i.MXプロセッサナレッジベース

ディスカッション

Overview i.MX28EVK Setup Build Yocto Project image Create a SDCARD from the Linux Host Boot i.MX28EVK Create file system on USB Create Mount Point and Mount the USB device Create 250 MB File Create Exports File Restart NFS Server Ubuntu Linux Host Setup Create Mount Directory Mount i.MX28EVK Exported Directory Access the NFS mounted directory Overview This document describes the steps for configuring a NFS Server running on an i.MX Application Processor - in this case the evaluation board i.MX28 EVK. Once the NFS server is running, an Ubuntu 12.04 Linux host is then configured to NFS mount the i.MX28EVK exported directory. The Ethernet interface is used for the connection transport. A block diagram of the connection setup is shown below: An Ethernet switch provided the Ethernet connection between the Linux Host and the i.MX28EVK. A thumb drive was connected to the USB port on the i.MX28EVK which was used for the exported directory. i.MX28EVK Setup Build Yocto Project image Use core-image-minimal and add packages to conf/local.conf to support NFS MACHINE=imx28evk source setup-environment mx28-evk echo "CORE_IMAGE_EXTRA_INSTALL += \"bash kernel-modules nfs-utils\" " >> conf/local.conf bitbake core-image-minimal When bitbake finishes the images are found in tmp/deploy/images/imx28evk Create a SDCARD from the Linux Host sudo dd if=/tmp/deploy/images/imx28evk/core-image-minimal-imx28evk.sdcard of=/dev/sdc bs=4M && sync Boot i.MX28EVK Insert the SDCARD into slot 0 on the bottom side of the i.MX28EVK and connect the serial console. Power-on and push the POWER button on the lower conner to turn on. The Login credentials are User Name: root There is no password configured by default. Create file system on USB The USB drive had one partition which was formatted with vfat file system: mkfs.vfat /dev/sdb1 Create Mount Point and Mount the USB device mkdir /mnt/usb mount /dev/sdb1 /mnt/usb Create 250 MB File dd if=/dev/zero of=/mnt/usb/file1.txt bs=512K count=500 Create Exports File echo "/mnt/usb *(rw,sync,no_root_squash,no_subtree_check)" > /etc/exports Restart NFS Server /etc/init.d/nfsserver stop /etc/init.d/nfsserver start Ubuntu Linux Host Setup Create Mount Directory sudo mkdir /mnt/remote Mount i.MX28EVK Exported Directory sudo mount -t nfs 10.85.1.10:/mnt/usb /mnt/remote Access the NFS mounted directory ls /mnt/remote

With Qt5 you will find the addition of new technologies that will make your development much easier. Qtquick2 SceneGraph Qml Qt5 is backwards compatible, that means you can run your Qt 4.8 applications, But that doesn't mean they will have the best performance, sometimes it is better to do a porting to use the newest features. Qt5, has two options to paint components into the screen. Painting in Qt 5 is primarily done either with: The imperative QPainter API Qt’s declarative UI language, QML, and its scene graph back-end. QPainter As this document mentions Qt5GraphicsOverview | Qt Wiki | Qt Project The Qpainter engine uses software to paint, and is used when drawing Qimages or Qwidgets. Its advantage over OpenGL is the high quality when antialiasing is enabled, and a complete feature set. The Qpainter can use an OpenGL engine, but as the document mentions it is more suceptible to state changes. And has to be used carefully. QML & Scene Graph. All visual QML items are rendered using the scene graph, a low-level, high-performance rendering stack, closely tied to OpenGL. Qt Quick 2 makes use of a dedicated scene graph based on OpenGL ES 2.0 or OpenGL 2.0 for its rendering. Using a scene graph for graphics rather than the traditional imperative painting systems (QPainter and similar), means the scene to be rendered can be retained between frames and the complete set of primitives to render is known before rendering starts. This opens up for a number of optimizations, such as batch rendering to minimize state changes and discarding obscured primitives. The QML scene graph is a new back-end for QML in Qt 5, and is based on OpenGL. It generally improves the performance of QML significantly over the QPainter-based back-end used in Qt 4. It achieves better performance in a number of ways: The scene graph uses OpenGL directly instead of going though a QPainter which could be using either the raster or OpenGL paint engine. This means that all the resources, like geometry, textures and shaders can be stored in a format suitable for OpenGL rather than using classes such as QPainterPath, QPixmap, QBrush, or QPen, which the QPainter would need to translate into OpenGL primitives and possibly cache. QML, being a declarative language, defines how the end result should look like, but it doesn’t define how and in which order each individual element is drawn. The drawing can therefore be reordered to reduce the number of state changes, or merged to reduce the number of draw calls. The scene graph uses a separate render thread, and synchronizes the animations with the vertical retrace on platforms where this can be supported. The render thread allows the preparation of the next frame to be done at the same time the current frame is being rendered. This has a positive effect also on single-core systems, since the render thread might block on OpenGL commands. The synchronization with the vertical retrace improves the perceived smoothness of the animations. We have tested on i.MX6 Both options, having the best results using QML Qtquick2 elements. When we tried using QtPainter via Widgets we face the problem that if not using a windowing system like X11 or Wayland the painter wont work well and will only show the QtGLWidget. With QML scene graph we are able to have an OpenGL element and a Qt element on the same environment, and there is an easy way to communicate one with the other and share variables. Please look at the example results here: I.MX6 scene graph Qt5 - YouTube And the great advantage, the sceneGraph is all accelerated via OpenGL.

