users

users@lists.genode.org

2 participants
2322 discussions

Activate Co-Processor for Bare-Metal via Genode
by Matthias Adams 06 Dec '18

06 Dec '18

Hello everyone, I am currently working on a Pandaboard ES, on which a Genode 18.02 runs using a L4 Kernel and I would like to use the Co-Processor of the Pandaboard (namely Cortex M3). I used an example like [1] in order to write my application for the M3, which works perfectly when loaded by the program suggested in [1]. [1] https://github.com/apopple/Pandaboard-FreeRTOS In order to get things working, when a Genode system is running on the main Processor (Cortex A9) I wrote a "driver" similar to the gpio driver, which should set-up the MMU for the M3 and start it, but absolutely nothing happens. The driver compiles and runs without any warnings or errors. Reads and writes to physical addresses look good to me, but the M3 does not seem to start (it should blink an LED for now). I triple checked all physical addresses, which I added to 'repos/base/include/drivers/defs/panda.h': /*M3 */ CM_MPU_M3_CLKSTCTRL_BASE = 0x4a008900, CM_MPU_M3_CLKSTCTRL_SIZE = 0x40, RM_MPU_M3_RSTCTRL_BASE = 0x4a306900, RM_MPU_M3_RSTCTRL_SIZE = 0x20, M3_L2MMU_BASE = 0x55082040, M3_L2MMU_SIZE = 0x40 I used these bases, in order to create my Registers like: struct M3_CM_MPU_reg : Genode::Attached_io_mem_dataspace, Genode::Mmio { M3_CM_MPU_reg(Genode::Env &env, Genode::addr_t const mmio_base, Genode::size_t const mmio_size) : Genode::Attached_io_mem_dataspace(env, mmio_base, mmio_size), Genode::Mmio((Genode::addr_t)local_addr<void>()) { } struct Clkstctrl : Register<0x00, 32> {}; struct Clkctrl : Register<0x20, 32> {}; }; struct M3_RM_MPU_reg : Genode::Attached_io_mem_dataspace, Genode::Mmio { M3_RM_MPU_reg(Genode::Env &env, Genode::addr_t const mmio_base, Genode::size_t const mmio_size) : Genode::Attached_io_mem_dataspace(env, mmio_base, mmio_size), Genode::Mmio((Genode::addr_t)local_addr<void>()) { } struct Rstctrl : Register<0x10, 32> {}; }; struct M3_L2MMU_reg : Genode::Attached_io_mem_dataspace, Genode::Mmio { M3_L2MMU_reg(Genode::Env &env, Genode::addr_t const mmio_base, Genode::size_t const mmio_size) : Genode::Attached_io_mem_dataspace(env, mmio_base, mmio_size), Genode::Mmio((Genode::addr_t)local_addr<void>()) { } struct Cntl : Register<0x4, 8> {}; struct Ttb : Register<0xc, 32> {}; }; The registers are then written and read in my driver like below: class Omap4_driver : public M3::Driver { private: Genode::Env &_env; M3_CM_MPU_reg _m3_cm_mpu; M3_RM_MPU_reg _m3_rm_mpu; M3_L2MMU_reg _m3_l2mmu; Omap4_driver(Genode::Env &env) : _env(env), // see repos/base/include/drivers/defs/panda.h _m3_cm_mpu(_env, Panda::CM_MPU_M3_CLKSTCTRL_BASE, Panda::CM_MPU_M3_CLKSTCTRL_SIZE), _m3_rm_mpu(_env, Panda::RM_MPU_M3_RSTCTRL_BASE, Panda::RM_MPU_M3_RSTCTRL_SIZE), _m3_l2mmu(_env, Panda::M3_L2MMU_BASE, Panda::M3_L2MMU_SIZE) { } public: static Omap4_driver& factory(Genode::Env &env); /****************************** ** M3::Driver interface ** ******************************/ void setup_l2mmu(Genode::addr_t ds_phys_addr, Genode::Ram_dataspace_capability ttb_cap, Genode::addr_t ttb_phys_addr) { Genode::addr_t ttb_addr = _env.rm().attach(ttb_cap); Genode::log("L2: attached dataspace"); /* Brings the L2 MMU out of reset and sets up virtual addresses */ unsigned clkctrl = _m3_cm_mpu.read<M3_CM_MPU_reg::Clkctrl>(); clkctrl = 0x01; _m3_cm_mpu.write<M3_CM_MPU_reg::Clkctrl>(clkctrl); unsigned clkstctrl = _m3_cm_mpu.read<M3_CM_MPU_reg::Clkstctrl>(); clkstctrl = 0x02; _m3_cm_mpu.write<M3_CM_MPU_reg::Clkstctrl>(clkstctrl); unsigned rstctrl = _m3_rm_mpu.read<M3_RM_MPU_reg::Rstctrl>(); rstctrl &= ~0x4; _m3_rm_mpu.write<M3_RM_MPU_reg::Rstctrl>(rstctrl); int *TTB = ((int*) ttb_addr); /* Initialise address translation table */ /* Every address will be mapped to M3_LOAD_ADDRESS */ for(int i = 0; i < 4096; i++) { TTB[i] = (((ds_phys_addr) & 0xfff00000) | 0x00c02); } /* Set the last entry to point to the GPIO1 peripheral to flash a light */ /* ie. Virtual address of GPIO1_DATAOUT is 0xfff1013c */ TTB[4095] = 0x4a300c02; /* Set the base address */ Genode::log("physAddr: ", Genode::Hex(ttb_phys_addr)); _m3_l2mmu.write<M3_L2MMU_reg::Ttb>(ttb_phys_addr); /* Enable MMU */ _m3_l2mmu.write<M3_L2MMU_reg::Cntl>(0x6); Genode::log("L2 enabled"); } void start_m3() { /* Release Cortex-M3 from reset */ unsigned rstctrl = _m3_rm_mpu.read<M3_RM_MPU_reg::Rstctrl>(); rstctrl &= ~1; _m3_rm_mpu.write<M3_RM_MPU_reg::Rstctrl>(rstctrl); } }; Any idea on how to get things working or on how to further debug would be great! Thanks, Matthias

