- l4-hackers - example.com Site

l4linux GPU direct access
by Abhishek Gupta 07 Jun '12

07 Jun '12

Hi, I was trying to run xserver on l4linux using vesa frame buffer (l4 is running on kvm) and virtual machine hangs and it seems that a infinite loop is started which is continuously try to do something but doesn't succeed. When I tried to debug it using gdb, it displays 0xb6ee9b28 in SetResetBIOSVars (pInt=0x820f5a0, set=<value optimized out>) at ../../../../hw/xfree86/int10/helper_exec.c:686 this code is from the xserver 1.7.7 static void SetResetBIOSVars(xf86Int10InfoPtr pInt, Bool set) { int pagesize = getpagesize(); unsigned char* base = xf86MapVidMem(pInt->scrnIndex, VIDMEM_MMIO, 0, pagesize); int i; if (set) { for (i = BIOS_SCRATCH_OFF; i < BIOS_SCRATCH_END; i++) MEM_WW(pInt, i, *(base + i)); } else { for (i = BIOS_SCRATCH_OFF; i < BIOS_SCRATCH_END; i++) /*line 685*/ *(base + i) = MEM_RW(pInt, i); /*line 686*/ //This line of code is not working } xf86UnMapVidMem(pInt->scrnIndex,base,pagesize); } The value of base becomes 0 after xf86MapVidMem func is called. So my Guess is that this func is failing and returning NULL but I am not able to figure out why Also I read in previous years mailing list that Linux tries to execute int10 to set the VESA mode but L4 applications are not allowed to execute real-mode code. Does this have anything to do with my problem?? How can I fix this problem? Also, Is there a way to allow L4 applications to execute code in real mode and then switch back? May be add some code to L4 to allow applications to execute code in real mode and then switch back. Abhishek Gupta Indian School of Mines

2 1

Fiasco.OC: sigma0 stucks in ipc-path
by Stefan kalkowski 07 Jun '12

07 Jun '12

Dear kernel-hackers, I think I came along a problem in the IPC-path of Fiasco.OC sometimes in the past, but it was somehow hard to reproduce. But now I've an example that quite reliable triggers the issue. The symptom in the past, and in the concrete example was that our roottask in Genode (called core) was requesting memory from sigma0, either implicitly just by touching some memory-area, like a ROM-module loaded by the bootloader, or explicitly by using the sigma0-protocol, e.g. to request I/O memory for the framebuffer. After that the request was received by sigma0, and also processed, but the answer never reached the faulter/client in this case the core-pager thread. The typical picture in the kernel-debugger then looks like the following: http://pastebin.com/cFvn1NV2 sigma0 is marked as ready, but it still stucks in the ipc-syscall. By instrumenting the kernel I could break down the point where sigma0 never returns. It is when trying to establishing the actual mapping in 'Thread::transfer_msg_items' in file src/kern/thread-ipc.cc. The spot is the following: ``` cpu_lock.clear(); L4_error err = fpage_map(snd->space(), sfp, rcv->space(), L4_fpage(buf->d), item->b, &rl); cpu_lock.lock(); ``` After the cpu lock is given away, sigma0 never aquires it back successfully. It looks to me like a race-condition, nevertheless simply letting the lock being locked doesn't solved the problem ;-). Maybe given your insight knowledge, you're much faster in tracking the problem down to its root? I've to add, that the problem occurred on x86 as well as ARM, on different QEMU versions, as well as real hardware. I also could reproduce it with slightly older versions of Fiasco.OC, and in different development stages of Genode. Nevertheless, it was never that reliable to reproduce. The current example which reproduces it reliable at least with my QEMU (qemu-kvm-0.14.0) version. You can find in form of an ISO-image here: http://dl.dropbox.com/u/82567292/avplay.iso You can try the iso-image like the following: qemu -no-kvm -m 256 -soundhw all -serial mon:stdio -cdrom avplay.iso Or, if you've to build it on your own, here is my topic branch, including the avplay run-script that triggers the problem: https://github.com/skalk/genode/tree/fiasco.oc-ipc-issue To compile and run it yourself you've to do the following steps (after installing the genode toolchain from http://genode.org/download/tool-chain) git clone git@github.com:skalk/genode.git cd genode git checkout -b issue origin/fiasco.oc-ipc-issue git clone git@github.com:genodelabs/linux_drivers.git make -C base-foc prepare make -C libports prepare PKG="libav libc sdl zlib" tool/create_builddir foc_x86_32 BUILD_DIR=build sed -i "/#REPOSITORIES.*libports/s/#//" build/etc/build.conf sed -i "/#REPOSITORIES.*linux_drivers/s/#//" build/etc/build.conf make -C build run/avplay Thank you in advance & best regards! Stefan