For default Android JB4.3 GA1.1.0 BSP, iMX6SL EVK board doesn't support BlueTooth, attached is the patch for the BSP to support AR3002 bluetooth. The hardware information can be found at: https://community.freescale.com/docs/DOC-95016 iMX6SL_EVK_BlueTooth_Patch_for_JB4.3_1.1.0.tar.gz: these are the patch files. iMX6SL_EVK_BlueTooth_Support_for_JB4.3_1.1.0.tar.gz: these are the patched source code files. Please select one of above to use.

Tested on imx28 EVK Rev. D. When plug-in or plug-out cable on eth0 port, eth1 port (and vice versa) will also be reset and the communication will be interrupted. Reason: Both Ethernet PHYs on EVK board share the same GPIO as their reset pin, in software the function name is mx28evk_enet_gpio_init. So any call to pdata->init() in fec.c will reset both PHY at the same time. In order to avoid such problem, you have to use 2 individual GPIO for the PHY reset.

Hello! Here's some CODE!!! #include <iostream> #include <stdio.h> #include <assert.h> #include <string.h> #include <fcntl.h> #include <malloc.h> #include <math.h> #include <stdlib.h> //#define EGL_USE_GLES2 #include <GLES2/gl2.h> #include <EGL/egl.h> #include <GLES2/gl2ext.h> #include <EGL/eglext.h> #include <termios.h> #include <unistd.h> #include <fcntl.h> #ifdef EGL_USE_X11 #include <X11/X.h> #include <X11/Xlib.h> #endif EGLDisplay egldisplay; EGLConfig eglconfig; EGLSurface eglsurface; EGLContext eglcontext; EGLNativeWindowType eglNativeWindow; EGLNativeDisplayType eglNativeDisplayType; EGLNativeDisplayType fsl_getNativeDisplay() { EGLNativeDisplayType eglNativeDisplayType = NULL; #if (defined EGL_USE_X11) eglNativeDisplayType = XOpenDisplay(NULL); assert(eglNativeDisplayType != NULL); #elif (defined EGL_API_FB) eglNativeDisplayType = fbGetDisplayByIndex(0); //Pass the argument as required to show the framebuffer #else display = EGL_DEFAULT_DISPLAY; #endif return eglNativeDisplayType; } EGLNativeWindowType fsl_createwindow(EGLDisplay egldisplay, EGLNativeDisplayType eglNativeDisplayType) { EGLNativeWindowType native_window = (EGLNativeWindowType)0; #if (defined EGL_USE_X11) Window window, rootwindow; int screen = DefaultScreen(eglNativeDisplayType); rootwindow = RootWindow(eglNativeDisplayType,screen); window = XCreateSimpleWindow(eglNativeDisplayType, rootwindow, 0, 0, 400, 533, 0, 0, WhitePixel (eglNativeDisplayType, screen)); XMapWindow(eglNativeDisplayType, window); native_window = window; #else const char *vendor = eglQueryString(egldisplay, EGL_VENDOR); if (strstr(vendor, "Imagination Technologies")) native_window = (EGLNativeWindowType)0; else if (strstr(vendor, "AMD")) native_window = (EGLNativeWindowType) open("/dev/fb0", O_RDWR); else if (strstr(vendor, "Vivante")) //NEEDS FIX - functs don't exist on other platforms { #if (defined EGL_API_FB) native_window = fbCreateWindow(eglNativeDisplayType, 0, 0, 0, 0); #endif } else { printf("Unknown vendor [%s]\n", vendor); return 0; } #endif return native_window; } void fsl_destroywindow(EGLNativeWindowType eglNativeWindowType, EGLNativeDisplayType eglNativeDisplayType) { (void) eglNativeWindowType; #if (defined EGL_USE_X11) //close x display XCloseDisplay(eglNativeDisplayType); #endif } void GLInit (void) { static const EGLint s_configAttribs[] = { EGL_RED_SIZE, 5, EGL_GREEN_SIZE, 6, EGL_BLUE_SIZE, 5, EGL_ALPHA_SIZE, 0, EGL_SAMPLES, 0, EGL_NONE }; EGLint numconfigs; printf("1"); eglNativeDisplayType = fsl_getNativeDisplay(); printf("2"); egldisplay = eglGetDisplay(eglNativeDisplayType); printf("3"); eglInitialize(egldisplay, NULL, NULL); printf("4"); assert(eglGetError() == EGL_SUCCESS); printf("5"); eglBindAPI(EGL_OPENGL_ES_API); printf("6"); eglChooseConfig(egldisplay, s_configAttribs, &eglconfig, 1, &numconfigs); assert(eglGetError() == EGL_SUCCESS); assert(numconfigs == 1); printf("7"); eglNativeWindow = fsl_createwindow(egldisplay, eglNativeDisplayType); assert(eglNativeWindow); printf("8"); eglsurface = eglCreateWindowSurface(egldisplay, eglconfig, eglNativeWindow, NULL); assert(eglGetError() == EGL_SUCCESS); printf("9"); EGLint ContextAttribList[] = { EGL_CONTEXT_CLIENT_VERSION, 2, EGL_NONE }; eglcontext = eglCreateContext( egldisplay, eglconfig, EGL_NO_CONTEXT, ContextAttribList ); assert(eglGetError() == EGL_SUCCESS); printf("10"); eglMakeCurrent(egldisplay, eglsurface, eglsurface, eglcontext); assert(eglGetError() == EGL_SUCCESS); } void GLEnd (void) { printf("Cleaning up...\n"); eglMakeCurrent(egldisplay, EGL_NO_SURFACE, EGL_NO_SURFACE, EGL_NO_CONTEXT); assert(eglGetError() == EGL_SUCCESS); eglDestroyContext(egldisplay, eglcontext); eglDestroySurface(egldisplay, eglsurface); fsl_destroywindow(eglNativeWindow, eglNativeDisplayType); eglTerminate(egldisplay); assert(eglGetError() == EGL_SUCCESS); eglReleaseThread(); } int main (int argc, char **argv) { GLInit(); for( int i = 0; i < 100000; ++i) { glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); // Clear The Screen And The Depth Buffer glClearColor (.0f, .0f, 1.0f, 1.0f); eglSwapBuffers (egldisplay, eglsurface); } GLEnd(); } Above code is stitched together from samples from GPU SDK. I've built an image using Yocto (dylan branch) and build the meta-toolchain-qt (and qte). Successfully managed to build above code ONLY FOR X11. Issues i am facing: 1) if build for X11 and run app on the board, window shows, but it stays white, when it should be BLUE...doesn't update no matter what color i glClear to. 2) when compiling with FB, fbCreateWindow and etc don't get recognized, i.e. undefined reference to `fbCreateWindow' .... WHAT header contains these functions???? 3) if even the basic samples don't work, how the hell is anybody supposed to build a GL application on this board?? --- more rhetorical than a real question, just frustrated here... what did i do wrong? 4) please show me a working tutorial or some code on how to get this EGL context initialized...i'm running at wit's end here...

