This tutorial teaches how to program bootloader on a SD Card using ATK. To program kernel and root file system to the SD card, please follow this i.MX35 PDK Linux Booting SD    ATK (Advanced Toolkit)       ATK (Advanced Toolkit) is a Windows software for programming the flash memory of i.MX boards. It can be downloaded here.     Using ATK       This section will describe the procedure to erase and program the bootloader in the SD Card.       1. Connect a serial cable between PC and i.MX board.       2. Set the switches:       Debug Board: SW9 -> 0         SW10 -> 0         Personality Board: SW1 and SW2 (All bits) -> 0     3. Run ATK going to Start -> Programs -> AdvancedToolKit -> AdvancedToolKit           Set the options:       i.MX CPU -> i.MX35_TO2       Device memory -> DDR2;       Communication Channel -> Serial Port (Usually COM1)     4. Click Flash Tools to erase, program, or dump the the memory and click GO     Erasing     1. To erase SD Card, select the parameters as below: Select MMC/SD as "Flash Model".     Select Erase on "Operation Type". 2. Turn on the board and press Erase.     3. ATK shows a message: "Flash erase successful!" when card is erased   Programming     Next, program the bootloader image into the memory card following the steps below:     1. Select the parameters: The bootloader binary image file can be found into your Board Support Package (click here to download) Select Program on "Operation Type".     Address: 0x00000000     File: "mx35_3stack_redboot_mmc.bin" (or similar name that indicates a MMC/SD image)     2. Press Program.     3. Close ATK, turn off the board and set switches to:     Debug Board: SW9 -> 0       SW10 -> 0 Personality Board: SW2 (Bits 1 and 2) -> 1       SW2 (All other bits) -> 0       SW1 (All bits) -> 0     4. Open Hyper Terminal, set it 115200,N,8 and see RedBoot Prompt.      

  IMX6 SL boot process is described in Chapter 8 (System Boot) of the Reference Manual. Shortly, the loading boot data from boot SD card is performed in two stages : first read IVT, DCD, then read executable code, using the Boot Data Structure.    At first, boot ROM copies 4K byte (containing IVT and DCD ) from sector 0 of the boot SD card to internal buffer in OCRAM, located in reserved area (0x00900000 - 0x00907000). This area must not be used by user application.   Then, “after checking the Image Vector Table header value (0xD1) from Program Image, the ROM code performs a DCD check. After successful DCD extraction, the ROM code extracts from Boot Data Structure the destination pointer and length of image to be copied to RAM device from where code execution occurs”.   The IVT contains field entry - absolute address of the first instruction to execute from the image.   Note : according to Figure 8-3 (Internal ROM and RAM memory map), only OCRAM Free Area (68KB) from 0x00907000 till 0x00918000 may be used by user’s application.   The attachment contains SD-bootable example.

