l4-hackers

l4-hackers@os.inf.tu-dresden.de

2827 discussions

Do L4 Linux lack source code about Freescale i.MX6 Quad ?
by Taeung 04 Aug '14

04 Aug '14

Hello , L4 hackers I always appreciate L4 hackers. I've succeed boot "helloworld + L4/Fiasco" on Freescale i.MX6q_sabre_sd. So, I've try to build L4 linux (|Repository : svn co http://svn.tudos.org/repos/oc/l4linux/trunk l4linux|). I've configured for build L4 Linux such as.. ====================================================== System type ---> ARM system type (Allow multiple platforms to be selected) Multiple platform selection ---> ARMv7 based platforms (Cortex-A, PJ4, Scorpion, Krait) Build type Freescale i.MX family Freescale i.MX support ---> i.MX6 Quad/DualLite support ====================================================== But I've meet errors even so often. such as .. 1. ERROR: CONFIG_ARM_PATCH_PHYS_VIRT must not be enabled. => So, I've force to change CONFIG_ARM_PATCH_PHYS_VIRT=y to =n . 2. irq.c:16:26: error: 'L4X_NR_IRQS_V_DYN' undeclared here irq.c:46:14: error: 'L4X_IRQS_V_DYN_BASE' undeclared and irq.c:48:14: error: 'NR_IRQS_HW' undeclared => So, I've included "#include <asm/generic/irq-vectors.h>" and "#include <asm/mach-arm/generic/mach/irqs.h>". 3. l4.c:57:13: error: 'L4X_IRQS_V_STATIC_BASE' undeclared ...etc. => So, I've included "#include <asm/generic/irq-vectors.h>" and "#include <asm/mach-arm/generic/mach/irqs.h>". 4. drivers/built-in.o: In function `clk_disable': clk.c:867: multiple definition of `clk_disable' arch/l4/built-in.o:/home/taeung/workspace/l4linux/arch/l4/kernel/arch-arm/l4.c:73: first defined here drivers/built-in.o: In function `clk_enable': clk.c:919: multiple definition of `clk_enable' arch/l4/built-in.o:/home/taeung/workspace/l4linux/arch/l4/kernel/arch-arm/l4.c:65: first defined here make[1]: *** [vmlinux] Error 1 make: *** [sub-make] Error 2 => So, I delete this lines in l4linux/drivers/clk/clk.c. 7. undefined reference to `__pv_phys_offset' from signal.c , init.c, mm.h ....etc. My question is.... Is it's problem that my config is wrong or l4 linux lack source code about Freescale i.MX6 Quad ? * Refer to > /home/taeung/workspace/l4linux/arch/l4/kernel/*irq.c* /home/taeung/workspace/l4linux/arch/l4/include/asm/generic/*irq-vectors.h* /home/taeung/workspace/l4linux/arch/l4/kernel/*pm.c */home/taeung/workspace/l4linux/arch/l4/kernel/arch-arm/*l4.c** */home/taeung/workspace/l4linux/arch/l4/l4lxlib/generic/*irq_gen.c* /home/taeung/workspace/l4linux/drivers/clk/*clk.c */home/taeung/workspace/l4linux/arch/l4/kernel/arch-arm/*signal.c */home/taeung/workspace/l4linux/arch/l4/mm/arch-arm/*init.c */home/taeung/workspace/l4linux/include/linux/*mm.h *

3 2

seL4 boot issue
by Jorge Ventura 04 Aug '14

04 Aug '14

I am trying to boot seL4 on beaglebone white. I have an initial code just to configure the DDR2 memory at range 0x80000000-0xbfffffff. As we can see below, the start address is the virtual address 0xf0000000 but at this point I have no MMU configured. I was thinking that the segment .boot should be the code to prepare all details about MMU/D-CACHE/I-CACHE but it's not what gdb is understanding based in the elf header. Someone could explain to me how to boot this kernel; what do I have to consider. I guess that beaglebone is not supported but this is exactly my goal, to make seL4 boot on beaglebone. Any help or suggestion is very welcome. Sincerely, Jorge Ventura (gdb) load kernel.elf > Loading section .boot, size 0x10000 lma 0x80000000 > Loading section .text, size 0xf1ec lma 0x80010000 > Loading section .text.unlikely, size 0xd98 lma 0x8001f1ec > Loading section .rodata, size 0x4170 lma 0x8001ff84 > Loading section .data, size 0x8 lma 0x800240f4 > Start address 0xf0000000, load size 147708 > Transfer rate: 319 KB/sec, 3787 bytes/write. > (gdb) monitor poll > background polling: on > TAP: am335x.dap (enabled) > target state: halted > target halted in ARM state due to debug-request, current mode: System > cpsr: 0x6000019f pc: 0xf0000000 > MMU: disabled, D-Cache: disabled, I-Cache: disabled >