1. Description 1) Support HDMI interlaced display mode, the followed format had been verified. CEA format 5: 1920x1080i @60Hz CEA format 6&7: 720(1440)x480i @60Hz CEA format 20: 1920x1080i @50Hz CEA format 21&22: 720(1440)x576i @50Hz 2) Support LCD interface for interlaced display mode, 1920x1080i @50Hz(CEA format 20) had been verified. 2. File List -- 0001-IPUv3-support-interlaced-display-mode.patch Patch to support interlaced display output for iMX6 ipuv3. -- 0002-iMX6-HDMI-support-interlaced-display-mode.patch Patch to support interlaced display mode for iMX6 HDMI driver. -- 0003-iMX6-LCD-interface-supports-1920x1080i50-mode.patch Patch to support interlaced display mode for iMX6 LCD interface driver. -- readme.txt this file, please refer to it before use the patches 3. Requirement - iMX6 SabreSD board. - L3.0.35_4.1.0_GA_iMX6DQ kernel. 4. How to use -- Copy the patch files to kernel folder. $ cd ~/ltib/rpm/BUILD/linux-3.0.35/ $ git apply ./0001-IPUv3-support-interlaced-display-mode.patch $ git apply ./0002-iMX6-HDMI-support-interlaced-display-mode.patch $ git apply ./0003-iMX6-LCD-interface-supports-1920x1080i50-mode.patch -- Build the new kernel image: $ cd ~/ltib/rpm/BUILD/linux-3.0.35 $ export CROSS_COMPILE=/opt/freescale/usr/local/gcc-4.6.2-glibc-2.13-linaro-multilib-2011.12/fsl-linaro-toolchain/bin/arm-fsl-linux-gnueabi- $ export ARCH=arm $ make imx6_defconfig $ make uImage -- Uboot parameters for video mode Output 1080i50 display mode on HDMI: "video=mxcfb0:dev=hdmi,1920x1080Mi@25,if=RGB24,bpp=32" Output 1080i60 display mode on HDMI: "video=mxcfb0:dev=hdmi,1920x1080Mi@30,if=RGB24,bpp=32" Output 576i50 display mode on HDMI: "video=mxcfb0:dev=hdmi,1440x576Mi@25,if=RGB24,bpp=32" Output 480i60 display mode on HDMI: "video=mxcfb0:dev=hdmi,1440x480Mi@30,if=RGB24,bpp=32" Output 1080i50 display mode on LCD interface: "video=mxcfb0:dev=lcd,LCD-1080I50,if=RGB565,bpp=32" -- Switch HDMI interlaced mode $ echo S:1920x1080i-50 > /sys/class/graphics/fb0/mode $ echo S:1920x1080i-60 > /sys/class/graphics/fb0/mode $ echo S:1440x480i-50 > /sys/class/graphics/fb0/mode $ echo S:1440x576i-60 > /sys/class/graphics/fb0/mode 5. Know issue 1) When the interlaced display and another display work on same IPU, blank and unblank the interlaced display will get the followed IPU warning, but the display still works due to IPU can revover from the error. imx-ipuv3 imx-ipuv3.0: IPU Warning - IPU_INT_STAT_5 = 0x00800000 imx-ipuv3 imx-ipuv3.0: IPU Warning - IPU_INT_STAT_10 = 0x00080000 2015-05-13 update: Replace the fourth patch to make interlace display mode follow CEA-861-specification The patch "0004-IPU-fine-tuning-the-interlace-display-timing-for-CEA.patch" was fine tuned for CEA-861-D specification on interlaced mode display. Please use this patch to replace the old 0004 patch. 2016-05-20 Update: For 3.0.35 BSP, add patch 0005-IPU-update-interlaced-video-mode-parameters-to-align.patch Align the interlaced video mode parameters to progressive mode. 0006-IPU-update-IDMAC-setting-for-interlaced-display-mode.patch Udate the IDMAC setting for interlaced display mode, output odd field data from memory first, it aligns with IPU DI timing, odd field first. For 3.14.52 BSP, created the new patch L3.14.52_1.1.0_GA_HDMI_Interlaced_Mode_Patch_2016_05_20.zip.