Requirements: Host machine with Ubuntu 14.04 UDOO Quad/Dual Board uSD card with at least 8 GB Download documentation and install latest Official Udoobuntu OS (at the moment of writing: UDOObuntu 2.1.2), https://www.udoo.org/downloads/   Overview: This document describes how to install and test Keras (Open source neural network library) and Theano (numerical computation library for python ) for deep learning library usage on i.MX6QD UDOO board.  Installation: $ sudo apt-get update && sudo apt-get upgrade update your date system: e.g. $ sudo date -s “07/08/2017 12:00” First satisfy the run-time and build time dependencies: $ sudo apt-get install python-software-properties software-properties-common make unzip zlib1g-dev git pkg-config autoconf automake libtool curl  python-pip python-numpy libblas-dev liblapack-dev python-dev libatlas-base-dev gfortran libhdf5-serial-dev libhdf5-dev python-setuptools libyaml-dev libpython2.7-dev $ sudo easy_install scipy The last step is installing scipy through pip, and can take several hours. Theano First, we have a few more dependencies to get: $sudo pip install scikit-learn $sudo pip install pillow $sudo pip install h5py With these dependencies met, we can install a stable Theano release from the git source: $ git clone https://github.com/Theano/Theano $ cd Theano Numpy 1.9 cause conflicts with armv7, so we need to change the setup.py configuration: $ sudo nano setup.py Remove line    #       install_requires=['numpy>=1.9.1', 'scipy>=0.14', 'six>=1.9.0'], And add setup_requires=["numpy"], install_requires=["numpy"], Then install it: $ sudo python setup.py install Keras The installation can occur with the command: (this could take a lot of time!!!) $ cd .. $ git clone https://github.com/fchollet/keras.git $ cd keras $ sudo python setup.py install $ LC_ALL=C $sudo pip install --upgrade keras After Keras is installed, you will want to edit the Keras configuration file ~/.keras/keras.json to use Theano instead of the default TensorFlow backend. If it isn't there, you can create it. This requires changing two lines. The first change is: "image_dim_ordering": "tf"  --> "image_dim_ordering": "th" and the second: "backend": "tensorflow" --> "backend": "theano" (The final file should look like the example below) sudo nano ~/.keras/keras.json {     "image_dim_ordering": "th",     "epsilon": 1e-07,     "floatx": "float32",     "image_data_format": "channels_last",     "backend": "theano" } You can also define the environment variable KERAS_BACKEND and this will override what is defined in your config file : $ KERAS_BACKEND=theano python -c "from keras import backend" Testing Quick test: udooer@udoo:~$ python Python 2.7.6 (default, Oct 26 2016, 20:46:32) [GCC 4.8.4] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import keras Using Theano backend. >>>  Test 2: Be aware this test take some time (~1hr on udoo dual): $ curl -sSL -k https://github.com/fchollet/keras/raw/master/examples/mnist_mlp.py | python Output: For demonstration, deep-learning-models repository provided by pyimagesearch and from fchollet git, and also have three Keras models (VGG16, VGG19, and ResNet50) online — these networks are pre-trained on the ImageNet dataset, meaning that they can recognize 1,000 common object classes out-of-the-box. $ cd keras $ git clone https://github.com/fchollet/deep-learning-models $ Cd deep-learning-models $ ls -l Notice how we have four Python files. The resnet50.py , vgg16.py , and vgg19.py  files correspond to their respective network architecture definitions. The imagenet_utils  file, as the name suggests, contains a couple helper functions that allow us to prepare images for classification as well as obtain the final class label predictions from the network Classify ImageNet classes with ResNet50 ResNet50 model, with weights pre-trained on ImageNet. This model is available for both the Theano and TensorFlow backend, and can be built both with "channels_first" data format (channels, height, width) or "channels_last" data format (height, width, channels). The default input size for this model is 224x224. We are now ready to write some Python code to classify image contents utilizing  convolutional Neural Networks (CNNs) pre-trained on the ImageNet dataset. For udoo Quad/Dual use ResNet50 due to avoid space conflict. Also we are going to use ImageNet (http://image-net.org/) that is an image database organized according to the WordNet hierarchy, in which each node of the hierarchy is depicted by hundreds and thousands of images. from keras.applications.resnet50 import ResNet50 from keras.preprocessing import image from keras.applications.resnet50 import preprocess_input, decode_predictions import numpy as np   model = ResNet50(weights='imagenet')   #for this sample I download the image from: http://i.imgur.com/wpxMwsR.jpg  img_path = 'elephant.jpg' img = image.load_img(img_path, target_size=(224, 224)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x)   preds = model.predict(x) # decode the results into a list of tuples (class, description, probability) # (one such list for each sample in the batch) print('Predicted:', decode_predictions(preds, top=3)[0]) Save the file an run it. Results for elephant image: Top prediction was 0.8890 for African Elephant Testing with this image: http://i.imgur.com/4FIOwAN.jpg Results: Top prediction was: 0.7799 for golden_retriever. Now your Udoo is ready to use Keras and Theano as Deep Learning libraries, next time we are going to show some usage example for image classification models with OpenCV. References: GitHub - fchollet/keras: Deep Learning library for Python. Runs on TensorFlow, Theano, or CNTK.  GitHub - Theano/Theano: Theano is a Python library that allows you to define, optimize, and evaluate mathematical expres…  GitHub - fchollet/deep-learning-models: Keras code and weights files for popular deep learning models.  Installing Keras for deep learning - PyImageSearch 

Patch for i.MX6 boards with LPDDR2 using single channel

This white paper is a discussion of random hangs and other issues using Windows Embedded Compact on Freescale i.MX6 application processor and how they were solved. All information in this document applies to Windows Embedded Compact 7 and 2013 as well as all variants of the i.MX6.      