1 0

How to pass command-line arguments to Genode application?
by Wouter van Oijen 03 Dec '18

03 Dec '18

Hello, I want to pass one or more command-line arguments to my Genode application. So instead of starting "mosquitto", I want to run "mosquitto -p 1888", for example. How do I do this? I've found run files that pass arguments to Noux applications (using a start-element inside the config-element), but this does not seem to work for my application. So how are command-line arguments passed to the application in general? Below is the part of the init configuration that I currently use to start the application. <start name="mosquitto"> <resource name="RAM" quantum="100M"/> <config> <vfs> <dir name="dev"> <log/> <rtc/> </dir> <dir name="socket"> <lwip ip_addr="10.0.2.55" netmask="255.255.255.0" gateway="10.0.2.1"/> </dir> </vfs> <libc stdout="/dev/log" stderr="/dev/log" rtc="/dev/rtc" socket="/socket" /> </config> </start> Regards, Wouter

2 2

Announcement: Genode OS Framework version 18.11 released
by Norman Feske 29 Nov '18

29 Nov '18

I am happy to announce the release of Genode 18.11. This version is focused on two topics that were prominently mentioned on the road map for 2018, namely software quality and the use of Genode for network appliances and server applications. For the users of Sculpt OS among you, the release comes with new window-management features that wait for your exploration. ;-) In short, the highlights of the new version are: - Quality assurance - Automated test infrastructure hosted on top of Genode - New tooling for obtaining test-coverage metrics - Static code analysis - Genode as a platform for Mirage-OS unikernels - Initial version of the Genode SDK - Component health monitoring - Modularized window system in Sculpt OS - Zero-copy networking on Xilinx Zynq - New SSH terminal server - New window layouter You can find the comprehensive documentation for the version 18.11 here: https://genode.org/documentation/release-notes/18.11 Have fun with the new version! Norman -- Dr.-Ing. Norman Feske Genode Labs https://www.genode-labs.com · https://genode.org Genode Labs GmbH · Amtsgericht Dresden · HRB 28424 · Sitz Dresden Geschäftsführer: Dr.-Ing. Norman Feske, Christian Helmuth