Question: How to enable HAB on the MX28, following the recommendations of AN4555 to get the "get_hab_status()" function working, but has run into an issue. Question #1 They believe they have all the HAB components worked out that are inputs to the efltosb tool as they are able to successfully run U-boot to the interactive prompt. However, at the point where they: - call the rvt_report_status() function, their board says "### ERROR ### Please RESET the board ###". - call the rvt_entry(), their board prints some garbage characters on the screen and then hangs. This suggests that there is something wrong with the clock that in turn affects the baudrate on the serial console causing the above behavior. Question #2 Is there a concept of a "Bound Signature" in HABv4 as there is in HABv3? Any chance the addresses for the rvt_ calls are incorrect? Can you provide the u-boot source? Either Bound signature verification or UID is never mentioned in the HABv4 Application Note. So I suppose it is not supported. We have made assumptions about the RVT function pointer offsets. The HAB 4 API does not explicitly say the offsets but uses a rvt_base::function_name notation. We have assumed that function pointers are placed in order, at every word offset beyond the RVT header. We have confirmed the RVT header exists at the latest address in the reference manual based on a memory dump but we cannot be certain the function offsets we have setup are correct. As far as source code, we modeled our changes for our mx28 board off of the hab.c and hab.h files available from the mainline u-boot for the mx6 architecture. This is basically the same code get_hab_status code that is written in the AN4555 document. We did HAB API function pointer addresses to match the updated RVT base address and assumed offsets. Answer: Here are the first 3 instructions from report_status(), could your customer check the instructions from the address which they called is correct? <report_status>: : b087b570 addlt fp, r7, r0, ror r5 : 1c0e1c05 stcne 12, cr1, [lr], {5} : 22182433 andscs r2, r8, #855638016 ; 0x33000000

Whenever possible is always better to avoid Virtual Machines when compiling Android as the building process might take several hours to even a day or two depending on the resources available to the Virtual Machine. Sometimes, however, a VM is the only option available and here are some useful considerations when using a VM as a host for Android builds. This document addresses some of the requirements for the Kit-Kat Freescale Android BSP although some may also apply to the general Android releases. It’s recommended to use Ubuntu 12.04 64-bits as Host OS. As for HDD space, which is often a limited resource when using VM, Google states that at least 30GB of disk space in order to build the Android Tree. In practice, however, it’s recommended to allocate at least 100GB to the VM HDD if possible, as additional packages and larger images can quickly deplete disk space. Installing Java (not VM specific) It’s recommended (albeit not mandatory) to uninstall any version of Java other than the recommended for the specific Android release for which we will build. In the case of Kit-Kat this is Java 6. Another option is to just switch the java alternatives to the ones we want to use for Android. There is a document with the exact instructions on how to install the required JKD on the following link. How to install Sun's JDK in Ubuntu for Android build Starting KitKat, however, it’s also necessary to install javap (Java Class File Dissasembler) so please also add these steps when following this document. $ sudo update-alternatives --install /usr/bin/javac javac /usr/lib/jvm/jdk1.6.0_45/bin/javac 1 $ sudo update-alternatives --config javac Memory considerations when using a Virtual Machine Using a Virtual Machine slows down the image building considerably. Besides that memory system is often also an issue. Having insufficient RAM especially on the linking part of the image build may cause a number of issues that are difficult to troubleshoot. In these cases it’s good to take a look at the resource monitor to see if indeed the RAM was depleted. One way to make up for the limited RAM is using a bigger swap. Google recommends at least 16GB of RAM/swap so it’s not uncommon to create a 10GB swap when working in VM, to do this please use the following commands. $ sudo fallocate -l 10g /mnt/10GB.swap $ sudo chmod 600 /mnt/10GB.swap $ sudo mkswap /mnt/10GB.swap $ sudo swapon /mnt/10GB.swap This will helps with the memory requirements but won’t speed up the build process. Building time with VM, as previously stated, will take several hours or even a couple of days depending on computing power.

