- l4-hackers - example.com Site

Trying to build L4 on NixOS
by Mateusz Czaplinski 09 Apr '17

09 Apr '17

I'm trying to build L4 on NixOS in a VM (long term goal: try to learn how to replace NixOS's kernel with L4Linux, then maybe Genode). I'm getting an error like below: undefined reference to `__stack_chk_fail' during linking of bootstrap.elf - in a GCC's error-barf just after: [bootstrap] ==> Linking bootstrap.elf I've put up the precise Nix expression/script I'm using to build it on github - at: https://github.com/akavel/l4.nix/blob/0108dfcf8d84780e1ba884c41ece0d2dbf4dd… but what it generally boils down to is a sequence of commands like this: $ wget http://os.inf.tu-dresden.de/download/snapshots/l4re-core-2016082114.tar.xz $ tar xf l4re-core-2016082114.tar.xz $ cd l4re-core-2016082114 $ mkdir out $ cp src/l4/mk/defconfig/config.amd64 out/l4/.kconfig $ make -C src/l4 O=out/l4 olddefconfig $ make -C out/l4 I've tried various additional changes, based on some snippets found here and there on teh Internets, like: - adding V=0 to the last make (no idea what it does yet, just some cargo culting) - adding NIX_CFLAGS_COMPILE = "-fno-stack-protector"; to the Nix expression/script (also not really sure where it applies in the stack); but none of the above helped. Any idea what I could try doing to push it further? Thanks, /Mateusz.

2 7

Questions about Fiasco.OC scheduler and L4Linux
by Marc CHALAND 02 Apr '17

02 Apr '17

Hi, I'm doing a study to port old software on top of Fiasco.OC. This soft uses routines which blocks thread scheduling between EnterRegion and LeaveRegion call. Is there any equivalent on TUDOS ? Did somebody already get this challenge ? I would like to know if L4Linux uses Fiasco.OC scheduler or a user land one ? For our i.MX7d board, Freescale provides Linux drivers on 4.1. I don't know yet how many drivers we will need. Do you think it is better to port drivers from 4.1 to L4Linux 4.3 or later or get a L4Linux 4.1. I have no idea about the complexity on doing this. Regards Marc

4 13

L4linux __raw_writesl/readl undefined
by Manolis Ragkousis 01 Apr '17

01 Apr '17

Hello everyone, I am trying to enable sd card support on my zedboard running l4linux and I get this error: ERROR: "__raw_writesl" [drivers/mmc/host/mmci.ko] undefined! ERROR: "__raw_readsl" [drivers/mmc/host/mmci.ko] undefined! make[2]: *** [../scripts/Makefile.modpost:91: __modpost] Error 1 make[1]: *** [/home/manolis/repos/l4linux/Makefile:1173: modules] Error 2 make[1]: *** Waiting for unfinished jobs.... LD ldrImage GZIP vmlinuz LOCINST vmlinuz make[1]: Leaving directory '/home/manolis/repos/l4linux/build' make: *** [Makefile:150: sub-make] Error 2 In order to enable sd card support I understand that I need to enable MMC/SD/SDIO card support in the kernel + ARM AMBA Multimedia Card Interface support. I have followed the instructions from previous emails in this list [1][2]. What am I missing? Is there a config option which enables raw writes/reads on top of L4? Thank you for you time and help, Manolis [1] https://os.inf.tu-dresden.de/pipermail/l4-hackers/2014/006296.html [2] https://os.inf.tu-dresden.de/pipermail/l4-hackers/2014/006417.html

2 2

Can't find device node in /dev/ directory!
by li94575 20 Mar '17

20 Mar '17

Hi, I added some drivers in l4linux, of course, I assigned the corresponding peripheral resources (iomem, irq) to these drivers. From the driver's output log, it seems that they all work well, but when entering the command line, I could not find any relevant device node in /dev/ directory. To my confusion, I can see these devices info in /sys/ directory. The command line args is showing below: mem=64M console=ttyLv0 l4x_rd=rom/ramdisk-" .. L4.Info.arch() .. ".rd ramdisk_size=4000 Any idea?

