Allwinner T113-S3 Getting Started Information Summary (Avoidance Guide) Part 1

aleksib

Allwinner T113-S3 Getting Started Information Summary (Avoidance Guide) Part 1 [Copy link]

 

I bought a Mango Pi T113 board out of nothing. I had worked on Linux before and my work was also related to Linux. I thought I could quickly set up a development environment, but it took me two weekends. So I organized the information and listed the key points. I hope you can avoid detours.  **Compilation environment** For the compilation environment, I basically refer to Mango Pi's github: [https://github.com/mangopi-sbc/Tina-Linux](https://github.com/mangopi-sbc/Tina-Linux) But I got stuck for a long time. I tried Ubuntu virtual machine, docker, and various environments repeatedly but failed. The reason was that github was too stuck, which made it impossible to download the complete project. Then I got various errors when compiling. If you don't want to use network disk files (in fact, network disk is recommended for beginners), but you have to do the same as me, here are some suggestions: 1. Be sure to use the Ubuntu 18.04 version, not other versions. 2. Try to use a virtual machine instead of docker, but it is said that docker can be successful. 3. The four files that need wget are very large, so you can download them separately and put them outside the Tina-Linux directory for backup. In this way, if git clone fails, you can copy them in again, saving time for each download. 4. During the git clone process, if there is a long-term jam or failure, it is recommended to delete and restart git, especially the update of the submodule git submodule update --init --recursive. Once there is a jam, delete all and restart, otherwise even if it is over, make will still report an error later. When deleting, copy the first four wget files out, and the latter will save download time. So the biggest pitfall here is that the two commands of git clone and git submodule must be seen to be executed very smoothly, rather than paused or jammed. Otherwise, the files are incomplete and make is likely to make mistakes later. Then for environment construction, there are several posts for your reference: "Mangopi MQ-R: T113-s3 compiles Tina Linux system" [https://blog.csdn.net/qq_28877125/article/details/126684892](https://blog.csdn.net/qq_28877125/article/details/126684892) "MangoPi-MQ Sparrow D1s Getting Started Tutorial" Tina system compilation, burning and basic analysis [https://whycan.com/p_72297.html](https://whycan.com/p_72297.html) "[Embedded AI] Allwinner R329 Tina system image compilation" [https://zhuanlan.zhihu.com/p/392832413](https://zhuanlan.zhihu.com/p/392832413) 《Allwinner Online Documentation》 [https://www.aw-ol.com/docs](https://www.aw-ol.com/docs) When reading the materials, I was very angry when I saw this online document. Why do R329, D1 and so on have tutorials, but T113 doesn't? Isn't it the son of the company? Hahaha, but many steps are universal, just refer to it. **Compile options** For students who have never developed Linux, commands such as git and make are enough to eat a pot, let alone something as complicated as make menuconfig. So we only need to know that after typing the make menuconfig command, the interface that appears is just to change the configuration items of our kernel. For the first time, we don't need to touch this place, just make it directly, and run it first. When compiling, just type: ```csharp make FORCE_UNSAFE_CONFIGURE=1 ``` Let's not worry about multi-process compilation for now, just make sure you can compile it. If you don't add FORCE_UNSAFE_CONFIGURE=1, you will get an error, but everything will be fine after adding it. **Burn** Burning is really a big pitfall. First of all, it is recommended that you use the ready-made firmware of Mango Pi to go through the burning and booting process to avoid failure later due to problems with the firmware you compiled yourself, and you don't know why. Ready-made firmware MQ-R-T113(RTL8189): [https://mangopi.org.cn/_media/undefined/mq-r-t113-rtl8189fs-5113-dns-uart0.zip](https://bbs.aw-ol.com/topic/2495/) MQ-R-T113(RTL8723): [https://mangopi.org.cn/_media/undefined/mq-r-t113-rtl8723ds-5135-dns-uart0.zip](https://bbs.aw-ol.com/topic/2495/) For burning, we can see that the official website video introduces several tools, such as PhoenixSuit, PhoenixCard, PhoenixUSBPro, and DragonSN, which is a burning tool. I am confused. Later I figured out that there is no SPI NAND FLASH soldered on the Mango Pi T113 board, so these mass production tools are not needed at all. I only need PhoenixCard to burn the system, and the others are not needed. Don't take detours. PhoenixCard download address: [https://www.aw-ol.com/downloads?cat=5](https://www.aw-ol.com/downloads?cat=5) PhoenixCard is a very simple tool. It divides the TF card into several areas according to the image configuration we packaged, and then writes the corresponding boot and rootfs into the corresponding areas. After that, unplug the TF card from the PC card reader, insert it into the card slot of T113, and you can start Tina. PhoneixCard video tutorial: (jump to 11:25 to start watching, the rest is useless) [https://open.allwinnertech.com/#/video/open?menuID=17&videoId=380](https://open.allwinnertech.com/#/video/open?menuID=17&videoId=380) There are a few key points about PhoenixCard: 1. Remember to select the startup card option, not the mass-produced card. 2. If you use a card produced by PhoenixCard and don’t want to use it for TIna later, you should use this tool to restore the TF card, otherwise the computer will not recognize it. **Startup** Tina’s startup process is more complicated. If you are interested, you can watch the following video: [https://open.allwinnertech.com/#/video/open?menuID=17&videoId=297](https://open.allwinnertech.com/#/video/open?menuID=17&videoId=297) If you are too lazy to watch the video, you can just listen to me. In fact, the startup of the Linux system is the same. Start uboot first. You can understand that uboot is a startup program that is responsible for cleaning the house and waiting for the Linux system to come in. We need to use the