.

> updates ---> fedora-17-x64 < firewall < netvm

Updates for VMs - right. As VMs are based on template, it is enough to update
only template.

> is there a difference between updates for dom0 and the vms ???

Dom0 have no direct network access and is not based on template, so updates
must be downloaded via some other VM. By default it is firewallvm, just
because it is always running and doesn't have much other tasks to do.
Note that, the UpdateVM is needed also for checking if updates are available
(which is done automatically every 6h), so this VM should be always running.
Updates for dom0 (as any other packages) are verified before installing.

> or does the current setting mean that dom0 & templates are updating via
> firewallvm ?

Yes, but in different mean: for template - firewallvm is just another router;
for dom0 - it downloads packages (using qubes-specific script, which basically
call yumdownloader) and then send it to dom0 (using special Qubes RPC service).

> is it explained in the wiki? got some connection problems currently ;)

I think no, but should be... Perhaps I will paste this email there ;)

> if yes, i will find it, please don't explain, if not or there is an error
> please tell me ;)
>
> network mouse keyboard works, USB 3.0 is now enabled in bios - but not tested
> yet because i got no USB 3 device... seems ok ;)
>
> best regards ;)
signature.asc

Marek Marczykowski

unread,
Aug 22, 2012, 2:12:55 PM8/22/12
to marv, qubes...@googlegroups.com
On 22.08.2012 19:31, marv wrote:
>> Space isn't allowed in VM name... If some tool allow you to set such name,
>> it is bug there.
> i used the normal edit vm dialog, you can enter spaces there ;)

Will fix it:
http://wiki.qubes-os.org/trac/ticket/656

> why is clockvm set to netvm - not behind the firewall or firewallvm used for
> clockvm, any reason for that?

Clock VM is used to synchronize time with network time servers. Same as in
case of Update VM - it should be always running. And as it listens for
incoming connections it's better to not introduce additional network service
in firewallvm (it currently have no open ports). But as in case of UpdateVM,
you can change this to whatever you want.

>> Note that, the UpdateVM is needed also for checking if updates are
>> available (which is done automatically every 6h), so this VM should be
>> always running.
> sorry, i got no UpdateVM ;( you mean the fedora template has to run for
> searching updates?

Not sure what do you mean... Update VM is VM role ("VM for downloading updates
on behalf of dom0"), not any specific VM type - it can be any VM, by default
"firewallvm".

>> Yes, but in different mean: for template - firewallvm is just another
>> router; for dom0 - it downloads packages (using qubes-specific script, which
>> basically call yumdownloader) and then send it to dom0 (using special Qubes
>> RPC service).
>>> This is VM used to download updates (or more general: any packages) for dom0
> and every proxyvm supports dom0 updates or is there something to change ip etc?

Yes, every VM can be set as UpdateVM.

> iam still searching where to configure the network ips?

Network connection settings can be configured in netvm. There is running
NetworkManager, you should have its icon in tray area.
Connections between VMs (including its IPs) are configured automatically based
on netvm setting in each VM. If you want to obtain IP address of VM, it is
shown in VM settings in "Basic" tab. You can also use qvm-ls cmdline tool
(qvm-ls -n).
signature.asc

Marek Marczykowski

unread,
Aug 23, 2012, 4:59:21 AM8/23/12
to marv, qubes...@googlegroups.com
On 23.08.2012 06:09, marv wrote:
> thx for the last mails i understand ;)
>
> last 3 hours i spend with 2 problems, clipboard does not work between vms, i
> try ctrl+shift+c keys and ctrl+shift+v
>
> with mouse copy+paste it does not work, any idea?

But inter-VM copy does work in general (you should see some notification about
copy and paste events), only the problem with mouse copy, right?
There are (at least) two types of clipboard in Linux: one is used by mouse
copy, the other by "normal" copy - Ctrl-C (or Ctrl-Ins in terminal window).
Qubes uses the second one for inter-VM clipboard. So in short - to copy some
some text from one VM to another, you must press Ctrl-C, Ctrl-Shift-C; to
paste it: Ctrl-Shift-V, Ctrl-V.
Very soon you will do it as fast as plain Ctrl-C/Ctrl-V ;)

> so i wrote down the whole script from joanna 2 x for the tor proxy setup, at
> the end i cant save the file in vim because
>
> of missing write permissions *aarrrghh* ;-)
>
> the last user comment on the site tells the same problem :-(

You can save the script somewhere (ex in /tmp) as normal user and then copy it
to desired directory as root. Or do whole operation as root, no matter.

> there is done in user mode, has i do it as root or the user?
>
> i cant copy the url or the error message from the user because no clipboard ;)
>
> i take some sleep ;)
>
> regards :-)
signature.asc

Marek Marczykowski

unread,
Aug 23, 2012, 11:03:40 AM8/23/12
to marv, qubes...@googlegroups.com
On 23.08.2012 16:21, marv wrote:
> good morning ;)
>
> in the same domain all works, copy with mouse or keyboard...

Only copying between different domains is restricted.

> i got a domain, green and a file in it, called green.txt and one domain called
> red, with red.txt
>
> if i copy red.txt (strg+c ---> shift+strg+c) and go to green file window,
> press shift+strg+v --> strg+v and nothing happens...
>
> if i go back to red folder and press strg+v it creates a new file called "red
> (copy) .txt) the clipboard also works but i cant insert it into
>
> green folder ;( and i can see or hear no sound or another notification while
> copying - nothing???
>
> the best, i can copy in green folder the green.txt to clipboard and at the
> same time i can copy the red.txt in red folder to clipboard... if i press
> strg+v in red it creates red (copy).txt - if i press strg+c in green window
> it creates green (copy).txt
>
> i have both in my clipboard... i think you are doing something with ha hash or
> a key, there must be a bug that divides the windows or something else ;(
>
> is there normally a sound or a message shown somewhere? maybe its a good idea
> to play i little significant beep something like "bip" & "bop" *lol*
>
> but for further using, how can i get files or txt from a -> b?