5 6

Re: Extending l4sys by reading/writing register values of a preempted thread
by Christian Ludwig 20 Mar '17

20 Mar '17

Hi, > I want to implement a new syscall for getting/setting stored register > values of a specific thread (identified by its cap). Why not use ex_regs() on the thread to trigger an exception. That way, the attached userland exception handler can introspect the thread's complete register set. It can also alter the registers, so that the thread continues to run with the modified register set. Is there anything more you need? > Before implementing the syscall, I need to better understand the > kernel-part of Fiasco.OC. Can you help me answer the following questions > and point me to the source code regarding each question: > > * Where (in the source code) are threads preempted and their register > state stored? A thread's userland registers are stored on every entry to the kernel. The class in Fiasco is called Entry_frame. The userland exception handler mentioned above modifies that Entry_frame. > * Where is the register state of a thread loaded to the CPU and the > thread started (e.g. after a preempt)? When the scheduler selects a thread, its userland state (Entry_frame) is restored on exit from kernel mode. > * How can I introspect the stored thread register of a thread in the > kernel debugger JDB? > ** My thoughts: Is it K<kobj_ptr> with kobj_ptr of the thread and the 3 > lines beginning with PC=...? > e.g. > " > PC=010579cc USP=200ffdb0 smlatbeq r9, r0, r2, r5 > [0] 00000003 000010c0 00218003 00000000 tsteq000r5, ip, rrx > [8] 200ffdb0 200ffe40 01086ef0 200ffe98 [c] 200ffef8 010579cc fffffff8 > 20000010 > " Yes, that PC line is the userland program counter. the USP is the userland stack pointer. All userland registers are located at the top of kernel stack for a particular thread. > I also have another, short, off-topic question: > * How can I introspect the capability space of a task inside the kernel > debugger? In short mode, press 's' for a list of all tasks. Select the task in question and the press 'o' to see the capability space for that task. Hope that helps, - Christian

3 2

Extending l4sys by reading/writing register values of a preempted thread
by Denis Huber 13 Mar '17

13 Mar '17

Dear L4 hackers community, I want to implement a new syscall for getting/setting stored register values of a specific thread (identified by its cap). Preliminary note: I'm using the ARM architecture. Before implementing the syscall, I need to better understand the kernel-part of Fiasco.OC. Can you help me answer the following questions and point me to the source code regarding each question: * Where (in the source code) are threads preempted and their register state stored? * Where is the register state of a thread loaded to the CPU and the thread started (e.g. after a preempt)? * Where is the register state stored of each thread? * How can I introspect the stored thread register of a thread in the kernel debugger JDB? ** My thoughts: Is it K<kobj_ptr> with kobj_ptr of the thread and the 3 lines beginning with PC=...? e.g. " PC=010579cc USP=200ffdb0 smlatbeq r9, r0, r2, r5 [0] 00000003 000010c0 00218003 00000000 tsteq000r5, ip, rrx [8] 200ffdb0 200ffe40 01086ef0 200ffe98 [c] 200ffef8 010579cc fffffff8 20000010 " I also have another, short, off-topic question: * How can I introspect the capability space of a task inside the kernel debugger? Kind regards, Denis

1 0

Run L4Re/Fiasco hello example under QEMU (ARM)
by Piotr Piwko 05 Mar '17

05 Mar '17

Hi, I would like to run hello example under qemu-system-arm, so I successfully built the images using gcc-linaro-6.3.1 toolchain with following configuration: L4Re: Target Architecture (ARM architecture) CPU variant (ARMv7A type CPU) Platform Selection (ARM Integrator) Fiasco: Architecture (ARM processor family) Platform (ARM Integrator Platform) CPU (ARM 926 Processor) Unfortunately, after booting boostrap_hello.elf image under qemu-system-arm (the latest 2.8.0 version) there is nothing printed out on the console: $ QEMU_OPTIONS="-machine integratorcp -m 256 -serial mon:stdio -nographic" make qemu E=hello MODULE_SEARCH_PATH=/path/to/fiasco/build/ Do you have any idea what I omitted or what is wrong in my setup? Thank you in advance for your help. -- Piotr Piwko http://www.embedded-engineering.pl/

