users January 2018

users@lists.genode.org

24 participants
27 discussions

Individual file access
by Nobody III 16 Mar '18

16 Mar '18

To start off, I'm planning on building a component that handles file open/save requests for other components, primarily graphical applications. My plan is to use a qt5 QFileDialog for the user to select one or more files, then modify the vfs server's configuration to attach the selected file(s) to the component's filesystem. To attach individual files to a component's filesystem, it seems like I should write a server-side link vfs plugin. The plugin would be like a symlink, but would transparently resolve the link within the vfs server, providing access to otherwise inaccessible files. Is this possible? If so, can you give me some tips on how to implement it?

2 7

Memory write tracing/logging of an application / Watchpoints in Genode/Fiasco.OC
by Stark, Josef 14 Mar '18

14 Mar '18

Hi all, I want to be able to 'trace' every memory write access of an arbitrary target application, in order to 'duplicate' those writes into another memory location. So that I would then have at all times one exact copy of the memory area (stack and heap) used by that application. My plan was to realize this with the help of a parent application, that acts as a session-proxy for the target, i.e. target requests a RAM-session from parent, which forwards the request to core and then provides the session to target; thus, the parent later knows which dataspaces/ memory areas are used by target. It could then flag that whole memory area with watchpoints. Hence, parent would be notified about all write accesses made by target and could consequently duplicate those writes in a secondary memory location. (I know that this would probably incur a slowdown factor of 10-1000 times, but that is allowed in this scenario.) Sadly, only now I have read that watchpoints are not implemented in Genode, and most probably neither are they in the underlying kernel, Fiasco.OC, which I have to use; breakpoints on the other hand seem to work just fine. My question is: Could there possibly be another way to achieve my goal, or do I have to bite the bullet and implement watchpoints myself? Thanks for bearing with me and for any hints at all. Best, Josef

4 16

Microkernel and alumni meetup
by Matthias Lange 11 Feb '18

11 Feb '18

Dear friends of the microkernel, it's about time for another microkernel and alumni meetup to catch up on all the developments of the last 6 months. Let's meet for a beer or two. To find a date that fits the most people please visit https://dudle.inf.tu-dresden.de/Microkernel_Meetup/ and add your preferred day. Regards, Matthias. -- Matthias Lange, matthias.lange@...272..., +49-351-41 888 614 Kernkonzept GmbH. Sitz: Dresden. Amtsgericht Dresden, HRB 31129. GeschÃ¤ftsfÃ¼hrer: Dr.-Ing. Michael Hohmuth

1 1

nitpicker focus
by Boris Mulder 09 Feb '18

09 Feb '18

Hi all, I'm writing a scenario with nitpicker where I want to support pop-ups e.g. for asking the user's password. I made a simple GUI app with a textbox that pops up and disappears when required. However, whenever this popup appears I want nitpicker to immediately change focus to this popup and put it in front of all other nitpicker clients and windows. Now the behaviour is either "click", "transient" or "none", meaning the focus is only given when the user clicks on the window, or never. This is a problem since if some other untrusted app is running in the background, the user could type in his password immediately when the pop-up appears, and the input is then directed to the app instead of the popup itself, allowing the untrusted app to read the password. The only way the user could enter the password correctly at the moment is by clicking on the pop-up before typing. One solution would be handling the focus via a focus report-rom, however when the pop-up is not there I want nitpicker to use its default behaviour (together with the window manager) so that it focuses windows when they are clicked. Reimplementing this default behaviour so it produces the report in this case seems quite ugly. Do you have any ideas what I could try to do to achieve the desired behaviour? From nitpicker's readme: > The focus ROM is expected to have a 'label' attribute with the value set to the label of the to-be focused client. When using an external focus policy, domains configured as "click" are internally handled like "transient" focusable domains. It is up to the external focus policy component to make a new focus permanent by providing an updated "focus" ROM." What if I want to switch between an external and internal focus policy based on which apps are running? Do you think it is easiest to extend nitpicker to have a focus="always" attribute that always focuses a certain client whenever it is up? Or to write a separate complicated focus policy component responsible for all focusing? -- Met vriendelijke groet / kind regards, Boris Mulder Cyber Security Labs B.V. | Gooimeer 6-31 | 1411 DD Naarden | The Netherlands +31 35 631 3253 (office)