First of all, copying files between domains isn't the same as copying text.
File copy is described here:
http://wiki.qubes-os.org/trac/wiki/CopyingFiles
signature.asc

Marek Marczykowski

unread,
Aug 23, 2012, 11:57:05 AM8/23/12
to marv, qubes...@googlegroups.com
On 23.08.2012 17:42, marv wrote:
>
>> Only copying between different domains is restricted.
> yes - copy&paste on the same domain isn't restricted, that's quite clear...
>
> my problem is the same if i try to copy text from trusted browser into torvm
> terminal or generally from domA <-> domB.. it does not work...
>
>> First of all, copying files between domains isn't the same as copying text.
>> File copy is described here:
>> http://wiki.qubes-os.org/trac/wiki/CopyingFiles
> aha ok, thats the reason, with files this works perfect, now i can send a file
> from domA to incoming-xyz folder of domB ;)
>
> but the proxyvm i.e. the torvm has no file-manager and you can't add
> applications like browser or file-manager...

You can start it manually: right click on VM->Run command in VM->nautilus

Regarding torvm, you also may have a look at:
http://groups.google.com/group/qubes-devel/browse_thread/thread/a83bdf073c8f5566/654b1be0a12bba9d


> and i am unable to copy the text from "trusted" Firefox to proxyvm (torvm)
> terminal, or from terminal to another dom.

Copying *text* should work as you tried before (Ctrl-Shift-C etc) - described
here: http://wiki.qubes-os.org/trac/wiki/CopyPaste
You should see some notification like this:
http://qubes-os.org/files/screenshots/release-1-alpha-2/copy-paste-2.png

If not - let me know, it can be some keyboard layout issue or sth like this.

PS Please send questions to qubes-devel, I'm not the only one who can answer it...
signature.asc

Marek Marczykowski

unread,
Aug 24, 2012, 11:15:29 AM8/24/12
to marv, qubes...@googlegroups.com
On 24.08.2012 16:33, marv wrote:
>> Regarding torvm, you also may have a look at:
>> http://groups.google.com/group/qubes-devel/browse_thread/thread/a83bdf073c8f5566/654b1be0a12bba9d
>>
> ok thanks for your patience ;) i have done all now, with one difference, i
> installed vidalia to a clone (tor-template) of the normal template, i don't
> want to change your standard template for security reasons.
>
> netvm (standard template) > tor-firewall (standard template) > tor-proxy (uses
> the clown tor-template where vidalia is installed) > tor-appvm with firefox
> (standard template?)

You can also use standard "firewallvm" instead of "tor-firewall". "tor-proxy"
vm will allow only traffic through tor anyway.

> i can start vidalia and i have removed the autostart like described, if i
> press the start tor button it works and a message is shown that it is
> connected to the tor network. but from appvm i get no access to tornet or the
> internet, do i have to change the settings of the tor-appvm in a different way
> or to configure the firefox or is that handled by qubes?

I think this some issue with torvm, as said in linked post - it was done in
times of Fedora14 template...
Quick check of original instruction: Fedora 17 have changed method of
starting/stopping services - it is managed by systemd now. In the meantime we
introduced some service management layer for qubes services - you can
enable/disable them from Qubes Manager ("services" VM tab). To disable
firewall and netwatcher services, you should add there "qubes-firewall" and
"qubes-netwatcher" entry and ensure that they are *un*checked.

If it doesn't help, try to manually investigate if tor process is running, if
iptables entries are set same as in start_tor_proxy.sh script etc. You can
also try to trace packets using tcpdump.

> And which template must be used for the appvm, the normal template shut work
> because it needs no tor code in it or i am wrong?

This is ok - appvm doesn't need to know anything about tor.

>> Copying *text* should work as you tried before (Ctrl-Shift-C etc) - described
>> here: http://wiki.qubes-os.org/trac/wiki/CopyPaste
> ok i try it again and again, no notification, no text copying possible between
> doms, file sending works good ;)
>> If not - let me know, it can be some keyboard layout issue or sth like this.
> think you're on the right way, i changed the kb layout to de ;) i try to
> change back and reboot, maybe it helps... but german users have to copy text
> too :-) :-)

Of course... You can try 'xev' in dom0 to see what events are generated by
Ctrl-Shift-C. Especially interesting is "state" field of 'C' KeyPress event.
On my system it is 0x05 (0x04 + 0x01, which mean Ctrl and Shift).

You can also try to change standard shortcut for something else - edit
/etc/qubes/guid.conf (there are comments with some instructions) and restart
some VM to test the changes.
signature.asc

stoffl4ever

unread,
Aug 26, 2012, 6:10:32 PM8/26/12
to Marek Marczykowski, qubes...@googlegroups.com

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
as promised my preliminary test report with custom build form qubes-os
with new xorg stuff and fc17 as template:
working on x230:
- - wlan networking
- - lan networking
- - sound via speaker
- - usb 2.0
- - volume buttons/mute
- - hardware wireless switch
- - fn+ display brightness
- - fn+ media buttons play/pause/forward/backward

not working:
- - wwan not working: in network manager there is another entry where,
in fedora 17/ubuntu 12.04 the h5321gw from ericsson was, but it says:
"Wired Network (vif2.0)" device not managed --- sounds like no driver
support for that umts/gsm/gps modem, or am i wrong?
- - fn + wireless on/off
- - fn + lock
- - fn + sleep
- - fn mircophone/webcam
- - sd card reader, see below
- - disposable vm not starting/crash:
/var/log/qubes/qubes-restore-log:
time=1345920482.142038, starting
time=1345920483.650083, cows restored
Loading new save file /var/run/qubes/current_savefile (new xl fmt info
0x0/0x0/948)
savefile contains xl domain config
xc: error Failed to allocate memory for batch.!: Internal error
libxl: error: libxl_dom.c:310:libxl__domain_restore_common restoring
domain: Resource temporarily unavailable
cannot (re-)build domain: -3
libxl: error: libxl.c:711:libxl_domain_destroy non-existant domain 3
Error starting VM

i was not able to mount a ntfs formated usbstick, possible solution
adding ntfs support in dom0?
the destination selection window wasn't even shown.

