users November 2018

users@lists.genode.org

15 participants
26 discussions

sd_card_drv quota exceeded for Panda
by Matthias Adams 20 Nov '18

20 Nov '18

Hello everybody, I am a student from Munich and I am currently working on building some bare metal code for the Pandaboard M3 co-processor, while a Genode system, using the L4 kernel, is running on the A9. In order to load the bare metal application form the SD card I read some blocks using the sd_card driver. During the initialization of the driver the following output is produced: init] child "sd_card_drv" [init] RAM quota: 104722432 [init] ELF binary [init] priority: 0 [init] provides service Block [init] ram_quota=104722432 Warning: Quota exceeded! amount=81920, size=4069, consumed=81920 The Quota waring is coming from the allocator guard in 'repos/base/include/base/allocator_guard.h', but I have no idea which quota could have been exceeded, because the ram_quota should definitely be enough when reading only a few KiB. I would be grateful for any hints or ideas on why this is happening and how to solve this issue. Thanks, Matthias Adams

2 1

sd_card driver for i.mx53 time out
by lzSun 19 Nov '18

19 Nov '18

Hi,all I mentioned in previous email that I am trying to use the file system in the tz_vmm demo and use the SD card as a storage device which is formatted as ext2. Now I have included sd_card driver, part_blk, rump_fs etc. in the tz_vmm.run script which I show it in [1]. But when I try to create a file in tz_vmm, I get this error: [init -> tz_vmm] --- start test --- [init -> tz_vmm] current working directory : / [init -> sd_card_drv] Error: wait till issuing a new command is allowed timed out I have no idea how to fix this problem, so I want to ask for your help. I also provide my test code in [2]. In addition, I saw in this link: https://sourceforge.net/p/genode/mailman/genode-main/thread/CABDqRrJei5p5Jy… , Martin said "Be aware, that 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. Use the platform 'hw_imx53_qsb' instead when issuing 'create_builddir'. I know he is targeting genode 15.08, so I want to know how about this situation in Genode's latest version? If genode can directly drive the SD card now , does this meas the SD card can not accessed by linux because the ESDHC-IRQ is configured to be secure ? Thanks for your help in advance! [1] part of tz_vmm.run script: <start name="platform_drv"> <resource name="RAM" quantum="3M"/> <provides> <service name="Regulator"/> <service name="Platform"/> </provides> </start> <start name="sd_card_drv"> <resource name="RAM" quantum="3M"/> <provides><service name="Block"/></provides> </start> <start name="part_blk"> <resource name="RAM" quantum="10M" /> <provides><service name="Block" /></provides> <route> <service name="Block"><child name="sd_card_drv" /></service> <any-service><parent/><any-child/></any-service> </route> <config> <policy label_prefix="rump_fs" partition="1" writeable="yes"/> </config> </start> <start name="rump_fs" caps="200"> <resource name="RAM" quantum="16M"/> <provides><service name="File_system"/></provides> <config fs="ext2fs"> <policy label_prefix="" root="/" writeable="yes"/> </config> <route> <service name="Block"> <child name="part_blk"/> </service> <any-service> <parent/> <any-child/> </any-service> </route> </start> <start name="tz_vmm" caps="400"> <resource name="RAM" quantum="160M"/> <config> <vfs> <tar name="darknet.tar"/> <dir name="dev"> <log/> <inline name="rtc">2000-01-01 00:00</inline> </dir> <fs/> </vfs> <libc stdout="/dev/log" stderr="/dev/log" rtc="/dev/rtc" /> </config> [2] test code: int fd; ssize_t n; off_t offset; static char buf[16384]; static char str[] = "hello,world!"; printf("--- start test ---\n"); getcwd(buf, sizeof(buf)); printf("current working directory : %s\n", buf); fd = open("file.txt",O_WRONLY|O_CREAT); if (fd == -1) { perror("open"); return; } n = write(fd, str, sizeof (str)); printf("write %d bytes.\n",n); close(fd); printf("--- test finished ---\n"); ------------------ Best wishes