i.MX 6 SoloX MCC MCC is a library for lightweight communication between cores Configured at compile time (mcc_config.h) Current version (2.0) is not backward compatible MCC works with shared memory area Communication is performed through ‘send’ and ‘receive’ functions Shared memory Cores communicate through shared memory Core structures of the MCC  MCC_ENDPOINT core Identifies a processor core, A9 is 0 and M4 is 1. node In Linux, each process using MCC has its own node number. MQX has only one node number. port Any number of ports per node is allowed. Value MCC_RESERVED_PORT_NUMBER is not allowed. The number of endpoints in system is defined during compile time by macro MCC_ATTR_MAX_RECEIVE_ENDPOINTS. MCC_BOOKEEPING_STRUCT Endpoint table I. MCC API  Functions Standard API −int mcc_initialize(MCC_NODE); −int mcc_destroy(MCC_NODE); −int mcc_create_endpoint(MCC_ENDPOINT*, MCC_PORT); −int mcc_destroy_endpoint(MCC_ENDPOINT*); −int mcc_send(MCC_ENDPOINT*, MCC_ENDPOINT*, void*, MCC_MEM_SIZE, unsigned int); −int mcc_recv(MCC_ENDPOINT*, MCC_ENDPOINT*, void*, MCC_MEM_SIZE, MCC_MEM_SIZE*, unsigned int); −int mcc_msgs_available(MCC_ENDPOINT*, unsigned int*); −int mcc_get_info(MCC_NODE, MCC_INFO_STRUCT*); MCC_SEND_RECV_NOCOPY_API_ENABLED −int mcc_get_buffer(void**, MCC_MEM_SIZE*, unsigned int); −int mcc_send_nocopy(MCC_ENDPOINT*, MCC_ENDPOINT*, void*, MCC_MEM_SIZE); −int mcc_recv_nocopy(MCC_ENDPOINT*, MCC_ENDPOINT*, void**, MCC_MEM_SIZE*, unsigned int); −int mcc_free_buffer(void*);

This PDF is training material for showing examples on video encoding, video decoding, video streaming on an i.MX53QSB board.