Here is a quick summary at booting Linux on the i.MX 6 sabre sd platform. This assumes you already have u-boot working on your platform as described here. This implies you already have a "working" Linux development environment with some ARM cross-compilers at hand (e.g. Debian + Emdebian). Get Linux sources We will use git to fetch Linux sources: $ git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git This should create a linux directory with all the latest sources (after a while). Note that for more stability you might want to checkout a release instead of the latest version; to do so, list the available release tags with e.g. git tag -l 'v*', and git checkout <the-desired-tag>. Compile Assuming your cross compiler is called e.g. arm-linux-gnueabihf-gcc, you can compile by doing: $ cd linux $ export ARCH=arm $ export CROSS_COMPILE=arm-linux-gnueabihf- $ make imx_v6_v7_defconfig $ make You then need to supply a LOADADDR (as joowonkim pointed out); do: $ make uImage LOADADDR=0x10008000 This should create a number of files, including arch/arm/boot/uImage and arch/arm/boot/dts/imx6q-sabresd.dtb. Put on SD We need a proper FAT partition on the SD card, from which u-boot will be able to load the kernel and dtb. Also, we need to make sure we leave some space for u-boot starting from offset 1024B. Here is an example SD card layout: +-----+------+--------+-----+---------------- | MBR | ... | u-boot | ... | FAT partition ... +-----+------+--------+-----+---------------- 0 512 1024 1M (offsets in bytes) Here is an example SD card layout, as displayed by fdisk: Device Boot Start End Blocks Id System /dev/sdc1 2048 8054783 4026368 c W95 FAT32 (LBA) (units: 512B sectors) You can format the FAT partition, mount, copy and unmount with: $ mkfs.vfat /dev/<your-sd-card-first-partition> $ mount /dev/<your-sd-card-first-partition> /mnt $ cp arch/arm/boot/uImage arch/arm/boot/dts/imx6q-sabresd.dtb /mnt/ $ umount /mnt Your SD card first partition is typically something in /dev/sd<X>1 or /dev/mmcblk<X>p1. Note that you need write permissions on the SD card for the command to succeed, so you might need to su - as root, or use sudo, or do a chmod a+w as root on the SD card device node to grant permissions to users. Also, be sure to have u-boot on the SD card as explained in this post. Boot! That's it; u-boot already knows how to deal with your kernel by default so you are good to go. Insert the SD card into the SD card slot of your i.MX6 sabre sd platform, connect to the USB to UART port with a serial terminal set to 115200 baud, no parity, 8bit data and power up the platform. You should see u-boot messages: U-Boot 2013.07-rc1-00014-g74771f4 (Jun 21 2013 - 16:27:39) u-boot should load the uImage and dtb from SD card and boot the kernel: (...) reading uImage 4215344 bytes read in 449 ms (9 MiB/s) Booting from mmc ... reading imx6q-sabresd.dtb 22818 bytes read in 22 ms (1012.7 KiB/s) ## Booting kernel from Legacy Image at 12000000 ... Image Name: Linux-3.10.0-rc6 Image Type: ARM Linux Kernel Image (uncompressed) Data Size: 4215280 Bytes = 4 MiB Load Address: 10008000 Entry Point: 10008000 Verifying Checksum ... OK ## Flattened Device Tree blob at 11000000 Booting using the fdt blob at 0x11000000 Loading Kernel Image ... OK OK Using Device Tree in place at 11000000, end 11008921 Starting kernel ... The kernel should boot: Booting Linux on physical CPU 0x0 Linux version 3.10.0-rc6 (vstehle@debian) (gcc version 4.7.2 (Debian 4.7.2-5) ) #1 SMP Fri Jun 21 18:09:26 CEST 2013 By default, the kernel will try to mount a root filesystem from the SD card second partition, as can be read in the default kernel command line: (...) Kernel command line: console=ttymxc0,115200 root=/dev/mmcblk1p2 rootwait rw ...but we did not prepare a root filesystem partition, so after a number of boot messages the kernel will wait indefinitely: (...) mmc1: new SDHC card at address b368 (...) mmcblk0: p1 (...) Waiting for root device /dev/mmcblk1p2... We will see in another post how to prepare this root filesystem on the second SD card partition. Enjoy! See also... If you plan to compile Linux often, you might want to use a C compiler cache; see this post. Once you have Linux booting on your platform the next step is to give it a root filesystem. See this post for a Debian root filesystem, this post for a minimal busybox filesystem and this post for generating a root filesystem with buildroot.