but this brings me to the next error the new formated drive was kindly
named "New Volume" guess what happened?
the " " aka blank space was not escaped when passing the mount command
to nautilus in the destination vm or something similar!
so the result were two errors saying that "New" and "Volumes" could not
be mounted! bug fix could be using ' aka single quote or ", double quto
around the name to escape special chars or a script that sanitizes the
string and escapes ?!$" etc... by putting an \ in front of the special
char! Sry, but i haven't started the dev docs so i don't know where to
look for the bug despite writting a patch;-) but hopefully i will be
able sometime soon!:-D

Uppsi, consider the above incomplete because i am not able to mount an
ext4 formated sdcard in an vm. fails with check spelling error
"/media/TEST" not found!
Same with usbstick as ext4... may be an additional problem!
but qvm-block -a untrusted dom0:sdb works so the stick shows up in nautilus!
with cmdline the mmcblk/sdcard is said to be unsupported... if right
then the sdcard couldn't work!

another error, but i guess more a general kde error, here is the rather
strange description and solution:
different browser windows and terminal from different vms were open when
suddenly all the entries for the open windows stack above each other on
the far right directly by sys tray bar and i was not able to pull them
back where they should be on the left side after the desktop switcher
but no success. after reboot they stop stacking above each other but now
the window bar gets populated from the right side each time i open a new
window. strange but i could life with it...;-)
suddenly during writing, i got an idea and i found out why they seam to
stack from the right... there were 3 spacers in the bar 2 with fixed
size one with dynamic size but as i said i didn't put them there...strange!

Not tested yet - but hopefully will in the near future:
- - microphone
- - audio input/output combo jack
- - usb 3.0
- - express card
- - vga
- - minidisplayport
- - fingerprint reader (but it seams my version the 14fe:2020 [upek] is
not supported under linux yet...)
- - webcam
- - docking port/connector

cya gn8 chris


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJQOp7XAAoJEA7+puZkoY1YkE4H/RNBfPt9vjx/59mLxv6Lz/Lw
xnFrUL6db+2PtF7u6pSjbZWWzd6+B/VKAckzpwYfuSUbdr5fSnadwrAL0qD3uPW2
mh6V4OW4B/+ixKEh6/Pe+K8gpAjaWYnLgHQcGYql4MJZfRTnROKLcXVq3OccBZPq
i2B1FxCIq/Ou3ZLFtv4pn1cvqXCeSatDIAtTRCwAJednLkuVVnQB6Lre5JaIi8hg
g+VhaiY2EuxEhxRaFYUW4aHW/VFrd/dxJuVmRsMJ+K44xNNZbtr36tGqPCzGM/0x
J5NRFaiwHHQMMmvMSCqQXzfmWnaSL5+Fcy7KamUx8B4CnBGsLuH2YkaAg9J6NoU=
=Mmi2
-----END PGP SIGNATURE-----

Marek Marczykowski

unread,
Aug 27, 2012, 9:29:45 AM8/27/12
to stoffl4ever, qubes...@googlegroups.com
On 27.08.2012 00:10, stoffl4ever wrote:
> as promised my preliminary test report with custom build form qubes-os
> with new xorg stuff and fc17 as template:
> working on x230:
> - wlan networking
> - lan networking
> - sound via speaker
> - usb 2.0
> - volume buttons/mute
> - hardware wireless switch
> - fn+ display brightness
> - fn+ media buttons play/pause/forward/backward
>
> not working:
> - wwan not working: in network manager there is another entry where,
> in fedora 17/ubuntu 12.04 the h5321gw from ericsson was, but it says:
> "Wired Network (vif2.0)" device not managed --- sounds like no driver
> support for that umts/gsm/gps modem, or am i wrong?

vif2.0 is irrelevant here (this is backend of firewallvm). I suspect your wwan
modem is internal USB device - you can check using lsusb in dom0. It can be
also PCI device - check lspci. When you found it, you can attach it to netvm
following this guide:
https://wiki.qubes-os.org/trac/wiki/AssigningDevices

> - fn + wireless on/off
> - fn + lock
> - fn + sleep
> - fn mircophone/webcam

Perhaps some additional kernel module or userspace program is needed in
dom0... You can find some useful information about it on thinkwiki.org.
Especially:
http://www.thinkwiki.org/wiki/How_to_get_special_keys_to_work
http://www.thinkwiki.org/wiki/How_to_configure_acpid

> - sd card reader, see below
> - disposable vm not starting/crash:
> /var/log/qubes/qubes-restore-log:
> time=1345920482.142038, starting
> time=1345920483.650083, cows restored
> Loading new save file /var/run/qubes/current_savefile (new xl fmt info
> 0x0/0x0/948)
> savefile contains xl domain config
> xc: error Failed to allocate memory for batch.!: Internal error
> libxl: error: libxl_dom.c:310:libxl__domain_restore_common restoring
> domain: Resource temporarily unavailable
> cannot (re-)build domain: -3
> libxl: error: libxl.c:711:libxl_domain_destroy non-existant domain 3
> Error starting VM

Is it happens all the time, or just once? You can try to recreate DispVM
template using qvm-create-default-dvm command.

> i was not able to mount a ntfs formated usbstick, possible solution
> adding ntfs support in dom0?

Do you really need this disk in dom0? You should be able to attach it to some
VM, where NTFS support (ntfs-3g package) is installed.

> the destination selection window wasn't even shown.

I don't understand this...
You can attack block device to VM using:
a) cmdline: https://wiki.qubes-os.org/trac/wiki/StickMounting
b) qubes manager: right click in destination VM -> "attach/detach block
devices" submenu.

> but this brings me to the next error the new formated drive was kindly
> named "New Volume" guess what happened?
> the " " aka blank space was not escaped when passing the mount command
> to nautilus in the destination vm or something similar!
> so the result were two errors saying that "New" and "Volumes" could not
> be mounted!

Ough, strange. But how you get it? I've just checked and Nautilus show correct
name "New volume" and even mounted it correctly (NTFS).

