users

users@lists.genode.org

2357 discussions

A short Nixcon talk
by Emery 16 Nov '15

16 Nov '15

I gave a talk on friday at a NixOS conference on saturday about the nix port, its very brief, and I may have mispoke in a nervous rush at times, but I got some good questions after the talk session was over. I even got a few questions about porting to the mill architecture from someone who already knew of the project. https://media.ccc.de/v/nixcon2015-8-en-Lightning_Ignite_talks_day_1#video&t… Starts at 12:40 The palor trick at the end was silly, but it was something I came up with a hour or two before to show how generally the concepts can be applied. I have two days of sprint starting tomorrow with my nix guys, and then I plan to start on some documentation. Emery

2 2

Turmvilla scenario
by Vincent Digital 13 Nov '15

13 Nov '15

Hi, With the previous Turmvilla scenario, I wish to setup two virtualbox sessions, one running LInux process and the other running Windows, concurrently. I created a new Win7 subsystem, emulating the LInux subsystem, with new vm_win7.vdi. However, I could startup only one of each sessions from NOUX, either Windows or Linux, but not both sessions at the same time. From CLI monitor, I could see that RAM is allocated separately for both sessions. I suspect some concurrency issue but could not find the issue. I've attached the log from captured from NOUX. Another issue I encountered is that the virtual box mouse pointer in the WIndows session and the Nitpicker mouse pointer can't seem to merge together unlike the Linux session where both pointers are merged. Any advise. Thanks in advance.

5 13

Troubles linking Wt library
by Reinier Millo Sánchez 11 Nov '15

11 Nov '15

Hi Genode Comunity I need porting a Qt4.8 application on GenodeOS. I'm using the old qt4 support on GenodeOS, and it's working fine with the examples. Now, my applications have another dependencies like WebToolkit (Wt) library. I'm working on the support of this library. I have followed the Porting Guide and have made some progress. Basically I have made a basic support of some Boost libraries needed to compile Wt: system, signals, random, date_time, regex and thread. These libraries compile fine on GenodeOS. I have tried to bring up the Wt library. To do this, i have reviewed the build log of the library. Using this build log I have prepares the build files for the library and make a dummy example (for test the compilation of the library) using one of the examples on repos/libports/src/test/libports/ Checking the dependencies of the compiled library on my pc I've got readelf -a libwt.so | grep NEEDED 0x0000000000000001 (NEEDED) Shared library: [libboost_thread.so.1.55.0] 0x0000000000000001 (NEEDED) Shared library: [libboost_random.so.1.55.0] 0x0000000000000001 (NEEDED) Shared library: [libboost_regex.so.1.55.0] 0x0000000000000001 (NEEDED) Shared library: [libboost_signals.so.1.55.0] 0x0000000000000001 (NEEDED) Shared library: [libboost_system.so.1.55.0] 0x0000000000000001 (NEEDED) Shared library: [libboost_filesystem.so.1.55.0] 0x0000000000000001 (NEEDED) Shared library: [libboost_date_time.so.1.55.0] 0x0000000000000001 (NEEDED) Shared library: [librt.so.1] 0x0000000000000001 (NEEDED) Shared library: [libpthread.so.0] 0x0000000000000001 (NEEDED) Shared library: [libstdc++.so.6] 0x0000000000000001 (NEEDED) Shared library: [libm.so.6] 0x0000000000000001 (NEEDED) Shared library: [libgcc_s.so.1] 0x0000000000000001 (NEEDED) Shared library: [libc.so.6] 0x0000000000000001 (NEEDED) Shared library: [ld-linux-x86-64.so.2] Based on it, i have set the dependencies to compile the Wt library on GenodeOS: LIBS += libc libc-stdlib libm cxx stdcxx pthread boost_system boost_date_time boost_random boost_signals boost_filesystem boost_thread boost_regex At this point the source code of the library compiles fine, but when to try to link the dummy application with the compiled library get this error: Program test/libports/wt/test-webtoolkit LINK test-webtoolkit /home/millo/XEOS/GenodeOS/build/foc_rpi/var/libcache/wt/wt.lib.so: undefined reference to `kevent' /home/millo/XEOS/GenodeOS/build/foc_rpi/var/libcache/wt/wt.lib.so: undefined reference to `sigwait' /home/millo/XEOS/GenodeOS/build/foc_rpi/var/libcache/wt/wt.lib.so: undefined reference to `kqueue' /home/millo/XEOS/GenodeOS/build/foc_rpi/var/libcache/wt/wt.lib.so: undefined reference to `sendmsg' collect2: error: ld returned 1 exit status Some body have any suggestion or any idea about this? I'm missign some library dependencie? Best regards -- Lic. Reinier Millo SÃ¡nchez Centro de Estudios de InformÃ¡tica Universidad Central "Marta Abreu" de Las Villas Carretera a CamajuanÃ Km 5 1/2 Santa Clara, Villa Clara, Cuba CP 54830 "antes de discutir ... respira; antes de hablar ... escucha; antes de escribir ... piensa; antes de herir ... siente; antes de rendirte ... intenta; antes de morir ... vive"