Gst-entrans.spec GStreamer gst-inspect Tool GStreamer gst-launch Tool i.MX27 Video GST Caps i.MX27 ADS Board Video GST Play i.MX27 ADS Board Video GST Encode Gst-inspect mfw vpuencoder 1109 2.0.2 GStreamer Miscellaneous GStreamer i.MX6 Decoding GStreamer i.MX6 Multi-Display GStreamer i.MX6 Encoding GStreamer i.MX6 Image Display GStreamer i.MX6 Camera Streaming GStreamer i.MX6 Image Capture GStreamer i.MX6 Multi-Overlay GStreamer i.MX6 Pipelines GStreamer ffmpeg GStreamer Streaming GStreamer Transcoding and Scaling Testing GStreamer Tracing GStreamer Pipelines Useful Gstreamer Commands Using a USB Camera with GStreamer

Agenda: 1. How-to M4 boot-up from TCM, DDR, OCRAM or QSPI in i.MX7D SABRE board 2. About multicore communication, Linux / Cortex-A and FreeRTOS / Cortex-M a. RPMsg Ping-Pong FreeRTOS demo b. RPMsg String Echo FreeRTOS demo 3. Multi-core Resource Sharing and Protection, RDC (Resource Domain Controller), Master Assignment Registers, Peripheral Mapping and Memory region Map 4. RDC settings in FreeRTOS BSP

by b47504 Overview This document is intended to introduce debug tips about i.MX power management based on i.MX Android software. The following topics are involved in this document: How to debug suspend/resume issues How to do power optimization How to debug suspend/resume issues General method: Capture more PM debug message Enable PM debug system to get more info about PM in kernel and debug interface Power management options ---> [*] Power Management Debug Support [*] Extra PM attributes in sysfs for low-level debugging/testing Enable wakelock debug_mask to capture more message about wakelock root@android:/ # echo 15 > /sys/module/wakelock/parameters/debug_mask root@android:/ # echo 15 > /sys/module/userwakelock/parameters/debug_mask Enable earlysuspend debug_mask to capture more message about early suspend and late resume. root@sabresd_6dq:/ # echo 15 > /sys/module/earlysuspend/parameters/debug_mask Add no_console_suspend=1 to the boot option for kernel This makes the system print more useful info before entry in suspend Eg: --- a/sabresd_6dq/BoardConfig.mk +++ b/sabresd_6dq/BoardConfig.mk -BOARD_KERNEL_CMDLINE := console=ttymxc0,115200 init=/init video=mxcfb0:dev=ldb,bpp=32 video=mxcfb1:off video=mxcfb2:off fbmem=10M fb0base=0x27b00000 vmalloc=400M androidboot.console=ttymxc0 androidboot.hardware=freescale +BOARD_KERNEL_CMDLINE := console=ttymxc0,115200 init=/init video=mxcfb0:dev=ldb,bpp=32 video=mxcfb1:off video=mxcfb2:off fbmem=10M fb0base=0x27b00000 vmalloc=400M androidboot.console=ttymxc0 androidboot.hardware=freescale no_console_suspend=1 System cannot enter suspend mode Check below setting items have been disabled: § Whether the usb cable has been removed(usb gadget will hold a wake lock) § Setting->Display->Sleep, check whether the inactivity timeout period setting is longer than your expected time. § Setting->System->Developer options->stay awake(stay awake not be set), check whether the option is disabled Check if all wake locks have been released(You can see which wake lock is held, and then debug into the specific module): root@sabresd_6dq:/ # cat /sys/power/wake_lock System could not resume from suspend/System crash when resume or suspend Check the PMIC_STBY_REQ signal. System use PMIC_STBY_REQ signal to notify power management IC to change voltage from standby voltage to functional voltage and vice versa. In general, pmic_stby_req pin is connected to pmic standby pin. So measure the pin to check whether the de-assert signal is triggered. If the signal is not triggered, we may consider whether wake-up sources are correctly setup. If the signal is triggered, we may double-check whether the pmic supply power normally. And not limited to the two points, we should also double-check everything we doubt according to the system log and hardware measured waves. Using Trace32 or ICE to locate the problem. Please view trace32 website to get more details. Track from mx6_suspend_enter in arch/arm/mach-mx6 . Track "state" value and try to map to different the low power mode via function mxc_cpu_lp_set. Check "mx6_suspend.S" which conduct the detailed operations in suspend: "MEM" is mapped to "dormant" mode. So goto "dormant" symbol and try to dump different operations to narrow down suspend/resume failure If this failure maybe related to DDR operation, try to dummy DDR IO relative low power operation. Using ram console to dump kernel log after reboot. Ram console will keep one kernel log copy into one certain memory space. You can use the following command to check last time kernel log, if memory power was not cut off during the reboot process. Eg(if it is the first time boot, you cannot find the /proc/last_kmsg file): root@sabresd_6dq:/ # cat /proc/last_kmsg Kernel resume back from suspend but android not This is usually introduced by the wrong key layout file Use getevent/sendevent tool to get power key scan code #getevent Correct the Keylayout file system/usr/keylayout/****.kl Correct the scandcode with your power key report value to Match the POWE key Suspend/Resume consume too much time: We can print the specific module name and time consume details, if the module's suspend/resume time consume more than the threshold parameter by read/write /sys/power/device_suspend_time_threshold file. By default, the parameter is setup to 0, via disabled the function. We can enable it by the following command: Eg: root@android:/ # echo 10 > /sys/power/device_suspend_time_threshold This command means that if the module's suspend/resume time consume more than 10 us, the system will print the module's detail out. If you want to know the more details how to implement it on kernel, please check kernel/power/main.c Notes: Can use the shell command to enter different system level low power modes for debug (For more details: you can check Linux_6DQ_RM.pdf): #echo mem > /sys/power/state #echo standby > /sys/power/state How to do power optimization Runtime Mode Check whether CPUFreq and Bus_freq scale are enabled root@android:/ # cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor root@android:/ # cat /sys/devices/platform/imx_busfreq.0/enable More details about this, please refer to "Documentation/cpu-freq/ governors.txt” . Check whether the system bus is working on your expected frequency. For MX6Q: root@android:/ # cat /sys/kernel/debug/clock/osc_clk/pll2_528_bus_main_clk/periph_clk/mmdc_ch0_axi_clk/rate Check CPU Loading and Interrupt(cat /proc/interrupts) root@android:/ # cat /proc/interrupts Through this command you can check whether some module will trigger interrupt frequently. And consider that whether we have some chances to reduce the interrupt count. Check clock tree carefully to see which clocks are not gated off but no any modules need them. root@android:/ # powerdebug -d -c Reduce GPU frequency.GPU also offered interface to modify the frequency. According to your own product, you can reduce the gpu frequency. Default gpu3DMaxClock is set to 64 in init.rc file, we can tuning a suitable value by ourselves. diff --git a/imx6/etc/init.rc b/imx6/etc/init.rc index 8c420b5..eb11ffe 100755 --- a/imx6/etc/init.rc +++ b/imx6/etc/init.rc @@ -397,6 +397,9 @@ on boot # Set GPU 3D minimum clock to 3/64 write /sys/module/galcore/parameters/gpu3DMinClock 3 +# Set GPU 3D maximum clock to 64/64 + write /sys/module/galcore/parameters/gpu3DMaxClock 64 + Suspend Mode Check whether all devices enter suspend mode or low power mode: Add debug message into devices drivers to check whether all devices driver suspend interface are called Use oscilloscope to measure the related signal (depend on specific device datasheet and custom hw design) to check whether every device enter low power mode Remove devices from the board(or rmmod the device driver) , and do hardware rework to exclude some hardware module if needed. Then we can figure out which module introduced the high consumption, and debug into the specific module. Add debug message in device drivers which may lead high power consumption, catch the waveform from these modules which may impact the high power consumption Check whether DDR enter in self-refresh mode(Please check the DDR datasheet to figure out which pin indicate self-refresh state, and check it with oscilloscope) Config GPIO PADs as output zero or input mode (depending to HW design) Cut off LDOs/DCDCs which no modules need (depending to HW design) Check all PLLs will cut off, just 32KHZ sleep clock living