1 0

Re: use sd_card driver in imx53_qsb_tz platform
by lzSun 28 Nov '18

28 Nov '18

Hi,Stefan >> I am wondering if I can config mmc0 and mmc1 to be secure and non-secure respectively.In this way, the mmc1 can be used by linux kernel directly. Thank you in advance! >Yes, this is possible. Our CSU class instantiation does not provide this however. We use the provided boolean to configure all SDHC controller registers to be secure or not. If you want to distinguish >in between them, please look into the CSU implementation where the boolean value is used, and inside the Security Reference Manual of the i.MX53 to target the corresponding registers. I am grateful for your help. I have achieve this goal by modify the first SDHC controller registers to be secure, and the genode in secure world can use SD card then. Thanks again. By the way, I find a little problem. This method can work correctly with the genode patch without using hard-float on TZ_VMM demo. When I changed to the FPU patch that you provided to enable kernel to use hard-float [1], the linux in normal world can not boot successfully. It will stuck at near this: " VFP support v0.3: implementor 41 architecture 3 part 30 variant c rev 2 mxc_rtc mxc_rtc.0: setting system clock to 1970-01-01 00:00:00 UTC (0) Freeing init memory: 192K " I am little puzzled why changing the SDHC register to secure will affect the normal startup of Linux in FPU patch though you have mentioned that this patch did not encounter the trustzone feature. Anyway, thanks for your effect and wish your work on this aspect progressing well. ------------------ Best wishes

1 0

geteuid / getpwnam not implemented
by Wouter van Oijen 27 Nov '18

27 Nov '18

Hello, I succeeded in building and starting Mosquitto under Genode, but it fails due to the not-implemented functions geteuid() and getpwnam(). See log below. Are these functions already implemented somewhere? How should I continue from here? Also, what does the warning about exec_static_constructors() mean? [init -> mosquitto] lwIP Nic interface up address=10.0.2.55 netmask=255.255.255.0 gateway=10.0.2.1 [init -> mosquitto] Warning: Don't call Genode::Env::exec_static_constructors() in components without static globals [init -> mosquitto] 1543318260: mosquitto version 1.5.4 starting [init -> mosquitto] 1543318260: Using default config. [init -> mosquitto] 1543318260: Opening ipv4 listen socket on port 1883. [init -> mosquitto] Warning: geteuid: geteuid not implemented [init -> mosquitto] Warning: getpwnam: getpwnam not implemented [init -> mosquitto] 1543318260: Error: Invalid user 'mosquitto'. [init] child "mosquitto" exited with exit value 1 Best regards, Wouter

3 3

Porting Mosquitto to Genode: no plugin found for socket()
by Wouter van Oijen 26 Nov '18

26 Nov '18

Hello, I'm currently porting Mosquitto (https://mosquitto.org/) to Genode. I've managed to compile and run the application, but I get the following error: (...) genode build completed using 'core-linux' as 'core' using 'ld-linux.lib.so' as 'ld.lib.so' spawn ./core Genode 18.08 17592186044415 MiB RAM and 8998 caps assigned to init [init -> mosquitto] Warning: rtc not configured, returning 0 [init -> mosquitto] Error: no plugin found for socket() [init -> mosquitto] Error: no plugin found for socket() [init] child "mosquitto" exited with exit value 1 This is probably because the socket() function in libc is just a stub and needs a plugin to work correctly. I'm not sure how to continue from here. As far as I know, there are different ways to implement the sockets in Genode (for example using VFS). Is there any (up-to-date) documentation about these implementations? What are the advantages and disadvantages of each approach, and what other steps are required to get it working (there seem to be a lot of dependencies when including the socket interface...). I've tried following the example in run/usb_net [1], which uses lib/vfs/lwip, but then I get the following error: invalid command name "usb_host_drv_binary" I cannot find where usb_host_drv_binary is coming from (it's referred to in this file, but I don't know where it is generated). Besides, I don't know if this example is working anyway, so maybe this is not the right direction... [1] https://github.com/genodelabs/genode/blob/master/repos/dde_linux/run/usb_ne… Following another example from the mailing list, I've tried adding LIBS += libc_lwip_nic_dhcp, but this currently results in the following errors: (...) genode build completed using 'core-linux' as 'core' using 'linux_timer_drv' as 'timer' using 'ld-linux.lib.so' as 'ld.lib.so' spawn ./core Genode 18.08 17592186044415 MiB RAM and 8998 caps assigned to init [init -> mosquitto] using the lwIP libc plugin [init -> mosquitto] init [init -> mosquitto] Using DHCP for interface configuration. [init] Warning: mosquitto: no route to service "Nic" [init -> mosquitto] Error: Nic-session creation failed (ram_quota=671744, cap_quota=8, tx_buf_size=204800, rx_buf_size=204800, label="") [init -> mosquitto] Warning: NIC not available, loopback is used as default [init -> mosquitto] Warning: rtc not configured, returning 0 [init -> mosquitto] Warning: geteuid: geteuid not implemented [init -> mosquitto] Warning: getpwnam: getpwnam not implemented Warning: blocking canceled in entrypoint constructor [init] child "mosquitto" exited with exit value 1 Based on this log output, to get this working I would need to: * provide the Nic service in some way * find a library which implements geteuid() and getwpnam() (I guess this would require Noux, right?) In both cases I also have the warning about rtc nog configured. How do I resolve this? Any pointers in the right direction are very welcome :-) Kind regards, Wouter