1 0

Re: optimization levels in genode
by lzSun 16 Nov '18

16 Nov '18

Hello,Stefan >Just out of curiousity, are you still using the tz_vmm.run script? And >if so, did you changed it to do: "run_genode_until forever", or does >it still use "run_genode_until <some_condition> <timeout>"? Yes, I am using tz_vmm.run script, and it sitll use the condition like this: # execute and wait for console run_genode_until {.*\/ #.*} 220 set serial_id [output_spawn_id] # wait for network to settle down send -i $serial_id "sleep 5\n" # test network send -i $serial_id "wget http://genode.org/\n" run_genode_until "Connecting to genode.org" 10 $serial_id run_genode_until {.*\/ #.*} 30 $serial_id Also, I tried to modify it to "run_genode_until forever" but I encountered the same problem.Please correct me if there are something wrong with those settings. ------------------ Best wishes

2 1

Request for feedback about "Genode Component Public License"
by Norman Feske 16 Nov '18

16 Nov '18

Dear Genode community, today, I'd like to present an idea that developed over the past three years during discussions with various contributors and users about the prospect of a Genode-based business ecosystem. As you certainly know, Genode Labs pursues a dual-licensing business model, which allows us to fund the development of Genode as an independent team. Our licensing business is enabled by the combination of the AGPLv3 as a strong copyleft license with the library-like nature of Genode. Until today, this model is limited to our company. To foster a sustainable ecosystem around Genode, we wish to enable others to pursue a similar business model while maintaining the spirit of open collaboration and free software. We eventually crafted a new license called "Genode Component Public License" (Genode CPL) specifically for components developed by 3rd parties, outside of Genode Labs. Please find the rationale, license text, and FAQ under the following link (which is not officially reachable from the website yet): https://genode.org/documentation/articles/component_public_license Before I go forward with publicly announcing the article at genode.org, I would very much appreciate your feedback, in particular: * How does the rationale and approach given in the description resonate with you? * Do you see any potential for controversies or uncertainties? * If you are in the business of producing proprietary software, would the Genode CPL create an incentive to open up your products without putting your business at risk? I'm very much aware that licensing is a delicate topic. To avoid hot-heated controversies and mis-interpretation, I encourage everyone to carefully read the "Questions and Answers" Section of the article before commenting. I'm looking forward to your feedback! Best regards 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 1

Re: rump_fs mount file system failed
by lzSun 14 Nov '18

14 Nov '18

Hi,Christian Thanks for your help. Now, the file system can be mounted. However, there's still errors in my test.That is File_system-session creation failed. Here's the output message: [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 14253 KB of RAM [init -> rump_fs] rump: /genode: file system not clean; please fsck(8) [init] child "rump_fs" announces service "File_system" [init -> test-file] Error: File_system-session creation failed (ram_quota=163840, cap_quota=12, tx_buf_size=131072, label="", root="", writeable=1) [init -> test-file] Error: failed to create <fs> VFS node [init -> test-file] --- start test --- [init -> test-file] Error: plugin()->open("file.txt") failed [init -> test-file] open [init -> test-file] : [init -> test-file] No such file or directory I don't know whether it's because there's some mistake in my configuration. Can you give me some hint? The run scripts is just like that I showed you before. And here's test file: #include <base/attached_rom_dataspace.h> #include <libc/component.h> /* libc includes */ #include <dirent.h> #include <fcntl.h> #include <stdio.h> #include <stdlib.h> #include <sys/stat.h> #include <sys/uio.h> #include <unistd.h> #include <string.h> #include <errno.h> static void test() { int fd; ssize_t n; off_t offset; static char buf[100]; static char str[] = "hello,world!"; printf("--- start test ---\n"); fd = open("file.txt",O_WRONLY|O_CREAT); if (fd == -1) { perror("open"); return; } n = write(fd, str, sizeof (str)); printf("write %d bytes.\n",n); close(fd); fd = open("file.txt", O_RDONLY); if (fd == -1) { perror("read open"); return; } n = read(fd, buf, sizeof (str)); printf("read %d bytes: ", n); for (size_t i = 0; i < sizeof (str); i++) printf("%c ", buf[i]); printf("\n"); close(fd); printf("--- test finished ---\n"); return; } struct Main { Main(Genode::Env &env) { Genode::Attached_rom_dataspace config_rom { env, "config" }; Libc::with_libc([&] () { test(); }); env.parent().exit(0); } }; void Libc::Component::construct(Libc::Env &env) { static Main main(env); } ------------------ Best wishes