> bug fix could be using ' aka single quote or ", double quto
> around the name to escape special chars or a script that sanitizes the
> string and escapes ?!$" etc... by putting an \ in front of the special
> char! Sry, but i haven't started the dev docs so i don't know where to
> look for the bug despite writting a patch;-) but hopefully i will be
> able sometime soon!:-D
>
> Uppsi, consider the above incomplete because i am not able to mount an
> ext4 formated sdcard in an vm. fails with check spelling error
> "/media/TEST" not found!
> Same with usbstick as ext4... may be an additional problem!
> but qvm-block -a untrusted dom0:sdb works so the stick shows up in nautilus!
> with cmdline the mmcblk/sdcard is said to be unsupported... if right
> then the sdcard couldn't work!

Regarding mmcblk devices I've just fixed it, so should work in next release.

> another error, but i guess more a general kde error, here is the rather
> strange description and solution:
> different browser windows and terminal from different vms were open when
> suddenly all the entries for the open windows stack above each other on
> the far right directly by sys tray bar and i was not able to pull them
> back where they should be on the left side after the desktop switcher
> but no success. after reboot they stop stacking above each other but now
> the window bar gets populated from the right side each time i open a new
> window. strange but i could life with it...;-)
> suddenly during writing, i got an idea and i found out why they seam to
> stack from the right... there were 3 spacers in the bar 2 with fixed
> size one with dynamic size but as i said i didn't put them there...strange!

The only one issue with KDE taskbar I'm aware of is incorrect window grouping.
When KDE decide to group windows, it group it based on color, not VM name - so
all VM with the same color are in the one, single group.
http://wiki.qubes-os.org/trac/ticket/375

> Not tested yet - but hopefully will in the near future:
> - microphone
> - audio input/output combo jack

Currently Qubes does not support passing audio input to VM, so microphone will
be available only in dom0, which is almost useless...
http://wiki.qubes-os.org/trac/ticket/234

> - usb 3.0
> - express card
> - vga
> - minidisplayport
> - fingerprint reader (but it seams my version the 14fe:2020 [upek] is
> not supported under linux yet...)
> - webcam

If you want to use webcam, you should also attach it somehow to VM. Currently
the only way to attach non-block USB devices it attach full USB controller as
PCI device.

> - docking port/connector

Thanks, I will add your hardware to HCL, but as "problematic" (because of
non-standard build requirement)...
signature.asc

stoffl4ever

unread,
Aug 27, 2012, 5:25:38 PM8/27/12
to Marek Marczykowski, qubes...@googlegroups.com

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

okay done;-) thx!
>
>> - fn + wireless on/off
>> - fn + lock
>> - fn + sleep
>> - fn mircophone/webcam
>
> Perhaps some additional kernel module or userspace program is needed in
> dom0... You can find some useful information about it on thinkwiki.org.
> Especially:
> http://www.thinkwiki.org/wiki/How_to_get_special_keys_to_work
> http://www.thinkwiki.org/wiki/How_to_configure_acpid
i know but installng software in dom0 could be challenging i guess or
can i just attach firewall vm to dom0? and yum install..
>
>> - sd card reader, see below
>> - disposable vm not starting/crash:
>> /var/log/qubes/qubes-restore-log:
>> time=1345920482.142038, starting
>> time=1345920483.650083, cows restored
>> Loading new save file /var/run/qubes/current_savefile (new xl fmt info
>> 0x0/0x0/948)
>> savefile contains xl domain config
>> xc: error Failed to allocate memory for batch.!: Internal error
>> libxl: error: libxl_dom.c:310:libxl__domain_restore_common restoring
>> domain: Resource temporarily unavailable
>> cannot (re-)build domain: -3
>> libxl: error: libxl.c:711:libxl_domain_destroy non-existant domain 3
>> Error starting VM
>
> Is it happens all the time, or just once? You can try to recreate DispVM
> template using qvm-create-default-dvm command.
happens always!
fix doesn't help same error only now domain 6 is missing and not 3 but
the rest is the same!
>
>
>> i was not able to mount a ntfs formated usbstick, possible solution
>> adding ntfs support in dom0?
>
> Do you really need this disk in dom0? You should be able to attach it
to some
> VM, where NTFS support (ntfs-3g package) is installed.
yeah i got what you mean but i was attaching stick via kde menue >
computer > left click on the device and a window pops up and asked where
to attached and this every time failed as described!
>
>
>> the destination selection window wasn't even shown.
>
> I don't understand this...
> You can attack block device to VM using:
> a) cmdline: https://wiki.qubes-os.org/trac/wiki/StickMounting
> b) qubes manager: right click in destination VM -> "attach/detach block
> devices" submenu.
or c) by clicking on the entry shown under kde>computer or it seams that
this is not the supposed way of doing it;-)
>
>
>> but this brings me to the next error the new formated drive was kindly
>> named "New Volume" guess what happened?
>> the " " aka blank space was not escaped when passing the mount command
>> to nautilus in the destination vm or something similar!
>> so the result were two errors saying that "New" and "Volumes" could not
>> be mounted!
>
> Ough, strange. But how you get it? I've just checked and Nautilus show
correct
> name "New volume" and even mounted it correctly (NTFS).
as above i tried to use kde>computer>click on device>window open select
dest vm and then error.
k
>
>> Not tested yet - but hopefully will in the near future:
>> - microphone
>> - audio input/output combo jack
>
> Currently Qubes does not support passing audio input to VM, so
microphone will
> be available only in dom0, which is almost useless...
> http://wiki.qubes-os.org/trac/ticket/234
>
>> - usb 3.0
>> - express card
>> - vga
>> - minidisplayport
>> - fingerprint reader (but it seams my version the 14fe:2020 [upek] is
>> not supported under linux yet...)
>> - webcam
>
> If you want to use webcam, you should also attach it somehow to VM.
Currently
> the only way to attach non-block USB devices it attach full USB
controller as
> PCI device.
>
>> - docking port/connector
>
> Thanks, I will add your hardware to HCL, but as "problematic" (because of
> non-standard build requirement)...
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJQO+XJAAoJEA7+puZkoY1Y4gQIAL45IJ2BrxSKVpHyS7RLDAxQ
q1F0nDWjmONrZfBVhWVSPoSV6JlbjyEMmFnFHybVneSG9XU+lAy35aWWzMURexaE
uz2mLmv00XKdqDqlNPxfydeyTUUraFz8JTkptPUCBpXCSHYsM9t6i7t2SkmqnluF
lT0PjH4wfcvPQDGlyhPNFsWgFKYrxImuaCz4q5ntwHpCCbqEL6kWlme4GpjhIgS3
k0hWx9v/bb6YZ7LN8pP4WAxWIUtQJfGnYX+Sd7xs9MDkQl8xAda/trITD7GrTK5j
dhBJ4UvRGpQ0nnzqGefmDSTxaFmg9rZ4cvWx1T3QCdLX/YmNcgmtvI2ooFg8NW8=
=T9tn
-----END PGP SIGNATURE-----