2 1

Best referance hardware at this point ?
by Peter Lindener 10 Nov '15

10 Nov '15

Dear Fellow Genodians- Is a Lenovo Thinkpad x201 is still the best hardware platform for syncing up with other's who are working with Turmvilla <https://github.com/genodelabs/genode/issues/1552> ? or is there something more modern that leading edge Genodians are working hard to make into there prefured day to day use Genode platform ? In the longer run... I am intrested in running the Gneode OS on Snickerdoodle <http://makezine.com/2015/10/12/snickerdoodle-dev-board-fpga-arm-processor/> (Dual Core ARM_A9 as well as the Mips Creator Ci40 <http://www.theregister.co.uk/2015/10/20/imagination_technologies_creator_ci…> .. It would be great if we might be able to deliver Genode to run on these boards via the Snappy package update manager <http://www.linuxjournal.com/content/more-stable-future-ubuntu>.... any thoughts on this? am I dreaming in the right direction? -Peter

7 10

TrustZone: transfering information cross-world using interrupts
by David Goltzsche 09 Nov '15

09 Nov '15

Hi, this is a follow-up question to an earlier thread [0] on this mailinglist regarding the injection of software interrupts from Genode into the normal world VM. The basic mechanism works fine, but we need to transfer some meta information with the interrupt to the normal world. One possible approach would be to write something to a certain location in the non-secure memory before triggering the interrupt and reading that location from Linux. Of course, a cross-world synchronization mechanism is required here to maintain the consistency of the reserved memory range. Another approach could use Linux's shared interrupts: Additional information can be transferred via the dev_id pointer passed to an interrupt handler. Are both approaches possible in general, which one would you recommend? David [0] http://sourceforge.net/p/genode/mailman/message/34244066/ -- David Goltzsche, M.Sc. Institute of Operating Systems and Computer Networks Distributed Systems Group TU Braunschweig www : https://www.ibr.cs.tu-bs.de/users/goltzsch mail: goltzsche@...290... tel : +49 531 391 3249 fax : +49 531 391 5936 xmpp: goltzsche@...369...

3 2

QEMU instead of VirtualBox
by Jookia 09 Nov '15

09 Nov '15

Hey there, Given VirtualBox requires nonfree dependencies to be built, would it be worth having a port of QEMU running on Genode instead? How hard would it be to set this up? Cheers, Jookia.

1 0

Nitpicker changes
by Norman Feske 05 Nov '15

05 Nov '15

Hello, with the incorporation of my recent work on issue #1690 [1] into the master branch, the use and configuration of the nitpicker GUI server has changed. [1] https://github.com/genodelabs/genode/issues/1690 Up to now, nitpicker used to handle the so-called X-ray mode internally. When activated by the user by pressing a configurable key, this mode tinted all views with domain colors and added labeling information to all views. It was meant to assist the user with uncovering Trojan Horses. However, the built-in policy started to stand in our way when implementing advanced scenarios like Turmvilla [2]. Here, we want to be more flexible with how the X-ray mode is activated, and how it affects the behavior of the user interface. For example, the delivery of hovering motion events to unfocused sessions, the ability to change the input focus by clicking on a unfocused view, the suppression of the labeling for trusted parts of the GUI. Building-in all those "features" into nitpicker would ultimately increase its complexity, and still fail to cover all thinkable use cases. To avoid that, I removed the notion of the X-ray mode altogether but added all the various aspects of the policy as explicit attributes of nitpicker's domain configuration. The idea is that the notion of the X-ray mode is provided by a separate component (xray_trigger), which reconfigures nitpicker each time when the X-ray mode is toggled. The X-ray trigger component contains the policy of when the X-ray mode is activated. E.g., by interpreting certain global keys. It simply produces an X-ray report that tells whether X-ray is active or not. The reconfiguration of nitpicker is done by a separate component that takes the information of the X-ray report and generates a nitpicker configuration. To avoid this component to be nitpicker-specific, I came up with a generic 'rom_filter', which is able to generate a ROM module with arbitrary XML, depending on other ROM modules, like the X-ray report. The output of the ROM filter is fed to nitpicker as configuration. Granted, to achieve the original effect as present in the demo.run script, the scenario becomes more complicated because we have to start the xray_trigger and rom_filter in addition to nitpicker. However, we gain a lot of flexibility that can now be expressed by the means of combining and configuring those components. To see the interplay of those components to emulate the original behavior, please have a look at the os/run/demo.run script. Most other run scripts use nitpicker solely as an output facility. Here, the X-ray mode is actually unimportant and it suffices to set up the nitpicker configuration in a way that allows for basic interactions. For example, the following configuration can be used as a template: <config> <domain name="pointer" layer="1" content="client" label="no" origin="pointer" /> <domain name="default" layer="2" content="client" label="no" hover="always" focus="click" /> <policy label="pointer" domain="pointer"/> <policy label="" domain="default"/> </config> It assigns the pointer to its dedicated domain, using the pointer position of the origin of its coordinate space. All regular clients end up in the "default" domain, their content is presented as is (not tinted) without any labeling, motion events are delivered to the clients whether they are focused or not, and a client can be focused by clicking onto one of its views. As for a more sophisticated example, I have incorporated the new concept into my Turmvilla branch [2] where not only nitpicker but also the window decorator dynamically responds to the X-ray mode. Here, X-ray is activated whenever the pointer hovers a window decoration, which illustrates quite well the flexibility of the new approach. [2] https://github.com/genodelabs/genode/issues/1552 Cheers Norman -- Dr.-Ing. Norman Feske Genode Labs http://www.genode-labs.com · http://genode.org Genode Labs GmbH · Amtsgericht Dresden · HRB 28424 · Sitz Dresden Geschäftsführer: Dr.-Ing. Norman Feske, Christian Helmuth