2 1

Re: optimization levels in genode
by lzSun 14 Nov '18

14 Nov '18

Hello,Stefan Thanks for your effort! I have followed your advice to check to genode 18.08. I used the patch you provided to me: https://github.com/skalk/genode/tree/imx53-hf-neon At first, I just build the whole project as before. But there's an error at building stage: can't read "cc_arch": no such variable while executing "exec [cross_dev_prefix]gcc -print-libgcc-file-name {*}$cc_arch" (procedure "build_core" line 8) invoked from within Then, I add the "cc_arch" to the file: ./tool/run/run, like this: set asm_src [generate_boot_modules_asm [run_dir]/genode $modules] + set cc_arch "-mfloat-abi=hard -mfpu=neon" # determine the libgcc set libgcc [exec [cross_dev_prefix]gcc -print-libgcc-file-name {*}$cc_arch] After this, I can build the whole project successfully and measure the performance gain on the i.mx53 qsb board. However, there's still a problem. When my program have done and return to normal world, the system seems stuck. From the serial port, I can only see the linux's cursor blinking but I couldn't input any command.The serial port's output message like this: [init -> tz_vmm] task finished in 14588ms / # There are no errors which just like a command have executed normally. Do you have any idea about this? Thank you very much. ------------------ Best wishes

2 1

Multiboot with Genode Sculpt
by Johannes Schlatow 13 Nov '18

13 Nov '18

Fellow Sculpters, I'd like to install Sculpt in a multiboot setup in parallel to my Archlinux system. Do you have any recommendations/experiences w.r.t. where to put the grub files (esp. grub.cfg)? What's the best practice for maintainability/updateability in your opinion? Cheers Johannes

3 3

rump_fs mount file system failed
by lzSun 13 Nov '18

13 Nov '18

Hello, I want to use a file system in tz_vmm demo. I get to know that there's sd_card driver for i.mx53 quick start board and a stable file system in genode which is rump_fs. So my idea is to use rump_fs and use sd_card for storage. The sd_card is formatted as ext2 and it is also used to booting the system. I write a simple test client called "test-file" which just try to create a file and open it.However, when the system booting, it print the error message: [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 14249 KB of RAM [init -> rump_fs] Error: mounting file system failed (errno 13 ) [init -> rump_fs] Error: dry mount attempt failed, aborting [init -> rump_fs] Error: Uncaught exception of type 'Genode::Exception' [init -> rump_fs] Warning: abort called - thread: ep [init] child "rump_fs" exited with exit value 1 I searched the error number online which tells me that it seems related to "permission".Here's part of my configuration: <start name="sd_card_drv"> <resource name="RAM" quantum="3M"/> <provides><service name="Block"/></provides> </start> <start name="part_blk"> <resource name="RAM" quantum="10M" /> <provides><service name="Block" /></provides> <route> <service name="Block"><child name="sd_card_drv" /></service> <any-service><parent/><any-child/></any-service> </route> <config> <policy label_prefix="rump_fs" partition="1"/> </config> </start> <start name="rump_fs" caps="200"> <resource name="RAM" quantum="16M"/> <provides><service name="File_system"/></provides> <config fs="ext2fs"> <policy label="" root="/" writeable="yes"/> </config> <route> <service name="Block"> <child name="part_blk"/> </service> <any-service> <parent/> <any-child/> </any-service> </route> </start> <start name="test-file" caps="100"> <resource name="RAM" quantum="4M"/> <config> <vfs> <dir name="dev"> <log/> <inline name="rtc">2000-01-01 00:00</inline> </dir> <fs/> </vfs> <libc stdout="/dev/log" stderr="/dev/log" rtc="/dev/rtc" /> </config> </start> Any help will be ok, thanks in advance! ------------------ Best wishes