Marek Marczykowski

unread,
Aug 27, 2012, 5:58:05 PM8/27/12
to stoffl4ever, qubes...@googlegroups.com
On 27.08.2012 23:25, stoffl4ever wrote:
>
> On 08/27/2012 03:29 PM, Marek Marczykowski wrote:
>> On 27.08.2012 00:10, stoffl4ever wrote:
>>> - fn + wireless on/off
>>> - fn + lock
>>> - fn + sleep
>>> - fn mircophone/webcam
>
>> Perhaps some additional kernel module or userspace program is needed in
>> dom0... You can find some useful information about it on thinkwiki.org.
>> Especially:
>> http://www.thinkwiki.org/wiki/How_to_get_special_keys_to_work
>> http://www.thinkwiki.org/wiki/How_to_configure_acpid
> i know but installng software in dom0 could be challenging i guess or
> can i just attach firewall vm to dom0? and yum install..

You can do it without attaching firewallvm to dom0: "qubes-dom0-upgrade" tool
works almost the same as "yum install". You can for example call "sudo
qubes-dom0-upgrade tpb" and it will install tpb package.

>>> - disposable vm not starting/crash:
>>> /var/log/qubes/qubes-restore-log:
>>> time=1345920482.142038, starting
>>> time=1345920483.650083, cows restored
>>> Loading new save file /var/run/qubes/current_savefile (new xl fmt info
>>> 0x0/0x0/948)
>>> savefile contains xl domain config
>>> xc: error Failed to allocate memory for batch.!: Internal error
>>> libxl: error: libxl_dom.c:310:libxl__domain_restore_common restoring
>>> domain: Resource temporarily unavailable
>>> cannot (re-)build domain: -3
>>> libxl: error: libxl.c:711:libxl_domain_destroy non-existant domain 3
>>> Error starting VM
>
>> Is it happens all the time, or just once? You can try to recreate DispVM
>> template using qvm-create-default-dvm command.
> happens always!
> fix doesn't help same error only now domain 6 is missing and not 3 but
> the rest is the same!

Can you send xen log? You can obtain it from
/var/log/xen/console/hypervisor.log or by right-click on dom0 in qubes manager.

>>> i was not able to mount a ntfs formated usbstick, possible solution
>>> adding ntfs support in dom0?
>
>> Do you really need this disk in dom0? You should be able to attach it
> to some
>> VM, where NTFS support (ntfs-3g package) is installed.
> yeah i got what you mean but i was attaching stick via kde menue >
> computer > left click on the device and a window pops up and asked where
> to attached and this every time failed as described!

No, right click on device in KDE menu will not attach the device to VM. It
will mount it in dom0. I though I removed devices from KDE menu...

>>> the destination selection window wasn't even shown.
>
>> I don't understand this...
>> You can attack block device to VM using:
>> a) cmdline: https://wiki.qubes-os.org/trac/wiki/StickMounting
>> b) qubes manager: right click in destination VM -> "attach/detach block
>> devices" submenu.
> or c) by clicking on the entry shown under kde>computer or it seams that
> this is not the supposed way of doing it;-)

As said above, it isn't the same.
signature.asc

stoffl4ever

unread,
Aug 28, 2012, 3:10:53 AM8/28/12
to Marek Marczykowski, qubes...@googlegroups.com