2 6

Sockets
by Roman Iten 30 Jan '18

30 Jan '18

Hi I have a libc server application that binds a UDP socket and a libc client application to test the server. For this purpose I start two vfs and bridge them using the nic_bridge component: <start name="vfs_client"> <binary name="vfs"/> <provides> <service name="File_system"/> </provides> <config> <vfs> <dir name="socket"> <lxip ip_addr="127.0.0.2" netmask="255.0.0.0" gateway="0.0.0.0" nameserver="0.0.0.0"/> </dir> </vfs> <default-policy root="/socket" writeable="yes"/> </config> <route>  </route> </start> <start name="vfs_server"> <binary name="vfs"/> <provides> <service name="File_system"/> </provides> <config> <vfs> <dir name="socket"> <lxip ip_addr="127.0.0.1" netmask="255.0.0.0" gateway="0.0.0.0" nameserver="0.0.0.0"/> </dir> </vfs> <default-policy root="/socket" writeable="yes"/> </config> <route>  </route> </start> <start name="nic_bridge"> <provides> <service name="Nic"/> </provides> <config> <policy label_prefix="vfs_client" ip_addr="127.0.0.2" /> <policy label_prefix="vfs_server" ip_addr="127.0.0.1" /> </config> <route> <service name="Nic"> <child name="nic_drv"/> </service>  </route> </start> The server successfully binds a UDP socket to 0.0.0.0:1234 but when the client tries to connect to 127.0.0.1:1234 it gets errno 61 (connection refused). I then tried to simplify the scenario because what I actually need is just a loopback interface. I replaced vfs_client and vfs_server by a single vfs and routed the Nic session to nic_loopback. Which leads to the same result (connection refused) when the client tries to connect. What am I missing? Thanks, Roman

2 2

Sculpt: fail to build stdcxx port
by Boris Mulder 29 Jan '18

29 Jan '18

Hi all, I wanted to try out the sculpt scenario from https://github.com/nfeske/genode.git. However, when I create the depot archive for stdcxx with > tool/depot/create genodelabs/bin/x86_64/stdcxx CROSS_DEV_PREFIX=/usr/local/genode-gcc/bin/genode-x86- I get the following error: > Library stdcxx-c++98 > MERGE stdcxx-c++98.lib.a >/usr/local/genode-gcc/bin/genode-x86-ar: basic_file_stdio.o: No such file or directory I already tried deleting the contrib directory and running prepare_port again, as well as starting from a clean directory. Furthermore, it is just a clean checkout of the repo. Googling the error message did not give me any useful results. Any ideas what I should do? -- Met vriendelijke groet / kind regards, Boris Mulder Cyber Security Labs B.V. | Gooimeer 6-31 | 1411 DD Naarden | The Netherlands +31 35 631 3253 (office)

2 1

Time For "Community Development" ML?
by John J. Karcher 26 Jan '18

26 Jan '18

Hi Norman, What do you think of setting up a new "Community Development" mailing list, for discussion and coordination within the development community that doesn't need the attention of the core developers? We're probably already at the stage now where it would be helpful, but I think the "Year Of Sculpt" will generate much wider hobbyist interest than in the past, which could disrupt the S/N ratio of this list. (Eventually, an end-user oriented ML would probably be a good idea also, but that should be a year or more away.) Thanks! -- John J. Karcher devuser@...544...

2 2

Linux IO_MEM session
by Johannes Kliemann 26 Jan '18

26 Jan '18

Hi all, I'm currently implementing a IO_MEM session for base-linux. The basic idea is as follows: - core opens a file in /dev/ and calls an ioctl on the file to set a memory range. - it passes this file descriptor to the child process that requested the session - the child process mmaps the requested memory range from the file descriptor The required device on Linux is already available (if you're interested, I've written a kernel module to provide it in [1]) so that isn't the problem. I tried to implement the IO_MEM session on base-linux but lack the understanding of how the Linux_dataspace and the Io_mem_dataspace are different. On base the Io_mem_session_component gets the dataspace capability from > _ds_cap = static_cap_cast<Io_mem_dataspace>(_ds_ep->manage(&_ds)); This doesn't work with base-linux since I can't cast the Linux_dataspace to Io_mem_dataspace. Yet I don't understand what I need to do this. I also didn't find out what parts of the session are executed in which context (speaking what runs in core and what in the component itself), especially where I would put the actual mmap call. My current implementation of the IO_MEM session is at [2] and [3]. Regards, Johannes [1]: https://github.com/jklmnn/hwiodev [2]: https://github.com/jklmnn/genode/blob/0cfcec98ad4d4fdfc7bf6cc6e6df3567360b4… [3]: https://github.com/jklmnn/genode/blob/0cfcec98ad4d4fdfc7bf6cc6e6df3567360b4…

