l4-hackers

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

2827 discussions

Running Fiasco.OC in Bochs
by Norman Feske 28 Apr '12

28 Apr '12

Hello, for tracing/profiling purposes, I'd like to run Genode on Fiasco.OC in Bochs. My prior attempts with the OKL4 and Pistachio kernels went fine. However, when booting the setup with Fiasco.OC, I get the following kernel message right at the start: Freeing init code/data: 28672 bytes (7 pages) DOUBLE FAULT! EAX=ec7ff000 ESI=00000000 DS=0023 EBX=fcdcf000 EDI=fcdcf170 ES=0023 ECX=ec7ff000 EBP=f0060100 GS=0043 EDX=ec7ff000 ESP=fcdcffbc SS=0010 ESP0=00000000 EFL=00200246 EIP=f006592c CS=0008 Return reboots, "k" tries to enter the L4 kernel debugger... Has anyone successfully used Fiasco.OC with Bochs before? If yes, I would very much appreciate a known-good starting point for experimentation. For reference, I am using the following kernel configuration: https://raw.github.com/genodelabs/genode/master/base-foc/config/x86_32.kern… My 'bochsrc' file looks as follows: display_library: x romimage: file=/usr/local/share/bochs/BIOS-bochs-latest megs: 128 com1: enabled=1, mode=file, dev=serial.out boot: cdrom ata0-master: type=cdrom, path=var/run/demo.iso, status=inserted mouse: enabled=1, type=imps2 log: ./bochs.out panic: action=fatal vgaromimage: file=/usr/local/share/bochs/VGABIOS-lgpl-latest vga: extension=vbe pci: enabled=1, chipset=i440fx, slot1=pcivga The Bochs version is the official release 2.5.1 configured with the following arguments: --with-x11 --enable-plugins --enable-pci --enable-clgd54xx Regards Norman -- Dr.-Ing. Norman Feske Genode Labs http://www.genode-labs.com · http://genode.org Genode Labs GmbH · Amtsgericht Dresden · HRB 28424 · Sitz Dresden Geschäftsführer: Dr.-Ing. Norman Feske, Christian Helmuth

1 0

Performance of l4_task_unmap call
by Chen Tian 26 Apr '12

26 Apr '12

Hi, I did a small test: a parent task first maps and then unmaps a page to its child. Both parent and child task are running on the same core. I found the execution is very slow and 99% time are spent on l4_task_unmap( L4_FP_OTHER_SPACES flag is used). After some investigation, I found the flag "need_xcpu_tlb_flush" in map_util.cpp:766 is set to true. My guess is that this flag controls the tlb of the cpu at which the child task is running. However, in my case, both tasks are running on the same core, setting this flag can have huge negative impact on performance. I am thinking to put a check before this statement: if (current_cpu() != space's CPU) set the flag to true. Is this correct? If so, how to get space's cpu? space->cpu() ? Thanks for the help. Chen

2 1

Bootfs got overwritten
by Haohui Mai 26 Apr '12

26 Apr '12

Hi, I'm encountering a problem where the memory region of BOOTFS of L4Re got overwritten. Interestingly, the page is overwritten by things in the bootfs. Here are the physical memory mapping set up by BOOTFS: ... BOOTFS: [2227000-23297b8] [C:506000] rtc BOOTFS: [232a000-26d5411] [C:507000] mag BOOTFS: [26d6000-2848a94] [C:508000] fb-drv BOOTFS: [129d000-129daa4] [C:509000] vandroid-x86.cfg BOOTFS: [126b000-126b3b7] [C:50a000] x86-legacy.devs … The page 0x2239000 (which belongs to rtc) is magically overwritten by the first page of x86-legacy.devs. I confirmed it with both JDB and QEMU monitor. I'm running the newest snapshot of L4Re / L4Android on top of the UP version L4.Fiasco kernel. What would be the best approach to debug the problem? is it there a way to mark the whole BOOTFS as read-only so that I can figure out what's going on? Thanks! ~Haohui

3 6

Re: Fiasco.OC development status
by Robert Campbell 24 Apr '12

24 Apr '12

I'm interested in userland changes and possible ports of libraries and frameworks. I'll definitely be focusing on the Fiasco.OC kernel. However, I'm at a crossroads with choosing the userland i.e whether to go with L4Re or Genode. IIRC, the port of Qt to L4Re or L4Env was version 3.x. Genode has a port of Qt 4.7.x. Genode has taken their development public So I am able to watch their activity. However, with L4Re I'm not sure what to expect in new releases bar what the publications might imply. If you are able to provide information on possible ports and update to existing ports then that would aid my decision and I would appreciate it. P.S