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 08/27/2012 11:58 PM, Marek Marczykowski wrote:
> On 27.08.2012 23:25, stoffl4ever wrote:
>>
>> On 08/27/2012 03:29 PM, Marek Marczykowski wrote:
>>> On 27.08.2012 00:10, stoffl4ever wrote:
>>>> - fn + wireless on/off
>>>> - fn + lock
>>>> - fn + sleep
>>>> - fn mircophone/webcam
>>
>>> Perhaps some additional kernel module or userspace program is needed in
>>> dom0... You can find some useful information about it on thinkwiki.org.
>>> Especially:
>>> http://www.thinkwiki.org/wiki/How_to_get_special_keys_to_work
>>> http://www.thinkwiki.org/wiki/How_to_configure_acpid
>> i know but installng software in dom0 could be challenging i guess or
>> can i just attach firewall vm to dom0? and yum install..
>
> You can do it without attaching firewallvm to dom0:
"qubes-dom0-upgrade" tool
> works almost the same as "yum install". You can for example call "sudo
> qubes-dom0-upgrade tpb" and it will install tpb package.
ahh, nice worked, now I understand, multi purpose, smart! ;-)
>
>>>> - disposable vm not starting/crash:
>>>> /var/log/qubes/qubes-restore-log:
>>>> time=1345920482.142038, starting
>>>> time=1345920483.650083, cows restored
>>>> Loading new save file /var/run/qubes/current_savefile (new xl fmt info
>>>> 0x0/0x0/948)
>>>> savefile contains xl domain config
>>>> xc: error Failed to allocate memory for batch.!: Internal error
>>>> libxl: error: libxl_dom.c:310:libxl__domain_restore_common restoring
>>>> domain: Resource temporarily unavailable
>>>> cannot (re-)build domain: -3
>>>> libxl: error: libxl.c:711:libxl_domain_destroy non-existant domain 3
>>>> Error starting VM
>>
>>> Is it happens all the time, or just once? You can try to recreate DispVM
>>> template using qvm-create-default-dvm command.
>> happens always!
>> fix doesn't help same error only now domain 6 is missing and not 3 but
>> the rest is the same!
>
> Can you send xen log? You can obtain it from
> /var/log/xen/console/hypervisor.log or by right-click on dom0 in qubes
manager.
please find it attached!
>
>>>> i was not able to mount a ntfs formated usbstick, possible solution
>>>> adding ntfs support in dom0?
>>
>>> Do you really need this disk in dom0? You should be able to attach it
>> to some
>>> VM, where NTFS support (ntfs-3g package) is installed.
>> yeah i got what you mean but i was attaching stick via kde menue >
>> computer > left click on the device and a window pops up and asked where
>> to attached and this every time failed as described!
>
> No, right click on device in KDE menu will not attach the device to VM. It
> will mount it in dom0. I though I removed devices from KDE menu...
>
>>>> the destination selection window wasn't even shown.
>>
>>> I don't understand this...
>>> You can attack block device to VM using:
>>> a) cmdline: https://wiki.qubes-os.org/trac/wiki/StickMounting
>>> b) qubes manager: right click in destination VM -> "attach/detach block
>>> devices" submenu.
>> or c) by clicking on the entry shown under kde>computer or it seams that
>> this is not the supposed way of doing it;-)
>
> As said above, it isn't the same.
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJQPG7zAAoJEA7+puZkoY1Yw0kIAKDkFI59XTDyitY+5U59PXaG
Lgjh8G2xK+9n5Llvkl9Cs3vxEYFFOb89ECIu2h0gR4JmDzAeJ2mhIav5l8awif1a
RkvayQsNagXn9TZTYPCOximncSk8Ur2kzMoIGt9oD4wdG6IchOzqz+W7e8FdQj3E
wlKFZkz+qye+loHmq2gl6XzeREqagKfUfkMnK8WFXI1q9v+zNpx7WUGvF/GZoBWQ
FtIqTeK2VqpiLPdmpQqGaSII8OLdBOx7Jf22n2DSALx++8sgAXuMXBiknbHJE5wJ
Pp08SFomLFj6980+xECZKmDBwT2omwr/ALLCAc0+1nhZNWbpwg6WJcfXqTQjMto=
=nKrJ
-----END PGP SIGNATURE-----

hypervisor.log

stoffl4ever

unread,
Sep 2, 2012, 2:01:51 PM9/2/12
to qubes...@googlegroups.com

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello everyone!

i have managed to get some of the thinkpad related stuff to work but the
disposable vm problem is still present!

and a new error appeared every time i start/restart a vm and i am
connected to an external monitor via a dock, i get the kdialog warning
that says: vm is trying to do some illegal gui related stuff your vm may
have been compromised...etc. from the logs i don't get a clue. this also
happens when clicking on the network vm network-manager in the systray!

without the dock and monitor and after reboot it works again! any
suggestions?

have a nice week!
best regards,
chris
Comment: Using GnuPG with Mozilla - http://www.enigmail.net/

iQEcBAEBAgAGBQJQQ58HAAoJEA7+puZkoY1YVr0H/iLDeToffLA7RuuS/NfSeTYc
HUtPQeyZkLJiNeD/Bw9TjNICW5BYajslvWQMjAUmOq5dqY9FCdlP8/31332zkrfS
PxCOMoVvbwuWdSanj+WyFAPlnEDT6dYpOTpaCciznfdN6rgiW8qaz1kTy+Tl9bay
5XrxXjrmoDNRc1+87asLTMVK08b5lg6CPw+vANv//C8QCIeBDbSV9b3zPkG9EpQT
8oxzJhPsCvspld5eedmI6E4l53n0f1DLyMrQcmSQdP0aLhiyOSomf7aw1jTmdCH8
KdCkQU3nJw2oLvdSvkwbyHt+PFE9y2q0ra2ZJ9d2QqPSshfBTQc5CUYMufQeCZc=
=GTa/
-----END PGP SIGNATURE-----

Marek Marczykowski

unread,
Sep 3, 2012, 7:13:25 AM9/3/12
to qubes...@googlegroups.com, stoffl4ever
On 02.09.2012 20:01, stoffl4ever wrote:
>
> Hello everyone!
>
> i have managed to get some of the thinkpad related stuff to work but the
> disposable vm problem is still present!

Interesting is "(XEN) grant_table.c:362:d5 Bad flags (0) or dom (5). (expected
dom 5, flags 3)" line in your xen log. Can you check if it is correlated with
dispvm? Ex "tail -f /var/log/xen/console/hypervisor.log" during
qvm-create-default-dvm call.

> and a new error appeared every time i start/restart a vm and i am
> connected to an external monitor via a dock, i get the kdialog warning
> that says: vm is trying to do some illegal gui related stuff your vm may
> have been compromised...etc. from the logs i don't get a clue. this also
> happens when clicking on the network vm network-manager in the systray!

Can you post relevant part of gui log? (/var/log/qubes/guid.<VM ID>.log,
should be accessible in qubes manager until VM is running)

> without the dock and monitor and after reboot it works again! any
> suggestions?

I suspect this can be caused by changed screen size when you connect external
monitor. Screen size is sent to VM at startup, so I always make sure to
startup system with external monitor connected (later monitor can be
disconnected, important is startup time).
signature.asc

Marek Marczykowski

unread,
Sep 3, 2012, 7:16:40 AM9/3/12
to qubes...@googlegroups.com, stoffl4ever
On 03.09.2012 13:13, Marek Marczykowski wrote:
> On 02.09.2012 20:01, stoffl4ever wrote:
>>
>> Hello everyone!
>>
>> i have managed to get some of the thinkpad related stuff to work but the
>> disposable vm problem is still present!
>
> Interesting is "(XEN) grant_table.c:362:d5 Bad flags (0) or dom (5). (expected
> dom 5, flags 3)" line in your xen log. Can you check if it is correlated with
> dispvm? Ex "tail -f /var/log/xen/console/hypervisor.log" during
> qvm-create-default-dvm call.