4 7

[l4ankh] Problems with fixed l4ankh
by Timur 03 Jun '12

03 Jun '12

Hi! I had fixed the l4ankh driver and after that it started to work properly. Now only thread is used to receive packets. In lines 158-162 we clear the rcv buffer: that is necessary to be done as after opening a session we start to receive packets that should not be served; otherwise the response time will be quite slow. This is not the best solution, but I do not have time to make it better. Now I face the following problem: the driver in ankh (I use 8139cp) stops receiving packets after some time (for me it is usually 3-4 minutes). This situation occurs when the interrupt handler was called and the device register IntrStatus already equals "zero". An interrupt doesn't work after that. -- Galiullin Timur

3 5

I have a question about L4Linux
by Ednilson Ricardo 03 Jun '12

03 Jun '12

Hello, I have some doubts about the L4Linux, can you help me? *** What are the purposes of L4Linux?Hardware subjaecnte that OS (capacity and types of memory, processor, bus, etc.).*** With this operating system can work? What capacity and type of hardware it runs? What is the architecture that it needs to function? the type of memory itworks? processors that it works, buses, and everything involved in the operating system. What it needs hardware to run? What type of application it serves?Interaticas, batch, real-time, etc..*** What are the characteristics related to the accomplishment of the application,low response time, low turnaround time, etc.. (View aqielas applications, which it meets the requirements) as it meets these requirements? -> It is otimisado,blibioterca specific uses, he uses some dynamic code that allows it meets these requirements, then how it will work.*** What is your architecture? (Monolithic? Layered? Micro-kernel virtual machineor some other system) PREFERENCE TO EXPLAIN AND THE GREATERLEVEL POSSIBLE to detail.*** If the operating system is not supposed or threads (the core of User, hybrid?)And how the theads are implemented in the OS (if it meets it has a mechanism forimplementation (How it was implemented)

2 4

Debugging l4linux + l4re
by Haohui Mai 03 Jun '12

03 Jun '12

Hi folks, I'm running mag + L4Linux on a real machine that doesn't have a serial port. It's really difficult to debug -- I'm wondering if there're any ways to open a new window under mag, and to redirect all serial port I/O to that window? It would be really helpful for debugging. I appreciate if anyone could share if he happens wrote this program before, or giving me some suggestions. Your help is appreciated! ~Haohui

2 1

Ankh network
by Johan Dams 31 May '12

31 May '12

Hi. I've noticed that when sending more than one pbuf in low_level_output() that each pbuf gets sent individually over the network and is being treated as individual packets. I 'fixed' this problem by copying the different pbufs into one buffer and sending this buffer in one go (basically as a test to verify the issue). I guess the problem has to do with how l4ankh_send() calls l4shmc_rb_sender_commit_packet() after each pbuf instead of waiting until all pbufs of the packet in question are passed on with l4shmc_rb_sender_next_copy_in(). Best regards, Johan

1 0

PowerPC support
by G S Madhusudan 31 May '12

31 May '12

This is in reply to Oliver Horst's posting on PPC support. I am running a research program to develop variants of L4 for use in India at IIT-Madras. http://www.iitm.ac.in/ Our target platforms are Freescale QorIQ processors (P2041, P4080 and T4240 - with hypervisor support) and Cortex A9 processors. We will initially use std dev boards but we are also developing our own custom HW. As someone pointed out it helps if someone has knowledge of PPC and kernels. I have been hacking open source kernels since 1985 ! (BSD 4.x) and have been working on PPC development since 1998, so while I do not underestimate the task it can be done. We also have access to Freescale's processor architecture team so that helps a bit. The project goal is actually ambitious since we want to replace Linux in security sensitive installations. I lead a Mach 3.0 based effort in the 90s along with IBM research and I am all too painfully aware of how uKernel based projects can fail ! I am also coordinating a processor development effort and Fiasco+ Genode combination will also be ported on to that processor. But it is a specialized custom processor so I suspect no one else other than my team be interested in that port. But one aspect of teh processor is aimed at improving the MMU so that the cost of context switches can be improved. If you cannot fix the OS, just fix the CPU ! Any thoughts on this aspect are welcome. On top of Fiasco we are developing UTM + router type appliances. Later we plan to port the Drizzle RDBMS. We would be glad to work with anyone with similar goals. Currently it is a 4 member team along with 7 MTech students doing projects on various aspects of this research. We also have access to sufficient P2041 and p4080 boards. Once our boards are ready (the 2041 boards i being PCB routed now), we plan to create an engineering curriculum using this HW and focusing on secure systems and uKernel OSs. the plan is to offer it to Indian engineering schools but anyone can use it. Regards G S Madhusudan