1 0

L4Android and L4Linux parallel on an i.MX53
by Tobias Fink 22 Apr '12

22 Apr '12

Hello List, for my bachelor project I'm trying to run Android and Linux parallel on an i.MX53 Board. While the first approach with KVM seemed not to be the right way to go L4 looks promising. While I found many interesting and helpful topics in this list I still have some questions. I tried to first build the "Hello World" example which worked fine but failed in building other examples because of not fulfilled dependencies (first to note io). I already checked out "l4linux_requirements". What else do I need? The board is a custom board with very little documentation. I have some Linux drivers for the connected hardware. Is it possible to use them with L4Linux or would I have to get/write drivers for L4 itself? Sorry about these very basic questions. I've just started working with L4. Regards Tobi

4 9

Windows in QEMU on KVM-L4
by Rolf Sommerhalder 22 Apr '12

22 Apr '12

Hi, In their paper [1], the authors mention that they were able to run 32-bit versions of Windows XP and 7, among other guest OSes, in KVM-L4 on AMD x86 processors with AMD-V extensions. Does anyone from the list happen to have an .iso around which boots on such an AMD machine and demonstrates the performance of Windows (any version will do) running in KVM-L4? Alternatively, I would be grateful for any recipe how to adapt a demo CD or Fisco.OC/L4Re build environment so that it allows to start a Windows image in QEMU/KVM on L4. So far, I have managed to install and to run Windows 7 Pro in a QEMU/KVM image on my Linux host. As this image is over 4 GByte by itself, plus QEMU with its dependencies, there is no way of just extending the ramdisk image in order to load and run it in L4Linux. After searching this list, I so far found two possible ways to proceed: a) Extended the ramdisk so that it mounts a disk partition /dev/hdaX, and then chroot to that mounted partition. Then try to launch the QEMU image from that new root. However, a previous poster here ran into a problem while trying this, and it is unclear to me if was able to overcome it [2]. b) Abandon the ramdisk and have L4Linux mount a disk partition as its root file system directly upon boot. This was tried in 2006 and it is unclear if it finally worked [3]. In both cases, I presume that the binaries in that partition (QEMU and its dependencies) must match the version of the L4Linux kernel (which is 3.0). My Linux build environment uses Ubuntu 10.04 LTS (which uses kernel 2.6.x). My only interest at this early stage of a quick Proof of Concept is just to get a feeling if a faithfully virtualised Windows run on top of KVM-L4 still performs reasonably well on a recent (Business-) PC, so that it is still acceptable to the average user. Previous results published suggested that real-world Linux applications incur (only) a 4..5% overhead when run on top of L4Linux, compared to running them natively on (bare metal) Linux. Now I would like to know if this surprisingly low overhead also continues throughout KVM-L4 with Windows as one of its "applications". Thanks in advance for any hints, Rolf [1] Virtual Machines Jailed - Virtualization in Systems with Small Trusted Computing Bases Michael Peter, Henning Schild, Adam Lackorzynski, Alexander Warg VTDS'09: Workshop on Virtualization Technology for Dependable Systems (Eurosys 2009 affiliated workshop) , March 2009 http://os.inf.tu-dresden.de/papers_ps/peterschild09_vtds_virtual_machines_j… [2] Using hard disk as root filesystem http://www.mail-archive.com/l4-hackers@os.inf.tu-dresden.de/msg00851.html [3] Booting L4Linux/Fiasco with root=/dev/hda http://os.inf.tu-dresden.de/pipermail/l4-hackers/2006/002396.html

2 1

l4_task_cap_equal broken?
by Stefan kalkowski 20 Apr '12

20 Apr '12

Dear Fiasco.OC developers, while working with Genode on top of Fiasco.OC, I've encountered the following problem. When I transfer ipc-gates via ipc between tasks, it comes to the point where a task receives a gate it already owns. These duplicates are detected due to an additional identifier that is transfered together with the gate. To detect fraud (whether somebody has send an invalid gate/id combination) the received gate gets compared with the locally available one via "l4_task_cap_equal". Now, the problem is "l4_task_cap_equal" often fails even when both capabilities point to the same ipc-gate. The following code-snippet demonstrates this issue: ... tag = l4_task_cap_equal(L4_BASE_TASK_CAP, i->kcap(), rcv_cap); if (!l4_msgtag_label(tag)) { unsigned long id1 = l4_debugger_global_id(i->kcap()), id2 = l4_debugger_global_id(rcv_cap); printf("my_cap=%lx (id=%lx) rcv_cap=%lx (id=%lx)", i->kcap(), id1, rcv_cap, id2); if (id1 == id2) enter_kdebug("FAIL"); } ... This snippet is executed after a thread in the roottask received from another thread of the same task a capability via IPC. Thereby rcv_cap is the cap that was received via ipc and i->kcap() is the cap that was retrieved via the additional transfered identifier. When executed I get the following output: ... my_cap=20f000 (id=12) rcv_cap=205000 (id=12) --------------------------------------------------------------------- CPU 0 [00145e81]: FAIL [ entrypoint] jdb: The debugger's global id (here: 12) indicates that both capabilities reference the same object. Nevertheless, the equality test fails. Can you confirm that this is undesired syscall's behaviour, or do I miss something important? The attached kernel patch fixes the problem for me, but may introduce other problems (e.g.: not sure whether obj() returns always valid pointers). Regards Stefan