2 1

Re: optimization levels in genode
by lzSun 12 Nov '18

12 Nov '18

Hello,Stefan >did you followed my advice to first enable the FPU in the kernel? >Otherwise it is totally clear to not work at all, because -O3 compiledcode make heavily use of the >FPU. Thanks for your patient reply. I have tried to compile the hard-float genode toolchain by modifying the repos/tool/tool_chain like this: -TARGET_NAME_arm = arm-none-eabi +TARGET_NAME_arm = arm-none-eabihf And I use tool_chain script to compile hard-float compiler by "./tool/tool_chain arm" command. Then , I followed your advice to enable the FPU in the kernel according to this patch[1]. But when I want to build the project by "make run/tz_vmm", the compiling stage seems completed, but there's error occurred in linking stage.The error message like this: genode build completed using 'hw_timer_drv' as 'timer' using 'ld-hw.lib.so' as 'ld.lib.so' core link address is 0x80000000 /usr/local/genode-gcc2/bin/../lib/gcc/arm-none-eabihf/6.3.0/../../../../arm-none-eabihf/bin/ld: error: var/run/tz_vmm.core uses VFP register arguments, /usr/local/genode-gcc2/bin/../lib/gcc/arm-none-eabihf/6.3.0/libgcc.a(_fixsfdi.o) does not I am a little confused. I think I have build the hard-float compiler and enabled the FPU in the kernel. Did I missed something? I also tried to change the flag "-mfloat-abi=hard" to "-mfloat-abi=softfp" in repos/base/mk/global.mk and repos/base/mk/spec/arm_v7a.mk, then the project can be build successfully but still stuck at "Starting kernel ..." at booting stage. >I've compiled a hard-float compiler like advised to you, took that topic branch, and recognized that for the i.MX53 one needs to set the >-mfpu flag to neon. Then I could measure the performance gain on the i.MX53 Quickstart board. I've updated my topic branch to be >ready-to-merge, and prepared a topic branch that uses hard-float neon FPU at least on i.MX53 here: Thank you very much for providing such a patch to me. But since the patch you provided is based on genode 18.08 and I am using 18.05 now, I can't use "git diff" command to see which files are modified. Because there are too many file changes. And it seems that the patch you provided can not be used by me directly. So, as I said before, I refer to this patch[1] to enable FPU in kernel.At beginning, I just modify the file that I think related to i.mx53 ,here's the list of those file. Then, I also tried to modify all the file related to the patch[1] but the result have not changed . I know my questions are quite more and odd, please do not mind and thanks for your patience. [1] https://github.com/skalk/genode/commit/cf02346cef59a6752c9a07494d812c8ada70… [2] repos/base-hw/lib/mk/spec/armv7/core-hw.inc repos/base-hw/src/bootstrap/spec/arm/crt0.s repos/base-hw/src/core/spec/arm/cpu_support.h repos/base-hw/src/core/spec/arm/exception_vector.s repos/base-hw/src/core/spec/arm/kernel/thread.cc repos/base-hw/src/core/spec/arm/vfpv3-d32.s repos/base/mk/spec/arm_v7a.mk ------------------ Best wishes

2 1

sd_card driver in genode
by lz Sun 07 Nov '18

07 Nov '18

Hello, I am now using tz_vmm demo on imx53-qsb board. Now, I want to use file system session in tz_vmm and use SD card for storage medium. I know there's a SD card driver [1] for imx53 board and I also find a related topic in this link: https://sourceforge.net/p/genode/mailman/message/34667551/. However, this link may be a few times ago, so the reference file [2] can not be find in my Genode source tree(18.05). I can only find some related file like [3] ,but I am a little confuse. Because in repos/libports/run/libc_filesystem_test.inc, there's no target like "server/libc_fatfs" , but in boot_modules , it contains " libc_fatfs.lib.so". I have tried append target "test/libc_fatfs test/libc_block" like in libc_filesystem_test.inc ,but when building the project, it print the error messge like: Error: Ports not prepared or outdated: fatfs And when I use ./tool/ports/prepare_port fatfs, there's another error: Error: Hash sum check for fatfs failed /genode/tool/ports/mk/install.mk:197: recipe for target 'fatfs.file' failed I have tried ./tool/ports/update_hash fatfs but got the same result.So I wonder how libc_fatfs can be used? Besides, I find the file repos/libports/run/libc_block.run which is a script for testing sd card driver I think. So I modify tz_vmm.run script as follows : 1. append targets " drivers/sd_card" and append boot_modules "sd_card_drv" 2.add config <start name="sd_card_drv"> <resource name="RAM" quantum="1M" /> <provides><service name="Block"/></provides> </start> <start name="my_server"> <vfs> <dir name="dev"> <log/> <block name="blkdev"/> </dir> </vfs> </start> 3. In my own program, I write a test file like in repos/libports/src/test/libc_block/main.cc, but it print the message like this: [init -> my_server] offset: 8193 [init -> sd_card_drv] Error: wait till issuing a new command is allowed timed out. I have no idea why this error occurred. In general , could you please give me some hint that how I can use file system session in tz_vmm? Thanks. [1] repos/os/src/drivers/sd_card/spec/imx53 [2] repos/libports/run/libc_ffat_fs.run repos/libports/run/libc_ffat.run [3] repos/libports/run/libc_fatfs.run repos/libports/run/libc_filesystem_test.inc

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

users November 2018