On Wed, Oct 26, 2011 at 6:25 PM, HUNG BUI <buihung@xxxxxxxxx> wrote: > Hi everybody, > > I am trying to patch and compile realtime patch for kernel 3.0.0-1205 > (TI-OMAP4) on Pandaboard. However I faced lots of error, and couldn't > finish the compiling process yet. > > I followed the instruction of compiling Pandaboard kernel from here > > http://adventuresinsilicon.blogspot.com/2011/02/pandaboard-ubuntu-how-to-recompile.html > http://www.omappedia.com/wiki/Ubuntu_kernel_for_OMAP4 > > My kernel was downloaded with the command : > > sudo apt-get build-dep linux-image-$(uname -r) > > which is arm-linux-3.0.0-1205-omap4 or linux-ti-omap4 > tar file : linux-ti-omap4_3.0.0-1205.10.tar.gz > The patch I used is : patch-3.0-rt6.patch.bz2 > > Pandaboard running on Ubuntu Oneiric Desktop. GCC 4.6 > > I tried the following commands : > > patch -p1 < ../patch-3.0-rt6.patch > fakeroot debian/rules editconfigs > (# Kernel Features -> Preemption Model -> Fully Preemptible Kernel (RT) ) > fakeroot debian/rules clean > dpkg-buildpackage -B -uc -us > > First the patch showed that there are some : > > Reversed (or previously applied) patch detected! Assume - R Apply > anyway n] with 2 files : > drivers/rtc/interface.c > drivers/rtc/rtc-tegra.c > > (Don't have this problem if i apply rt8 or latter However they are for > kernel 3.0.1 which I don;t know how to get) > > I answer No for both assume -r and Apply anyway , I am not sure it's > correct or not > > Also the patch showed that there are some hunk failed with 3 files : > > arch/arm/mm/cache-l2x0.c > init/main.c > kernel/stop_machine.c > I tried to fix this manually with the .rej files (otherwise it will > cause this error when compile: > > kernel/stop_machine.c:176:2: note: each undeclared identifier is > reported only once for each function it appears in > kernel/stop_machine.c: In function ‘cpu_stopper_thread’: > kernel/stop_machine.c:290:3: error: ‘stopper_lock’ undeclared (first > use in this function) > > > Then when I compile the kernel, i got some more errors which i can > search and find the way to fix them however when it come to the > following error, I couldn't find the solution for it : > > drivers/dsp/syslink/multicore_ipc/transportshm.c:261:2: error: request > for member ‘lock’ in something not a structure or union > drivers/dsp/syslink/multicore_ipc/transportshm.c:261:2: warning: > dereferencing ‘void *’ pointer [enabled by default] > drivers/dsp/syslink/multicore_ipc/transportshm.c:261:2: error: request > for member ‘lock’ in something not a structure or union > make[2]: *** [drivers/dsp/syslink/multicore_ipc/transportshm.o] Error 1 > > I am not sure if the patch (3.0 rt6), kernel(linux-ti-omap4 3.0.0) , > gcc (4.6), and the commands i am using is correct, or there is some > mismatches between them that make more and more errors happened when > compiling. > > Thanks in advance, > > Hung Bui > -- > To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html > Hi Hung, I recommend building the RT patch against the mainline Linux kernel, as it would apply clean. If you need to use OMAP kernel git tree, then you may have to resolve the hunks by yourself. I am not sure how far off it is from the mainline and how much work applying the patch would be. I can give you the instructions that I follow to build the RT-patched kernel for panda board. Hope it is useful to you. (1) Use a cross-compiler for ARM on x86 development Linux machine (or a Linux VM). I found code-sourcery toolchain to be simple and easy to install and use on my x86 Linux development machine. You can get that from https://sourcery.mentor.com/sgpp/lite/arm/portal/subscription3053 (I think I use "Sourcery G++ Lite 2009q1-161" and use the binary installer "IA32 GNU/Linux Installer"). Add the code-sourcery tool-chain to the PATH. (Or, you can use OpenEmbedded toolchain which is a very common one too) (2) git clone git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git my3.0.y-stbl-rt (3) git checkout -b my3.0.6 v3.0.6 (4) As suggested by Frank Rowand in http://marc.info/?l=linux-rt-users&m=131352000121791&w=2 you can create a default kernel configuration for panda board as follows or use the .config file he had attached in the same post and tweak it. (4.1) make ARCH=arm CROSS_COMPILE=arm-none-linux-gnueabi- omap2plus_defconfig (4.2) make ARCH=arm CROSS_COMPILE=arm-none-linux-gnueabi- menuconfig # this allows USB thumb drive # Device Drivers -> USB support -> EHCI HCD (USB 2.0) support CONFIG_USB_EHCI_HCD=y # Device Drivers -> USB support -> EHCI support for OMAP3 and later chips CONFIG_USB_EHCI_HCD_OMAP=y # ethernet device # Device Drivers -> Network device support -> USB Network Adapters -> # Multi-purpose USB Networking Framework -> # SMSC LAN95XX based USB 2.0 10/100 ethernet devices CONFIG_USB_NET_SMSC95XX=y # Preeemption model # Kernel Features -> Preemption Model -> Preemptible Kernel (Low-Latency Desktop) (5) make ARCH=arm CROSS_COMPILE=arm-angstrom-linux-gnueabi- uImage (5.1) ssh this uImage to your panda board and see whether it works. panda board sometimes hangs on boot. Resetting and/or power-cycling a few times will make it boot. If it does not boot at all after several resets, then it is possible you did not build the kernel correctly. I usually do this step as a sanity check that there is no major issue with building and booting the vanilla kernel without the RT patch. (6) patch -p1 --merge -i <path_of_the_patch_file>/patch-3.0.6-rt17.patch (7) Run menuconfig again as in Step-4 and choose "Fully Preemptible Kernel (RT)" as the preemption model. (8) Build the kernel image and test it as in Step-5. (9) Run the cyclictest and see the latencies. Latencies are huge without the high resolution timer. (10) You can turn on the high-resolution timer (by disabling the 32 KHz Timer and applying the high-resolution timer patch) and measure the latencies. There is a bug on panda board (or OMAP4430 in general) about time-stamps jumping backwards (which appears to be an issue with the clock source). If you search for my other posts in this mailing list, you will find info and links about that. Hope this helps. Thanks, Sankara -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