[中文翻译版] 见附件   原文链接： https://community.nxp.com/docs/DOC-343054 

We will cross compile L5.4.3_1.0.0 on ubuntu18.04 LTS platform in three steps. 1. Installing Ubuntu 18.04 to vmplayer14 & Configuring Ubuntu 18.04 LTS 2. Compiling L5.4.3_1.0.0 BSP According to Steps in i.MX_Yocto_Project_User's_Guide.pdf 3. Handling tensorflow-lite compilation errors. --BEGIN No.1 Installing Ubuntu 18.04 to vmplayer14 & Configuring Ubuntu 18.04 LTS 1. Install 18.04 on vmplayer After installation is done, root user should be set at first. # sudo passwd root Then follow these steps to configuration ubuntu 18.04 for environment of compiliation --Changing sources of ubuntu 18.04 mirror If you are Chinese users, you can do the step, which can improve your system performance. # sudo cp /etc/apt/sources.list /etc/apt/sources.list.bak # sudo geit /etc/apt/source.list Comment I.MX customers outside China do not need to modify Ubuntu source list, or can modify it to local mirror site of Ubuntu 18.04, which can improve the speed of software upgrade. Delet all sources and copy following lines here, Then save it and exit Update ubuntu source deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic main restricted universe multiverse # deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic main restricted universe multiverse deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic-updates main restricted universe multiverse # deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic-updates main restricted universe multiverse deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic-backports main restricted universe multiverse # deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic-backports main restricted universe multiverse deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic-security main restricted universe multiverse # deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic-security main restricted universe multiverse # deb https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic-proposed main restricted universe multiverse # deb-src https://mirrors.tuna.tsinghua.edu.cn/ubuntu/ bionic-proposed main restricted universe multiverse Then running these 2 commands to update sources and packages # sudo apt-get update # sudo apt-get upgrade 2. Installing packages for compiliation Packages For Compiling BSP # sudo apt-get install flex bison gperf build-essential zlib1g-dev lib32ncurses5-dev x11proto-core-dev libx11-dev lib32z1-dev libgl1-mesa-dev tofrodos python-markdown libxml2-utils xsltproc # sudo apt-get install uuid-dev:i386 liblzo2-dev:i386 gcc-multilib g++-multilib subversion openssh-server openssh-client uuid uuid-dev zlib1g-dev liblz-dev lzop  liblzo2-2 liblzo2-dev git-core curl # sudo apt-get install u-boot-tools mtd-utils android-tools-fsutils openjdk-8-jdk device-tree-compiler aptitude libcurl4-openssl-dev nss-updatedb # sudo apt-get install chrpath texinfo gawk cpio diffstat No.2 Compiling L5.4.3_1.0.0 BSP According to Steps in i.MX_Yocto_Project_User's_Guide.pdf          The detailed BSP compilation steps are given in i.MX_Yocto_Project_User's_Guide.pdf. Here are just some tips and simple steps. For Chinese customers, the system may not respond when the repo init or repo sync command is executed after obtaining the repo script. Detailed description is below: (1)Getting repo script # mkdir ~/bin (this step may not be needed if the bin folder already exists) # curl https://storage.googleapis.com/git-repo-downloads/repo > ~/bin/repo # chmod a+x ~/bin/repo # export PATH=~/bin:$PATH (2)Getting Yocto source # mkdir imx-yocto-bsp # cd imx-yocto-bsp # repo init -u https://source.codeaurora.org/external/imx/imx-manifest -b imx-linux-zeus -m imx-5.4.3-1.0.0.xml # repo sync          After running repo init or repo sync, you may encounter a situation where there is no progress for a long time, and repo sync stops there without any prompt. This may be caused by googlesource site being inaccessible.          At this time, you can modify the googlesource URL to the URL of Tsinghua University. # gedit ./repo          Changing REPO_URL = 'https://gerrit.googlesource.com/git-repo'          To be:                    REPO_URL = 'https://mirrors.tuna.tsinghua.edu.cn/git/git-repo'          Save and exit. Then run “repo init -u https://source.codeaurora.org/external/imx/imx-manifest -b imx-linux-zeus -m imx-5.4.3-1.0.0.xml”. or repo sync No.3 Handling tensorflow-lite compilation errors       If you encounter a tensorflow-lite compilation error during the compilation process, the following processing methods may be helpful to you. Cause Analysis:          Compiling tensorflow-lite requires three steps: fetching packages, configuring environment, and compiling code.          During the configuring environment, the following 8 software packages need to be downloaded. If any of the 8 files are not successfully downloaded, an error will be reported and compilation will be terminated.          ①9f48e814419e.tar.gz          ②38ebac7b059e84692f53e5938f97a9943c120d98.zip          ③release-1.8.0.tar.gz          ④389ec3f906f018661a5308458d623d01f96d7b23.tar.gz          ⑤master.zip          ⑥816a4ae622e964763ca0862d9dbd19324a1eaf45.tar.gz          ⑦1f5eae5d6a135ff6811724f6c57f911d1f46bb15.tar.gz          ⑧fft.tgz          If you encounter this problem, you can refer to the following methods to solve it. Open the file with editor: ~/imx-yocto-bsp/build-xwayland/tmp/work/aarch64-poky-linux/tensorflow-lite/1.13.2-r0/git/tensorflow/lite/tools/make/download_dependencies.sh URLs EIGEN_URL="https://bitbucket.org/eigen/eigen/get/9f48e814419e.tar.gz" GEMMLOWP_URL="https://github.com/google/gemmlowp/archive/38ebac7b059e84692f53e5938f97a9943c120d98.zip" GOOGLETEST_URL="https://github.com/google/googletest/archive/release-1.8.0.tar.gz" ABSL_URL="https://github.com/abseil/abseil-cpp/archive/389ec3f906f018661a5308458d623d01f96d7b23.tar.gz" NEON_2_SSE_URL="https://github.com/intel/ARM_NEON_2_x86_SSE/archive/master.zip" FARMHASH_URL="https://github.com/google/farmhash/archive/816a4ae622e964763ca0862d9dbd19324a1eaf45.tar.gz" FLATBUFFERS_URL="https://github.com/google/flatbuffers/archive/1f5eae5d6a135ff6811724f6c57f911d1f46bb15.tar.gz" FFT2D_URL="http://www.kurims.kyoto-u.ac.jp/~ooura/fft.tgz" Save and exit. [comment]     Don’t run “bitbake tensorflow-lite -c cleanall before you modify it”, otherwise files will be deleted.  Then try to compile tensorflow-lite like below: # bitbake tensorflow-lite -c compile   At the same time, we can open this directory and check if the following directories are created. If these 8 directories are successfully created, it means that links we modified are no problem.          After tensorflow-lite is successfully built, we can continue to compile full images. # bitbake imx-image-full --END NXP TIC team Weidong Sun 2020/3/30