3 5

Re: use sd_card driver in imx53_qsb_tz platform
by lzSun 26 Nov '18

26 Nov '18

Hi,Stefan >Surely this is the right way to go. All ingredients are already in place. Have a look at the usb_armory platform, which uses exactly the same SoC i.MX53 with a trustzone Linux using the sd-card client-wise from the Genode secure-world. >You need to change the sd-card device assignment in: `repos/base-hw/src/bootstrap/spec/imx53_qsb/platform_trustzone.cc` >The fourth parameter of the Csu object decides over the sd-card assignment. And of course you need to change the Linux kernel to use the paravirtualized block device of the trustzone VMM. >Please use the Linux kernel version + adaptations from here as inspiration: https://github.com/m-stein/linux/tree/genode_hw_usb_armory_tz_vmm Thank you very much for your help. Now I am trying to modify the linux kernel according to your advice. And I have another question want for your help. Since i.mx53 quick start board have two SD card slot, which are MicroSD card slot(mmc0) and SD card slot(mmc1) respectively.So, can I configure them to be secure or non-secure seperately?In other words, does the CSU object in the repos/base-hw/src/bootstrap/spec/imx53_qsb/platform_trustzone.cc configure both mmc0 and mmc1 to be secure or non-secure at the same time? Because, from the system booting message, I can only find the mmc0 which is also used for booting that is found by the SD card driver, but there's no information about mmc1. I have insert SD card to both mmc0 and mmc1 and part of the booting message are as follows: [init -> platform_drv] --- i.MX53 platform driver --- [init -> sd_card_drv] --- SD card driver --- [init] child "platform_drv" announces service "Platform" [init] child "sd_card_drv" announces service "Block" [init -> sd_card_drv] CID: 0x3534453 0x55303447 0x80b0970d 0x4000e700 [init -> sd_card_drv] RCA: 0xe624 [init -> sd_card_drv] SD card detected [init -> sd_card_drv] capacity: 3781 MiB [init -> part_blk] Partition 1: LBA 10240 (7733248 blocks) type: 83 [init] child "part_blk" announces service "Block" [init -> part_blk] session opened at partition 1 for 'rump_fs -> ' [init -> rump_fs] Using ext2fs as file system [init -> rump_fs] asserting rump kernel 14109 KB of RAM [init -> rump_fs] rump: /genode: file system not clean; please fsck(8) I am wondering if I can config mmc0 and mmc1 to be secure and non-secure respectively.In this way, the mmc1 can be used by linux kernel directly. Thank you in advance! ------------------ Best wishes