Also /var/log/xen/console/guest-fedora-17-x64-dvm.log can be useful.
signature.asc

stoffl4ever

unread,
Sep 4, 2012, 4:40:59 AM9/4/12
to Marek Marczykowski, qubes...@googlegroups.com

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 09/03/2012 01:16 PM, Marek Marczykowski wrote:
> On 03.09.2012 13:13, Marek Marczykowski wrote:
>> On 02.09.2012 20:01, stoffl4ever wrote:
>>>
>>> Hello everyone!
>>>
>>> i have managed to get some of the thinkpad related stuff to work but the
>>> disposable vm problem is still present!
>>
>> Interesting is "(XEN) grant_table.c:362:d5 Bad flags (0) or dom (5).
(expected
>> dom 5, flags 3)" line in your xen log. Can you check if it is
correlated with
>> dispvm? Ex "tail -f /var/log/xen/console/hypervisor.log" during
>> qvm-create-default-dvm call.
nothing happens in the hypervisor.log during the qvm-create-default-dvm
command, it succeeds with no errors!
>
> Also /var/log/xen/console/guest-fedora-17-x64-dvm.log can be useful.
>
a lot of out of memory problems it seams! log, attached!

regarding the resolution problem still persists even with external
monitors connected during boot of qubes and startup of vms, also logs
attached!

Thanks, best regards,
Chris

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://www.enigmail.net/

iQEcBAEBAgAGBQJQRb6TAAoJEA7+puZkoY1Yp3UH/irRrgxb7HC7Uf80Sk9MmEAa
L48PBAF4uTfGygPa4CWP8d9NV2n6is6C2AHrg/KRj0R8Stg7X1ISVrfLGTqG9hP9
9NN8tT9/b2B2BvixJesnvFyF2yzCR0NLCuvw8TaTUjbrrB1tROEHSKbUdHgfSv8G
+xiHLPTX3pLM2/IwF3w577uQq49yQXj7fT9f6vJR38GRu4wh/pOl61haLSO2Fq0R
ERiVenlEQcZ9NDL2RSp+YBRl/VQ+ywDsIHFRyrOmIkiWWrsaT0kCvtGb/F1Y6EWj
iXLD25VGJJpI6YIa7rFzCH8iNQ6Ooq8aZyxCfcfVhmhKR6avpAcnRrsisaxkmx0=
=BON9
-----END PGP SIGNATURE-----

guest-fedora-17-x64-dvm.log
guest-untrusted.log
guid.4.log

Marek Marczykowski

unread,
Sep 4, 2012, 1:13:44 PM9/4/12
to stoffl4ever, qubes...@googlegroups.com
On 04.09.2012 10:40, stoffl4ever wrote:
>
> On 09/03/2012 01:16 PM, Marek Marczykowski wrote:
>> On 03.09.2012 13:13, Marek Marczykowski wrote:
>>> On 02.09.2012 20:01, stoffl4ever wrote:
>>>>
>>>> Hello everyone!
>>>>
>>>> i have managed to get some of the thinkpad related stuff to work but the
>>>> disposable vm problem is still present!
>>>
>>> Interesting is "(XEN) grant_table.c:362:d5 Bad flags (0) or dom (5).
> (expected
>>> dom 5, flags 3)" line in your xen log. Can you check if it is
> correlated with
>>> dispvm? Ex "tail -f /var/log/xen/console/hypervisor.log" during
>>> qvm-create-default-dvm call.
> nothing happens in the hypervisor.log during the qvm-create-default-dvm
> command, it succeeds with no errors!
>
>> Also /var/log/xen/console/guest-fedora-17-x64-dvm.log can be useful.
>
> a lot of out of memory problems it seams! log, attached!

This looks like two separate issues:
Regarding the first one (fedora-17-x64-dvm and oom-killer):
It looks like used memory is somehow miscalculated. The easy workaround is to
remove ballooning down DispVM template (then DispVM will always need 400MB to
startup):
Edit /usr/lib/qubes/qubes_prepare_saved_domain.sh and comment out line "xl
mem-set (...)".

Regarding the second one (untrusted VM and xen_balloon problem):
---
[ 350.408481] System RAM resource 118000000 - 127ffffff cannot be added
[ 350.408492] xen_balloon: reserve_additional_memory: add_memory() failed: -17
---
Looks like some kernel problem... You can try to switch to official Qubes 1.0
kernel (3.2.7) in VM (still using experimental 3.4.4 in dom0 for your GPU).
Just call:
sudo qubes-dom0-update kernel-qubes-vm-3.2.7
and then set it as default kernel (either using qubes manager - global
settings or qubes-prefs).

Let me know if this helps.

> regarding the resolution problem still persists even with external
> monitors connected during boot of qubes and startup of vms, also logs
> attached!

---
Verify failed: (int) untrusted_shmcmd->width < 4096 && (int)
untrusted_shmcmd->height < 3072
---

Ah, indeed, there is hardcoded window size limit:
#define MAX_WINDOW_WIDTH 4096
#define MAX_WINDOW_HEIGHT 3072

How big screen size do you have?
signature.asc

Chris

unread,
Sep 4, 2012, 1:34:04 PM9/4/12
to Marek Marczykowski, qubes...@googlegroups.com
See below! The Rest will be check when at home;-)
1920x1200+1920x1200+1366x768= max height 1200 max width 5206
>--
>Best Regards / Pozdrawiam,
>Marek Marczykowski
>Invisible Things Lab

--
Sent from my Android phone with K-9 Mail. Please excuse my brevity.

Marek Marczykowski

unread,
Sep 4, 2012, 2:48:33 PM9/4/12
to qubes...@googlegroups.com, stoffl4ever
On 04.09.2012 19:13, Marek Marczykowski wrote:
> Regarding the second one (untrusted VM and xen_balloon problem):
> ---
> [ 350.408481] System RAM resource 118000000 - 127ffffff cannot be added
> [ 350.408492] xen_balloon: reserve_additional_memory: add_memory() failed: -17
> ---
> Looks like some kernel problem... You can try to switch to official Qubes 1.0
> kernel (3.2.7) in VM (still using experimental 3.4.4 in dom0 for your GPU).
> Just call:
> sudo qubes-dom0-update kernel-qubes-vm-3.2.7
> and then set it as default kernel (either using qubes manager - global
> settings or qubes-prefs).