Q: When trying to mount his SDIO WiFi module (an Azurewave module containing a Marvel 88W8790)  in 4 bit mode, and got error wifi module on iMX6 Smart SD dev board using the alpha kernel (3.5.7+3285970). mwifiex_sdio mmc0:0001:1: WLAN FW is active mwifiex_sdio mmc0:0001:1: mwifiex_cmd_timeout_func: Timeout cmd id (2004.239824) = 0xa9, act = 0x0 mwifiex_sdio mmc0:0001:1: num_data_h2c_failure = 0 mwifiex_sdio mmc0:0001:1: num_cmd_h2c_failure = 0 mwifiex_sdio mmc0:0001:1: num_cmd_timeout = 1 mwifiex_sdio mmc0:0001:1: num_tx_timeout = 0 mwifiex_sdio mmc0:0001:1: last_cmd_index = 1 mwifiex_sdio mmc0:0001:1: last_cmd_resp_index = 0 mwifiex_sdio mmc0:0001:1: last_event_index = 0 mwifiex_sdio mmc0:0001:1: data_sent=1 cmd_sent=1 mwifiex_sdio mmc0:0001:1: ps_mode=1 ps_state=0 +++++++++++++++++++++++ There are only two known issues with the SDHC driver as noted in the release notes and these don't seem to match.  The Linux Reference Manual states the operation of the SDIO was veriifed usign the AR6003.  My assumption is that this is the Silex module we have standardized on.  Was 4 bit mode verified using this module? Does anyone have any idea what could be happening here? A: L3.5.7 is an alpha release, and WiFi function is not stable. The test report shows that open WiFi issue exists. Wifi module sometime can't work on mx6q_smd. After insmod ath, ath6kl_core and ath6kl_sdio. Insert the wifi card, sometime(50%) will display -------------------------- ath6kl: unable to read RX_LOOKAHEAD_VALID                                      ath6kl: Unable to recv target info: -84                                        ath6kl: Failed to init ath6kl core                                             ath6kl_sdio: probe of mmc0:0001:1 failed with error -84    -------------------------- reinsert the wifi card may solve the problem. when execute "udhcpc -i wlan0", sometime (30%) will cause program exception. Sometime can't get the wlan0 ip(program hang). Environment(OS,Platform,Driver, etc): HW: MX6Q_SMD Num014 and Num017 MX6Q_ARD don't have this problem SW: Kernel 3.5.7-1.0.0 GNU/Linux Case ID: TGE-LV-WIFI-0043 Reproduce Steps: #modprobe ath #modprobe ath6kl_core #modprobe ath6kl_sdio insert the wifi card # iwconfig wlan0 mode managed # iwlist wlan0 scanning | grep MAD-wifi #iwconfig wlan0 key 00112233445566778899123456 #iwconfig wlan0 essid MAD-wifi #udhcpc -i wlan0 Attached MX6 ARD WiFi issue also. [Kernel3.5.7_MX6QARD]Wifi:wifi card can't work during suspend and resume. 100% -- Bug detailed description: With wifi card inserted in the board. doing system suspend and resume test. wifi card can't work well after the system suspend and resume. Always report : ath6kl: Unable to decrement the command credit count register: -84             ath6kl: Unable to write to the device: -84                                     ath6kl: bmi_write_memory for uart debug failed                                 ath6kl: Failed to boot hw in resume: -5          Environment(OS,Platform,Driver, etc): HW: MX6QARD -023 Only tried on this platform SW: root@imx6qsabreauto:~# uname -a                                                Linux imx6qsabreauto 3.5.7-1.0.0+3285970 #1 SMP PREEMPT Sat Jun 29 10:20:45 CDT 2013 armv7l GNU/Linux Case ID:  TGE-LV-WIFI-1060 and TGE-LV-WIFI-1062 Reproduce Steps: 1. boot the kernel with wif card inserted 2. doing wifi stress test 3. doing suspend and resume

i.MX6X_内核驱动代码与定制_V2-20150518.doc: 3.0.35

Hello Linux and i.MX6 users Tushar has posted on the Element14 community some clear tutorials (accessible following the link below) to create a bootable SD-Card from a Linux image. Linux host:             http://www.element14.com/community/community/designcenter/single-board-computers/riotboard/blog/2014/07/31/upgrading-riotboard-with-kernel-31017-linux-host Windows host:      RIoTboard: Upgrading RIoTboard with Kernel 3.10... | element14 Happy SD-Card Programming Greg