Here are my experiences for compiling Qt5.3.0-beta1 on Yocto. Special thanks to Martin Jansa, the maintainer of the meta-qt5 layer and his help on this. My original procedure was based on this tutorial: Building Qt5 using yocto on Wandboard - Wandboard Wiki Reason: Qt5.3 contains a nice new plugin that allows the use of gstreamer output for textures without the CPU intensive step of copying them (Gerrit Code Review). This allows to play even full HD videos and apply all the power of Qt5 (e.g. shaders) to them. Steps: Setup your repo: repo init -u https://github.com/Freescale/fsl-community-bsp-platform -b master-next; repo sync Download meta-qt5 branch: cd sources; git clone -b jansa/qt5-5.3.0-beta1 https://github.com/meta-qt5/meta-qt5.git Checkout a specific revision: cd meta-qt5; git checkout 92be18a3a14deed9d38b8fc6e89f09ba4d730597 Apply the following patch (maybe later no longer needed): diff --git a/recipes-qt/qt5/qt5.inc b/recipes-qt/qt5/qt5.inc index dfc1c76..a2f9a73 100644 --- a/recipes-qt/qt5/qt5.inc +++ b/recipes-qt/qt5/qt5.inc @@ -54,6 +54,7 @@ FILES_${PN}-tools-dbg = " \ " FILES_${PN}-plugins-dbg = " \ ${OE_QMAKE_PATH_PLUGINS}/*/.debug/* \ + ${OE_QMAKE_PATH_PLUGINS}/*/*/.debug/* \ " # extra packages @@ -98,6 +99,7 @@ FILES_${PN}-tools = " \ " FILES_${PN}-plugins = " \ ${OE_QMAKE_PATH_PLUGINS}/*/*${SOLIBSDEV} \ + ${OE_QMAKE_PATH_PLUGINS}/*/*/*${SOLIBSDEV} \ " FILES_${PN}-mkspecs = "\ ${OE_QMAKE_PATH_ARCHDATA}/mkspecs \ Define your machine: export MACHINE=xxx (replace with your board) Setup build environment: cd .. ; . setup-environment build edit your local layer conf ("conf/bblayers.conf") and add the following two lines: ${BSPDIR}/sources/meta-openembedded/meta-ruby \ ${BSPDIR}/sources/meta-qt5 \ edit your local.conf and add the following lines: DISTRO_FEATURES_remove = "x11 wayland" IMAGE_INSTALL_append = " \ firmware-imx-vpu-imx6q \ firmware-imx-vpu-imx6d \ " IMAGE_INSTALL_append = " \ cpufrequtils \ nano \ packagegroup-fsl-gstreamer \ packagegroup-fsl-tools-testapps \ packagegroup-fsl-tools-benchmark \ gstreamer \ gst-plugins-base-app \ gst-plugins-base \ gst-plugins-good \ gst-plugins-good-rtsp \ gst-plugins-good-udp \ gst-plugins-good-rtpmanager \ gst-plugins-good-rtp \ gst-plugins-good-video4linux2 \ qtbase-fonts \ qtbase-plugins \ qtbase-tools \ qtbase-examples \ qtdeclarative \ qtdeclarative-plugins \ qtdeclarative-tools \ qtdeclarative-examples \ qtdeclarative-qmlplugins \ qtmultimedia \ qtmultimedia-plugins \ qtmultimedia-examples \ qtmultimedia-qmlplugins \ qtsvg \ qtsvg-plugins \ qtsensors \ qtimageformats-plugins \ qtsystems \ qtsystems-tools \ qtsystems-examples \ qtsystems-qmlplugins \ qtscript \ qt3d \ qt3d-examples \ qt3d-qmlplugins \ qt3d-tools \ qtwebkit \ qtwebkit-examples-examples \ qtwebkit-qmlplugins \ cinematicexperience \ " PACKAGECONFIG_append_pn-qtmultimedia = " gstreamer010" QT5_VERSION = "5.2.1+5.3.0-beta1+git%" PREFERRED_VERSION_qtbase-native = "${QT5_VERSION}" PREFERRED_VERSION_qtbase = "${QT5_VERSION}" PREFERRED_VERSION_qtdeclarative = "${QT5_VERSION}" PREFERRED_VERSION_qtjsbackend = "${QT5_VERSION}" PREFERRED_VERSION_qtjsbackend-native = "${QT5_VERSION}" PREFERRED_VERSION_qtgraphicaleffects = "${QT5_VERSION}" PREFERRED_VERSION_qtimageformats = "${QT5_VERSION}" PREFERRED_VERSION_qtmultimedia = "${QT5_VERSION}" PREFERRED_VERSION_qtquick1 = "${QT5_VERSION}" PREFERRED_VERSION_qtquickcontrols = "${QT5_VERSION}" PREFERRED_VERSION_qtsensors = "${QT5_VERSION}" PREFERRED_VERSION_qtserialport = "${QT5_VERSION}" PREFERRED_VERSION_qtscript = "${QT5_VERSION}" PREFERRED_VERSION_qtsvg = "${QT5_VERSION}" PREFERRED_VERSION_qttools-native = "${QT5_VERSION}" PREFERRED_VERSION_qtwebkit = "${QT5_VERSION}" PREFERRED_VERSION_qtwebkit-examples = "${QT5_VERSION}" PREFERRED_VERSION_qtxmlpatterns = "${QT5_VERSION}" build an image: bitbake core-image-minimal This image will build QT5.3 for framebuffer. If you want to use it with X11, then adapt according to this tutorial: Integrate Qt5 into yocto sato image on Wandboard - Wandboard Wiki Please tell me, if I missed something. I wrote this as I remembered the steps.