This doesn't looks promising in context of "Qubes 1.0 disposable VM fails" thread.
signature.asc

Mantis

unread,
Sep 7, 2012, 1:07:04 PM9/7/12
to qubes...@googlegroups.com
Greetings,
I seem to be running into the same initial problem the OP had with anaconda. However, I don't have an ivy bridge processor. I have a AMD 1100T x6. My graphics card in is Radeon 6870. I've tried quite a few different kernel parameters with no luck. I'm kind of stuck actually. Is it common to have problems with the installer when using an ATI graphics card?

Marek Marczykowski

unread,
Sep 7, 2012, 1:51:37 PM9/7/12
to qubes...@googlegroups.com, Mantis
On 07.09.2012 19:07, Mantis wrote:
> Greetings,
> I seem to be running into the same initial problem the OP had with
> anaconda. However, I don't have an ivy bridge processor. I have a AMD 1100T
> x6. My graphics card in is Radeon 6870. I've tried quite a few different
> kernel parameters with no luck. I'm kind of stuck actually. Is it common to
> have problems with the installer when using an ATI graphics card?

Have you tried basic video mode? There was also some option to use text mode
install (which will be only partial solution, as you will have the same
problem on installed system). I don't remember the exact option name
("text"?), should be the same as for Fedora installer.

Be aware of that text mode install have much less configurable options,
especially sometimes can produce broken installed system... This is why we
disabled automatic fallback to text mode install.

Anyway on installed system there is much more options to try to solve the
problem. The first one was in thread "problem with xorg on ATI card".
signature.asc

Marek Marczykowski

unread,
Sep 7, 2012, 1:55:23 PM9/7/12
to qubes...@googlegroups.com, Mantis
Of course you can also try to build ISO with newer Xorg server and drivers,
instruction here: http://wiki.qubes-os.org/trac/wiki/QubesBuilder
This solve most problems with ivy bridge systems (and brings some new like
fullscreen not working...), but not sure how about yours.
signature.asc

syd bris

unread,
Sep 8, 2012, 7:11:31 AM9/8/12
to qubes...@googlegroups.com
Same problem here on a Toshiba laptop with ATI graphics. You have 3
install options: standard (produces a snowy display); vesa (produces
x-startup failure); and repair installation (progresses OK). Tried the
nopat option on the standard install and after a screen resolution
change the installation progressed flawlessly. However, when rebooting
for first time into qubes the display went snowy again. So the
installed system is there but how do I get into it now?

Marek Marczykowski

unread,
Sep 8, 2012, 7:54:26 AM9/8/12
to qubes...@googlegroups.com, syd bris
On 08.09.2012 13:11, syd bris wrote:
> Same problem here on a Toshiba laptop with ATI graphics. You have 3
> install options: standard (produces a snowy display); vesa (produces
> x-startup failure); and repair installation (progresses OK). Tried the
> nopat option on the standard install and after a screen resolution
> change the installation progressed flawlessly. However, when rebooting
> for first time into qubes the display went snowy again. So the
> installed system is there but how do I get into it now?

Have you added the nopat option in grub menu (on installed system)?
signature.asc

Marek Marczykowski

unread,
Sep 10, 2012, 4:27:28 PM9/10/12
to syd2...@gmail.com, qubes...@googlegroups.com
On 10.09.2012 21:57, syd2...@gmail.com wrote:
>
> "nomodeset nopat" allowed the boot-up screen to proceed, but after that I
> just got a black screen with a white cursor. Is there a way to get into the
> installed system via the repair option on the install DVD?

There is recovery mode on installer DVD. You can also pass "init=/bin/bash"
option to the kernel.
signature.asc

Marek Marczykowski

unread,
Sep 28, 2012, 8:59:20 PM9/28/12
to Chris, qubes...@googlegroups.com
On 04.09.2012 19:34, Chris wrote:
>> Ah, indeed, there is hardcoded window size limit:
>> #define MAX_WINDOW_WIDTH 4096
>> #define MAX_WINDOW_HEIGHT 3072
>>
>> How big screen size do you have?
>>
> 1920x1200+1920x1200+1366x768= max height 1200 max width 5206

This is now fixed by recent updates (qubes-gui-dom0 package version 1.4.20).
signature.asc

Franz

unread,
Dec 14, 2012, 12:52:17 PM12/14/12
to qubes...@googlegroups.com


On Wed, Aug 22, 2012 at 7:36 AM, Marek Marczykowski <marm...@invisiblethingslab.com> wrote:
To summarize the thread:
Unfortunately ivy bridge based systems are not supported by Qubes 1.0-rc1 (our
GPU driver is to old). Now it is too late to test required (complicated)
changes properly, so final 1.0 release will also have no support for it.

The good news: there are separate branches in our source repository with
development version, which should work on such hardware - totally untested and
without any warranty. If you want a try, building instructions are here:
http://wiki.qubes-os.org/trac/wiki/QubesBuilder

Instruction on wiki covers above case as example and is more complete than in
my previous message (previous lack of some important steps - adding
dependencies for xorg driver).

Hi
does qubes 2 beta1 support ivy bridge based system?
best
Franz

Marek Marczykowski

unread,
Dec 19, 2012, 10:10:08 AM12/19/12
to qubes...@googlegroups.com, Franz
Should work (it have new Xorg), but I don't have hardware to test it. You can
try - if installer will launch in graphics mode, everything should be fine.
signature.asc

Franz

unread,
Dec 19, 2012, 12:41:47 PM12/19/12
to Marek Marczykowski, qubes...@googlegroups.com
I'll try with an external  USB3 SSD and let you know
Best
Franz

Marek Marczykowski

unread,
Dec 19, 2012, 8:28:30 PM12/19/12
to Franz, qubes...@googlegroups.com
Actually there are problems with USB3 support (some module missing or sth),
but it should work if you connect disk via USB2 port.
signature.asc
Reply all
Reply to author
Forward
0 new messages