1 0

Using pthread on L4/Fiasco
by 석현철 23 May '12

23 May '12

Thank you. I made a stupid mistake. When I added “REQUIRES_LIBS = libpthread” into Makefile, I modified three Makefiles. There are hello/Makefile, hello/server/Makefile, and hello/server/src/Makefile. I tested one by one. But I found I made a mistake. At last try, I wrote REQUIRES = libpthread instead of REQUIRES_LIBS = libpthread on hello/server/src/Makefile. After correcting my mistake, I can test pthread application. Thank you! Marcus Hyunchul Seok.

1 0

Using pthread on L4/Fiasco
by 석현철 23 May '12

23 May '12

Hi. Hackers I’m trying to use pthread in an application which runs on L4. I modified “main.c” in Hello application, including “#include <pthread.h>”. And when I compiled L4Re, I got error messages: main.o: In function `main': /root/l4re-core-2011081207/src/l4/pkg/hello/server/src/main.c:39: undefined reference to `pthread_create' /root/l4re-core-2011081207/src/l4/pkg/hello/server/src/main.c:44: undefined reference to `pthread_join' make[5]: *** [hello] Error 1 I think this is because it cannot fine pthread.h header file. But I don’t know how to fix it. When I looked up previous lists, I could find one thread related to pthread in 2011 list. (the thread title is “Silly Pthread Issues”) I think that it is the same problem, but the reply is only one. Said “Add REQUIRES_LIBS = libpthread to the Makefile.” Actually, I tried it, but I couldn’t solve the problem. How can I use the pthread? Please, Help me!! Thank you! Hyunchul Seok The modified code is below: #include <stdio.h> #include <unistd.h> //#include <pthread-l4.h> #include <pthread.h> void *process_test(void *arg) { int i = (int)arg; int j; for (;;) { for (j = 0; j < i; j++) { printf(" "); } printf("%d thread\n", i); sleep(i); } } int main(void) { pthread_t threads[10]; int status[10]; int i; for (i = 0; i < 3; i++) { pthread_create(&threads[i], NULL, process_test, (void *)i); puts("New thread creation!"); } for (i = 0; i < 3; i++) { pthread_join(threads[i], (void **)&status[i]); } return 0; }

2 1

[Announcement] Karma VMM first public release
by Matthias Lange 18 May '12

18 May '12

We are happy to announce the first public release of the Karma Virtual Machine Monitor [0]. Karma is a virtual machine monitor (VMM) that runs Linux in a virtual machine on top of the Fiasco.OC [1] microkernel. Its main design directives are speed and simplicity. Unlike KVM, VirtualBox or VMWare, Karma does no emulation, but resorts to aggressive paravirtualization and therefore has low complexity and shows superior performance for a number of benchmarks. Karma requires hardware assisted CPU virtualization. A feature that is available on all recent x86 CPUs (SVM or VT). Optionally, it can make use of hardware assisted memory virtualization (Nested Paging or ePT), if available. The Karma VMM is currently being used as a research vehicle at Technische Universität Berlin and Technische Universität Dresden. It main features are: - Trustworthy isolation of virtual machines by means of the microkernel. VMM compromise does no harm to other virtual machines as one instance of Karma runs exactly one virtual machine. - Clean, tiny code base, less than 9000 source lines of code. - Near native (guest)-Linux performance. - Supports: * SMP * Networking * Direct harddisk access via AHCI * RTC and Hpet support Karma is released under the terms of the Gnu General Public License Version 2. [0] http://karma-vmm.org [1] http://os.int.tu-dresden.de/fiasco [2] http://www.isti.tu-berlin.de/security_in_telecommunications [3] http://os.inf.tu-dresden.de -- Dipl.-Inf. Matthias Lange <mlange(a)sec.t-labs.tu-berlin.de> Security in Telecommunications TU Berlin / Telekom Innovation Laboratories Ernst-Reuter-Platz 7, 10587 Berlin Phone: +49 - 30 - 8353 58 553 Mobile: +49 - 160 - 587 28 07 Web: http://www.t-labs.tu-berlin.de/sect

5 5