This is the prototype solution to enable second display showing different things on JB4.2.2 SabreSD. Make use of Class Presentation provided by android to be embedded into Status bar. When unlock the screen, the Presentation will show on second display. Now, the solution requires one .mp4 video placed in root sdcard. Of course, you may change it to show anything. The attached Files are a layout xml file, a patch and a recorded video. The layout file should be put into android/frameworks/base/packages/SystemUI/res/layout/ folder. The patch should be applied to frameworks/base.git. The recorded video shows the dual display demo as a reference.

i.MXプロセッサナレッジベース

i.MX Processors Knowledge Base

ディスカッション

i.MX Create NFS Server and Export USB Drive: Yocto Project Distribution

Qt5 QPainter vs. QML & Scene Graph.

Patch to support BlueTooth on iMX6SL EVK board, Android JB4.3 GA1.1.0 BSP

[imx28] plug-in/out on eth0 will reset eth1 and vice versa

i.mx6 Yocto OpenGL ES 2 X11 (or FB)

Patch for iMX6 BSP to support interlaced display on HDMI and LCD interface

Q&A: How to enable MX28 HAB Problems?

I.MX处理器在Linux BSP上独立运行的详细操作

Android Virtual Machine Host Setup Recommendations

快速，简便地设置i.MX评估板，以通过USB对i.MX板进行编程

Linux on the i.MX6 sabre sd platform in a few commands

MX6SLL_LPDDR3_Script_Aid

Gstreamer

i.MX 7D with FreeRTOS

GStreamer ffmpeg

All Boards TFTP on OpenSuse

i.MX Android Debug Tips for Power Management

Compiling Qt5.3.0-beta1 on Yocto

How to Enable Second Display Showing Different Things on JB4.2.2 SabreSD

OpenVX视觉图像扩展API简介 – 基础API

i.MXプロセッサ ナレッジベース

i.MX Processors Knowledge Base

ディスカッション

i.MXプロセッサナレッジベース