2 1

hash sum check failed for atari800 port
by Wouter van Oijen 22 Nov '18

22 Nov '18

Hello, I was just trying to prepare the "atari800" port (from genode-world) but the hash sum check failed (see logging below). I can easily fix the hash to make it work at my system, but I'm surprised that the current hash is only 40 characters (instead of 64 characters that are expected for a sha256sum). If it should be a sha256sum then the port will obviously not work for anybody, and should fail in any automated regression test. I have the feeling I'm missing something here... If I'll update the hash, will that break the build on another system? Why is the current hash only 40 characters long? Regards, Wouter LOGGING: vagrant@ubuntu-bionic:~/genode$ ./tool/ports/prepare_port atari800 atari800 update src/app/atari800 Error: Hash sum check for xformer failed /home/vagrant/genode/tool/ports/mk/install.mk:197: recipe for target 'xformer.file' failed make[2]: *** [xformer.file] Error 1 /home/vagrant/genode/tool/ports/mk/prepare_single_port.mk:38: recipe for target '_install_in_port_dir' failed make[1]: *** [_install_in_port_dir] Error 2 tool/ports/prepare_port:33: recipe for target 'atari800' failed make: *** [atari800] Error 2 vagrant@ubuntu-bionic:~/genode$ grep xformer repos/world/ports/atari800.port DOWNLOADS := atari800.git xformer.archive URL(xformer) := http://prdownloads.sf.net/atari800/xf25.zip SHA(xformer) := e65923fd3a4845c1c72d1677db0fda969187eb42 DIR(xformer) := src/app/atari800/xformer vagrant@ubuntu-bionic:~/genode$ sha256sum -b - < contrib/atari800-*/xf25.zip 98ae0ad10413dd6f35ed80f5662dba6d790def70c7829046e52012a03b574b8a *-

2 3

use sd_card driver in imx53_qsb_tz platform
by lzSun 22 Nov '18

22 Nov '18

Hello ,Genode Community I have written an email before which has a similar question to this email, but I haven't receive any response yet. Because solving this problem is very important to my work , so I decide to put it out again and hope to receive some help. The thing is I want to use sd_card driver in i.mx53 quick start board with trustzone. I know Martin said " you should not run 'sd_card' with Genodes Trustzone feature-set (platform 'hw_imx53_qsb_tz'), because Genode will then declare the ESDHC-IRQ to be unsecure to enable Linux to drive SD cards."[1]. But unfortunately, my work is just needs to use sd_card driver with trustzone. So I wonder is this possible? I am now using the latest Genode version(18.08). Form my point of view, if ESDHC-IRQ is still to be unsecure, this's two ways to solve this problem: 1. set ESDHC-IRQ to be secure. In this way ,Genode OS in secure world can directly drive the SD card. Also, the linux in the normal world must be modified to the stat that could not drive SD card. If this is right, could you provide me some hint to this method? 2. forward the sd card request to linux in the normal world.So all the sd card operation will be performed by linux. However, I think this will introduce more world switch between two worlds and you should provide a linux driver to serve for Genode's file operation. I will be grateful if you could give me some help, thanks a lot! ------------------ Best wishes

2 1

Re: optimization levels in genode
by lzSun 22 Nov '18

22 Nov '18

Hi,Stefan >The run-script you are using is an automated test, which remote-controls the test board by using >Tcl/expect. Therefore, if you use the unmodified run-script to control the Linux instance via serial line >manually, it won't work. The 'expect' mechanism holds control over the serial line I/O. That means >everything you type in will not be sent to the other side. >If you want to use the test-scenario in a non-automated fashion, you either replace the very first >"run_genode_until ..." with "run_genode_until forever", or you simply do not use the run-script to >connect to your test-board, but use a terminal emulator like "picocom" directly. >Please, first evaluate whether it accepts input events with "direct" serial-line access. I'm wondering that >you see a _blinking_ terminal but no input events are shown. Sorry,but I can not fully understand what you mean. Now I am using i.mx53 quick start borad and the borad is connected to my PC through it's UART DB9 connector. I managed this through a USB to serial converter. Also , I use a serial debug tool that is "minicom" to get the output of the board and input my command.Through the minicom , I can input command to linux , such as "ls mkdir wget... etc" , also ,I can open the kernel module that I implemented to issue "smc" to switch to tz_vmm to execute some code. It works fine when I use the genode 18.05. However ,when I changed to the hard-float version that you provided , I found this odd problem. Also, I have found that even I din't input any command the linux will get stuck after a period of time after startup. I have tried to replace the "run_genode_until..." with "run_genode_until forever" but the result haven't changed. This problem has bothered me for a few days, I will be very grateful if you could give me some help. ------------------ Best wishes

2 1

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

users