There are two optionsou 1. Remove all source code after building: You can indicate bitbake to remove all source code after a build adding this line INHERIT += "rm_work" our to your conf/local.conf file. Of course, when baking another image, ALL source code has to be fetched again and it can take a lot of time, specially on machines with low connections to the Internet. The smallest image you can build is core-image-minimal, containing the minimal set of packages to board a particular machine. The largest is fsl-image-gui (in fact, there is one bigger 'fsl-image-gui-sdk' used for developing purposes inside the target). Between these two, there are a lot of images you can choose. You can even create your own image using hob. In case you do not want to build anything, there are some pre-built images on this site, just download the sdcard file and flash it to an SD card. 2. Select only those  IMAGE_FSTYPES you need. There is a variable named IMAGE_FSTYPES  which indicates the output formats you want for the resulting root file system images, by default is has this value: IMAGE_FSTYPES ?= "tar.bz2 ext3 sdcard" and it is located on /sources/meta-fsl-arm/conf/machine/include/imx-base.inc machine's header file. If you want to change it, place this variable on the build/conf/local.conf file. In case you have NFS mounting, remove the 'sdcard' (and either 'ext3' or 'tar.bz2') string, resulting in IMAGE_FSTYPES = "tar.bz2"

In the i.MX51 default WINCE6  release, the eCSPI doesn't support multiple bursts mode and set the wait states. Attached was the document and code for how to enable the multiple bursts mode and how to set the wait states between two burst.

We have a ATK tool which can program image, also it can burn fuse for i.MX51. Since fuse is one time program, so please take care the fuse can't be turn back after programmed.

Very often, customers need to disable the framebuffer auto blank. There are several ways to do so. Modify the code in drivers/tty/vt/vt.c or remove the  “CONFIG_VT_CONSOLE” from the kernel configuration. All these work, but inconvenient. Review the code following: static int vesa_blank_mode; /* 0:none 1:suspendV 2:suspendH 3:powerdown */ static int vesa_off_interval; static int blankinterval = 10*60; core_param(consoleblank, blankinterval, int, 0444); The blank is controlled by blankinterval, which can be set with the name consoleblank. And the consoleblank is a "core_param". core_param in linux can be recognized by kernel. Also can be passed to kernel command line from uboot with bootargs. we could add this to the bootargs that the framebuffer will not go blank: consoleblank=0 Example(verified with imx6 L3.0.35_4.1.0_130816): setenv bootargs_mmc 'setenv bootargs ${bootargs} root=/dev/mmcblk1p1 rootfstype=ext4 rootwait video=mxcfb0:dev=hdmi,1920x1080M@60,if=RGB24,bpp=32 fbmem=28M consoleblank=0'

View Webinar Recording

Brief introduction on the aarch64 linux kernel memory mapping layout and basic management stuffs.  Contents include: Kernel's virtual memory layout and mapping after running i.MX8QM/QXP kernel reserved memory layout Kernel memory allocation method and technology (Buddy, cma, ION...) DMA buffer management, SWIOTLB, IOMMU GPU memory management How to customize the memory for different use cases How to avoid using CMA for a better stability and performance

These questions and answers are about interrupt generation at a dedicated (configurable) video output port. The i.MX6D manual (Rev. 0) Image Processing Unit (IPU) chapter mentions: Every DI has 10 timing generator counters. The IPU Interrupt Generator has 10 DI0 counters (1...10) and just 2 DI1 counters (3 & 😎 as interrupt sources. The Interrupt Control Register lists 11 DI0 counters (0...10) Q1. Are the DI timing-generator counters linked to the counters in the interrupt controller, or are they different counters? A1. Yes, the DI timing generator counters are linked to the counters in the interrupt controller. Q2. Why are there 11 counters listed in the interrupt controller, but just 10 counters in the timing generator? A2. There is disp_clk_en_pre in the interrupt controller. Thus the 11 counters: 10 timing generator counters and 1 disp clock generator counter. Q3. Is configurable timing feasible for DI0 by using the timing generator counters? A3. Yes, using the 10 internal timing counters you can generate various timing relationships. In addition, you can detect any of the interrupt counters. For example, if you use counter 8, then you can detect the interrupt associated with counter 8. Q4. Explain the impact of the DI1 counter access of only channels 3 and 8. A4. DI1 also has 10 timing generator counters and 1 disp clock generator counter, which you can use to generate desired waveforms. This is similar to DI0. The difference is only 2 of the 10 counters (plus another disp_clk) are connected to the interrupt controller for DI1. Therefore, there is a restriction for detection. If you use counter 7, read out the counter 7 interrupt of DI1 is not possible. However, 2 channels should be sufficient. These interrupts are usually used to indicate a frame start or a frame end. We usually use counter 3 to represent Vsync. So normally we only use counter 3 interrupt. DI1 has only 3 accesses because this covers the anticipated use case and the desire was to restrict register size. The extra counters facilitate flexible DI1 timing generation.