2 2

Fiasco.OC development status
by QiuYi 19 Apr '12

19 Apr '12

> On 10.04.2012 07:20, Robert Campbell wrote: >> Thank you Bjoern. >> Any idea when new release will be made and what changes and improvements to >> expect? > Unfortunately, there is no more specific answer than: There will be a > new release soon (maybe this summer). No major new features, but all > components will have undergone further developments & fixes. > Are there any features you are interested in specifically and that you > are missing from or cannot find in the current release? > Bjoern I am interested in the drivers of touchscreen, keypad and NANDflash on the beagleboard at the L4RE and L4linux. qiuy.

1 0

FPU state management in L4Fiasco.OC program
by Mai, Haohui 18 Apr '12

18 Apr '12

Hi, I'm writing a new OS kernel atop of L4Fiasco.OC. I'm having some difficulties to let my OS support user-level processes that uses floating-point instruction. So far, My OS closely follows L4Linux, it creates a new task for each user-level process, and marks as all threads of the task as alien thread to capture all page faults and exceptions. My OS works fines with page fault and syscalls. However, I'm confused with what I should do with floating-point exception (exception #7), which is issued to my kernel at the first time it tries to perform floating-point arithmetic. I tried to return to the user-level process with an empty message ( l4_msgtag_t tag = l4_msgtag(0, 0, 0, 0) ), or an empty message with L4_MSGTAG_TRANSFER_FPU (l4_msgtag_t tag = l4_msgtag(0, 0, 0, L4_MSGTAG_TRANSFER_FPU) ), but I had no luck. The user-level process immediately calls "int $0x3" and stops. I read through L4Linux's code and it seems it's doing similar thing, except that it did initialize the FPU state before asking L4 to transfer the FPU state to user-level process. Am I missing something? What's the recommend way of doing it? Your comments are highly appreciated. ~Haohui

2 8

shmc/l4shmnet improvements
by Stefan Fritsch 17 Apr '12

17 Apr '12

Hi, the attached shmc.diff fixes many bugs in shmc and adds some new functionality. l4shmnet.diff then uses this to implement a protocol that does not require one end to wait for the other end during boot, and that allows to set the link state up and down. The old protocol is still supported. The changes may be used under the MIT license (which is included in the include/shmbuf.h file. You can find the detailed changelog for shmc.diff below. The diff is on top of the diff I have sent in November. Cheers, Stefan Add include/shmbuf.h which implements the l4shmnet packet protocol, to be used by l4linux and others. include/shmc.h: - l4shmc_iterate_chunk(): New function to iterate over all chunk names - l4shmc_chunk_size(): change return value to allow returning an error - l4shmc_area_size_free(): New function to get remaining free space - l4shmc_area_overhead(): New function to get overhead not usable by chunks - l4shmc_chunk_overhead(): New function to get size overhead per chunk include/types.h: - Make l4shmc_chunk_desc_t members volatile to avoid races - l4shmc_area_t: add _size member for fast access - l4shmc_chunk_t: add _capacity member that stores the size outside of the shm area and is therefore safe from tampering by remote end shmc/include/internal.h: - l4shmc_chunk_ready(): need HW memory barrier - l4shmc_chunk_size(): Check for overflow - l4shmc_chunk_capacity() adapt to new _capacity member - l4shmc_chunk_consumed(): don't need a memory barrier lib/src/shmc.c: - l4shmc_create(): protect from integer overflows, initialize lock - l4shmc_area_overhead(): new function - l4shmc_chunk_overhead(): new function - next_chunk(): utility function that gets the next chunk, with proper sanity checks to avoid invalid memory access/loops if the other side is malicious - l4chmc_iterate_chunk(): new function - l4shmc_add_chunk(): Add sanity checks, use next_chunk(), initialize complete chunk struct properly before inserting into linked list - l4shmc_area_size_free(): new function - l4shmc_get_chunk_to(): Add sanity checks, use next_chunk()

2 4

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