3 8

noux_bash failure
by genode user 25 Jan '18

25 Jan '18

I have tried to build the noux_bash run file using the sel4 , nova and linux kernel options and so far have not had any success. Nova and sel4 both fail with: [init -> drivers -> acpi_drv] Error: ROM-session creation failed (ram_quota=6144, cap_quota=3, label="platform_info") [init -> drivers -> acpi_drv] Error: Could not open ROM session for "platform_info" Warning: no io_mem support on Linux (args="base=0xe0000, size=0x20000, wc=no, diag=0, label="init -> drivers -> acpi_drv -> ", ram_quota=3827, cap_quota=2") [init -> drivers -> acpi_drv] Segmentation fault (signum=11), see Linux kernel log for details This looked similar to a problem I found here: https://sourceforge.net/p/genode/mailman/message/35973788/ but changing the amount of ram in the suggested file did not help. I also tried changing the amount of ram specified for noux in the run file with no change in behavior. With the linux kernel I was told I am missing the genodelabs/pkg/drivers_interactive-linux archive but installing it fails with: /home/omavuto/projects/genode-17.11/genode-17.11/depot/ genodelabs/api/base-linux-2017-11-24/src/lib/syscall/linux_syscalls.h:383:27: error: ‘__sigword’ was not declared in this scope _value[__sigword(signum)] = __sigmask(signum); It may be that this is related to another problem I found. Earlier in the process I was told to install the base-linux archive with this command: /home/omavuto/projects/genode-17.11/genode-17.11/tool/depot/download genodelabs/bin/x86_64/base-linux but I get the error: Error: failed to download 'https://depot.genode.org/ genodelabs/src/base-linux.tgz' it looks like that this archive does not exist. I found a similarly named archive called base-linux-2017-11-27 which I was able to download. I then simlinked the base-linux-2017-11-27 to baselinux in depot/genodelabs/src/. That seemed to satisfy the requirement but perhaps the archive I used is somehow incompatible. Any suggestions on getting this to start? I am most interested in sel4 but suggestions for nova and linux are appreciated as well. Thanks!

2 3

Installing Genode within container
by Subhadeep Sarkar 24 Jan '18

24 Jan '18

Hi, I am a researcher at Inria France and am presently exploring Genode as a potential option to perform hands on with Sel4. I am relatively new to Genode and having an issue while installing it. Here is the issue: I am trying to install Genode on Ubuntu (16.04 LTS) running within a container . I am able to successfully see through all the installation steps. However, when I issue the final command: make -C build/x86_32 KERNEL=okl4 run/demo , I get this following error. make[1]: Leaving directory '/Genode/genode/build/x86_32' genode build completed using 'core-okl4.o' as 'core.o' using 'ld-okl4.lib.so' as 'ld.lib.so' using 'pit_timer_drv' as 'timer' install bootloader creating ISO image... xorriso 1.4.2 : RockRidge filesystem manipulator, libburnia project. spawn qemu-system-i386 -no-kvm -display sdl -serial mon:stdio -m 512 -cdrom var/run/demo.iso Could not initialize SDL(No available video device) - exiting Aborting, received EOF Log step failed, retry. spawn qemu-system-i386 -no-kvm -display sdl -serial mon:stdio -m 512 -cdrom var/run/demo.iso -cdrom var/run/demo.iso qemu-system-i386: -cdrom var/run/demo.iso: drive with bus=1, unit=0 (index=2) exists Aborting, received EOF Log step failed, retry. spawn qemu-system-i386 -no-kvm -display sdl -serial mon:stdio -m 512 -cdrom var/run/demo.iso -cdrom var/run/demo.iso -cdrom var/run/demo.iso qemu-system-i386: -cdrom var/run/demo.iso: drive with bus=1, unit=0 (index=2) exists Aborting, received EOF Log step failed, retry. I did install libsdl1.2-dbg . Clearly, I do not have a display connected, as I am trying to run it from within a container. Could you please help me with this issue? Thank you in advance! Yours sincerely, Subhadeep ------------------------------------------------------------------------------------------- Subhadeep Sarkar, PhD Postdoctoral researcher, Univ Rennes, Inria, CNRS, IRISA Inria Rennes - Bretagne Atlantique Campus universitaire de Beaulieu 35042 Rennes cedex, France Website: subhadeep.cf E-mail: subhadeep.sarkar@...552... ; subhadeepsarkarybs@...54... Phone: +33 2 99 84 73 52

4 10

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

users January 2018