1 0

Cache lock-down on Pandaboard.
by Daniel Sobański 04 Nov '15

04 Nov '15

Hi there, Is there a way to lock L2 cache on Pandaboard ES with running Ubuntu there? TRM says that it is possible, but I don't know it feasible on Pandaboard. I've tried to compile kernel object and set there bits in Auxiliary Control Register using cp15, but it's RO I suppose, because I cannot write there. CPSR says I am in a Privilige mode, but i guess that's a non-secure Privilige mode? How to use that PL310 cache controller to do that? Do I need to use TrustZone somehow? When compile SMC #1 instruction using asm volatile(...) when compiling natively on Panda, when trying to taskset this Panda is not responding. Cheers,

1 0

stdio.h header in Genode
by kas 31 Oct '15

31 Oct '15

Hey, When I try to build an application which includes stdio.h of libc, many linking error happens. I think it is due to the collision between Genode's base environment and 'libc' because when I remove stdio.h header file, the linking error message disappears. Can you please tell me what is missing here? thanks in advance kind regards, kas

2 1

Demo on i.MX53 QSB
by Tiago Brito 29 Oct '15

29 Oct '15

Hi, I've sent an email to Stefan asking some questions regarding the demonstration using an i.MX53 SABRE tablet. I was suggested by Stefan to ask questions like these at the mailing list. Anyway, thank you Stefan for answering me. My idea is to run a similar demonstration on a i.MX53 QSB, and this is the most relevant part of Stefan's answer: > > I tried to reproduce the demonstration on my Freescale i.MX53 Quick Start > > Board, following the instructions of the readme in your github repository, > > and I failed to do so. > Anyway, coming to your problem: you've tried to execute a binary version > of the Genode OS compiled to run on the secure side of an i.MX53 SABRE > tablet on top of an i.MX53 Quickstart board. These are different boards > with differently configured i.MX53 SoCs (different RAM areas, different > GPIO pins, different display port assignments, ...). > > If you want to execute exactly the same demonstration on top of the > Quickstart board, you'll have to do quite some adaption work. Although, > we have already a framebuffer driver that works for the LVDS extension > of the Quickstart board, a touchscreen driver for this board is still > missing. Moreover, you have to compile Android for the Quickstart Board > using my adapted Linuk kernel version for i.MX53 [2]. Even so, I still have some questions. I would like to know if it is possible to reproduce a similar demonstration, with Genode and Android, using an i.MX53 Quick Start Board, but this time using, a mouse, keyboard and a regular PC monitor (through VGA), instead of using a touchscreen. How should I proceed? And do I need to develop the required drivers? Do I need to build a similar demo using existing drivers that already support my intentions? Is it possible to change the vmm.run script in order to introduce the ps2 drivers instead of the touchscreen's drivers? And what about the display configurations? Stefan mentioned the QSB has different RAM areas, GPIO pins, and display port assignments. Does that mean it is difficult to see the demo through VGA? > I do not want to take the wind out of your sails, but I think there is > quite some work to do to port the demo to the Quickstart board. Maybe > you should start with a less rich example. In Genode's current version, > we have a run-script (test scenario) that boots a simple > virtual-machine-monitor on top of Genode in the secure world, which > boots a Linux kernel with a small initramfs. I think it is a good > starting point to build and test this scenario. > so long best regards > Stefan > > [1] https://lists.sourceforge.net/lists/listinfo/genode-main > [2] https://github.com/skalk/linux/tree/imx53-tz With my best regards Tiago

2 1

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

users