1 0

Task creation in ned or moe
by Stark, Josef 31 Jul '14

31 Jul '14

Hello, could maybe someone kindly explain to me (roughly), how task creation/startup is done in ned or moe? I assumed that the Factory methods create_task() and create_thread() would be used, but instead it seems that Elf_Loader is used for this. I skimmed through this class but I don't fully understand it. Basically, it seems that moe's call "elf_loader.start(_init_prog, cxx::String(""));" (in file main.cc) leads in the end to this call: "Ldr::Elf_loader<Moe_x_app_model, Dbg>::launch(&am, "rom/l4re", ldr);" (loader_elf.cc). So, to me it seems that just l4re gets started, not the desired binary (ned), but clearly I am missing something (a lot) here (because it actually works the way it should). Could someone enlighten me here? And why aren't create_task() and create_thread() used here? (In this text I am referring to moe, but in ned I couldn't find these two methods either.) Thanks in advance! Josef

2 1

LUA bindings / kill task
by Stark, Josef 31 Jul '14

31 Jul '14

Dear all, is there maybe a list somewhere containing all the LUA bindings that are exported by L4(Re)? I found a few functions in the file src/l4/pkg/lua/lib/l4/ll4lib.c ("ipc_call", "ipc_reply_and_wait" etc.) and also in src/l4/pkg/ned/server/src/ned.lua (mainly "start" and "startv") but specifically I am looking for a command to kill a task. It seems that the file src/l4/pkg/ned/server/src/ned.lua prepares to offer such a function (in the array "_task_ops[]" the function "__task_kill" corresponds to the name "kill"), however I didn't succeed in calling this kill function. What could I be doing wrong? If it is not yet implemented, what needs to be changed? I tried it in interactive ned mode as well as in the mode which interprets a config file (using the latest snapshot). Thanks in advance! Cheers, Josef Stark

2 2

Problems with python to compile seL4
by Jorge Ventura 29 Jul '14

29 Jul '14

I am still using Debian 6. I installed python 2.7.8 but I am having problems to compile the seL4. arduino@squeeze64:~/ARM/src/seL4$ TOOLPREFIX=arm-none-eabi- ARCH=arm PLAT=am335x ARMV=armv7-a CPU=cortex-a8 make Traceback (most recent call last): File "./tools/invocation_header_gen.py", line 18, in <module> import tempita ImportError: No module named tempita make: *** [arch/api/invocation.h] Error 1 What is this module "tempita"? What package is that?

3 3

seL4 is free!
by Gernot Heiser 29 Jul '14

29 Jul '14

A few hours ago we open-sourced seL4. The release covers the kernel source and proofs (under GPL v2) as well as tools and libraries (mostly BSD). http://sel4.systems contains details, including download and mailing list links. Enjoy, Gernot

1 0

Question about IPI
by Yuxin Ren 28 Jul '14

28 Jul '14

Hi All, Now I want to how many IPIs are sent during an IPC. So I add a counter for IPI in the kernel, and in the user level, there is a client sending IPC to a server on another core. But I found some behavior are very strange. 1. The total number of IPI are not multiple times of the number of IPC. It seems that there are some "extra IPIs", and with the number of IPC increases, the number of "extra IPIs" are also increasing. So my question is if there are some background program in the kernel or L4Re to send IPIs? 2. If I add printf statement in the kernel, the number of IPIs are also increasing. Does printf in the kernel cause sending IPI? If it does, is there any methods to avoid this. Like in Linux, it has printk function in the kernel. Thanks a lot!! Yuxin

2 5

Start a process with a specified pager
by Noah Zentzis 24 Jul '14

24 Jul '14

Hi, I'm trying to launch a new task and run it under a custom pager, but I can't seem to figure out how to go about it. I've looked through the Moe and libloader source code, but I'm not sure how the task actually ends up getting created, only where the ELF image is loaded into memory. Can anyone point me towards the documentation for task creation or explain how Moe actually ends up turning the executable image into a running task? Thanks! - Noah Zentzis

2 1

Question about irq
by Maria Soler 24 Jul '14

24 Jul '14

Hello, I have a question about how the irqs work in L4re. I am running FIASCO.OC + L4re on qemu for arm. I want to be able to raise an interruption from qemu and handle it from my VM. I am basing my work on the example at examples/libs/libirq (I am coding in C, so it is more convenient than other examples in C++). I have a qemu device that has the interrupt number #N in qemu. I rise the interrupt but the l4irq_wait function doesn't return. I have tried both with qemu_irq_pulse(state->irq); and with qemu_set_irq(state->irq, 1), also I have tried attaching the irq and requesting it; but there is no difference. My basic doubts are: - Is the number of the interruption the same within L4re as in qemu? and if not, how can I calculate/check what the number is in L4re? - If the number is OK, what can be the problem I am facing? The mentioned example doesn't work for me either (l4irq_request returns 0 in this case), so I don't know how to solve this exactly. - Is there a better example for this that I can use as a baseline? In case this is relevant, this is part of a more complex project, so I have a .devs file where I have a line like: Io.Res.irq(#N); I am also using this file in the present step to keep things as similar as possible to the real scenario. Thanks! Regards, María.

2 1

Why need a lock in memory/capability map
by Yuxin Ren 24 Jul '14

24 Jul '14

Hi All, I note that when a thread does memory or capability mapping, both over system call and IPC, it needs take a lock, which is the existence_lock of target task. I cannot understand it needs take a per task, as cap/page table are per thread. What data does this lock to protect? Thank you. Yuixn

2 3

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

l4-hackers