2 1

read/write to virtual address
by Daniel Krefft 17 Feb '17

17 Feb '17

Hello, while working on a checkpoint/restore mechanism for fiasco.oc, we want to trace read and write (kernel) calls to (virtual) memory. 1) Where are these calls defined? We're just able to find ARM specific calls inside the kernel, which allow writing and reading to/from a virtual address 2) Are these calls (if existent) platform agnostic (user-space)? Is there any kind of interface? Best regards -- Daniel Krefft MSc. Wissenschaftlicher Mitarbeiter Technische Universitaet Muenchen Fakultaet fuer Informatik Lehrstuhl fuer Betriebssysteme Arcisstrasse 21 80333 Muenchen Tel. +49 89 289 18791 daniel.krefft(a)tum.de www.os.in.tum.de/personen/krefft

3 7

About l4linux debug method
by li94575 17 Jan '17

17 Jan '17

Hi all, I have faced a thorny problem, l4linux died occasionally after running into the command line, and there are no errors displayed, pressing the ESC key also cannot enter fiasco jdb-mode. I don't know what caused the problem, can someone provide some methods to help locating this problem? Thanks in advance!

2 7

Memory usage analysis and optimization
by Jesse-SC Chou (周書正) 20 Dec '16

20 Dec '16

Hi L4-hackers, I’m analyzing the memory usage of Fiasco/L4Re/L4Linux. On my platform, there are 128MB physical memory. L4 bootstrap can detect it correctly. ================================================== L4 Bootstrapper Build: #164 拻 12堎 9 11:39:51 CST 2016, 4.8.4 Scanning up to 128 MB RAM, starting at offset 32MB Memory size is 128MB (20000000 - 27ffffff) RAM: 0000000020000000 - 0000000027ffffff: 131072kB Total RAM: 128MB ================================================== The memory regions after boot up and sigma0 loaded: ================================================== Regions of list 'regions' [ 20000000, 200001c7] { 1c8} Root mbi_rt [ 20001000, 20001aff] { b00} Kern fiasco [ 20002000, 20092fff] { 91000} Kern fiasco [ 200c0000, 200c9f43] { 9f44} Sigma0 sigma0 [ 200d0000, 200d6177] { 6178} Sigma0 sigma0 [ 20140000, 20168c27] { 28c28} Root moe [ 20170000, 2017b58b] { b58c} Root moe [ 21000000, 2100e5e3] { e5e4} Boot bootstrap [ 21100000, 21a2dfff] { 92e000} Root Module SIGMA0: Hello! KIP @ 20002000 allocated 4KB for maintenance structures SIGMA0: Dump of all resource maps RAM:------------------------ [4:20000000;20000fff] [0:20093000;200bffff] [0:200ca000;200cffff] [0:200d7000;2013ffff] [4:20140000;20168fff] [0:20169000;2016ffff] [4:20170000;2017bfff] [0:2017c000;210fffff] [4:21100000;21a2dfff] [0:21a2e000;26ffffff] ================================================== The largest unused continuous memory region are at “0x21a2e000 ~ 0x26ffffff”, which is about 85MB. I would like to make continuous memory region larger and have the following ideas and questions: 1. The memory used by bootstrap may be freed after the system bring up. If moving “bootstrap” program to 0x26000000, the “Module” programs can append to “MOE” program. The “Module” will be started “0x207c0000” (aligned to page size 4KB). Thus, I can get larger continuous memory area that originally separated by “bootstrap” program. Is this idea correct ? [ 20170000, 2017b58b] { b58c} Root moe [ 21000000, 2100e5e3] { e5e4} Boot bootstrap [ 21100000, 21a2dfff] { 92e000} Root Module 2. There are a reserved region “0x27000000 ~ 0x27FFFFFF” not shown in boot log. When I traced into program, this region seems used by sigma0. What is this memory used for? I think 16MB is a very large chunk in current embedded system. Can we compute the amount this area really use and shrink to a smaller size? Any suggestion is very appreciated. Thanks. BRs, Jesse

2 2