serial port to connect E2, E3, and GND on the board. You can use the usb-ttl tool to connect, so you can see the information output by the T113 serial port on the computer through usb-ttl. Why are the two pins E2 and E3 acting as RX and TX? You can see it by looking at the circuit diagram. These two pins are the default serial port 0 of T113, and the system prints and receives from here. When the system starts, we keep typing the 2 key on the keyboard on the serial terminal to stop it at the uboot step, and we can send uboot instructions to pull up the kernel and so on. If you are interested, you can study it yourself. There are a lot of uboot tutorials on the Internet. In Tina's uboot, there is an efex command, which can put the system into mass production state, so that the PhoenixSuit tool can be used to mass produce firmware. However, this mass production state is prepared for spi nand flash. There is no spi nand flash on the Mango Pi board, so this mass production state is useless to us. We use SD card to boot. At this point, I didn't understand it at first, and various mass production failed until later. It was a pitfall. Of course, there is a soldering pad for spi nand flash on Mango Pi. You can buy one and solder it on yourself. I haven't had time to check the model. If you have tried it, please leave me a message and I will add it. In Tina's boot process, there is a boot0 before uboot. This boot0 is used to pull up uboot. You can understand that building a suspension bridge between two canyons is to pull a kite line first, then use the kite line to pull a thicker rope, and then use the thick rope to pull the steel cable. Boot0 pulls uboot, and uboot pulls tina, which is actually the same principle. The reason why there are so many layers is mainly because we are afraid that a certain process will be burned incorrectly and the board will become bricked. Because boot0 exists and is fixed in the CPU, T113 will never become bricked, which is probably what it means. But because we are booting from an SD card, we can ignore the above and just play with our own. **Connecting to wifi** As long as the system can start normally, connecting to wifi is relatively simple, just a few commands, you can refer to the official information of R329: https://r329.docs.aw-ol.com/devboardstudy/r329evb5studywifi/ Search for wifi: wifi_scan_results_test Connect to wifi: wifi_connect_ap_test wifi name and password (parameters have spaces) After connecting, you can use the ifconfig command to view your ip address, or you can use the ping command to test connectivity. **PC driver** After the board is started, I feel that there are two types of drivers. If it is in mass production, there is a device driver corresponding to mass production, which we can ignore. But if you enter the Tina system, there will be an ADB driver that you need to install, so that you can use the ADB tool to push our compiled helloword to the board later. Driver download address: https://www.aw-ol.com/downloads?cat=5 Download the PhoeniSuit tool, there is a Drivers directory, there is an ADB driver, plug the board into the PC, the device manager pops up an exclamation mark, and then you can install it yourself. "ADB (tab auto-complete version)" is also downloaded, and it will be used to interact with the board later. Summary That's all I'll write today, and it's also a summary of my two weeks of playing with T113. I hope you can take fewer detours. In fact, I have played with many boards, including my current work, which is also running Linux on Qualcomm and Realtek platforms, including stm32, FreeRTOS and lwip. I have been playing for more than ten years. Why did I choose T113? I think there are several reasons: 1. I need a board to help me verify small Linux applications sometimes. I think the Raspberry Pi is too big. 2. T113 is not a BGA package. If you want to do some open source projects yourself, the pressure on the PCB is relatively small. 3. To be honest, AI and machine learning are far away from me. These things are only affordable by big companies. For our personal learning, unless it is in this industry, there is no need to join in the fun. It is exhausting to build a learning model. After it is done, I don’t know what I can do. The board is left to gather dust. My learning focus is still on doing some open source Internet of Things applications, watering flowers remotely, counting the reported environmental data, and building some remote servers. These are all practical things, and I feel they are closer to practicality. I will add cross-compilation and partitioning, io operations later when I have time. I am also familiar with them. I wish you all a happy time! \com/downloads?cat=5 Download the PhoeniSuit tool, which has a Drivers directory with ADB drivers. Plug the board into the PC, and the device manager will pop up an exclamation mark, then you can install it yourself. Download "ADB (tab auto-complete version)" as well, and use it to interact with the board later. Summary That's all I'll write today, and it's also a summary of my two weeks of playing with T113. I hope you can avoid detours. In fact, I've played with a lot of boards, including my current work, which also runs Linux on Qualcomm and Realtek platforms, including stm32, FreeRTOS and lwip. I've been playing with them for more than ten years. Why did I choose T113? I think there are several reasons: 1. I need a board to help me verify small Linux applications sometimes. I think the Raspberry Pi is too big. 2. T113 is not a BGA package. If you want to do some open source projects yourself, the pressure on the PCB is relatively small. 3. To be honest, AI and machine learning are far away from me. These things are only affordable by big companies. For our personal learning, unless it is in this industry, there is no need to join in the fun. It is exhausting to build a learning model. After it is done, I don’t know what I can do. The board is left to gather dust. My learning focus is still on doing some open source Internet of Things applications, watering flowers remotely, counting the reported environmental data, and building some remote servers. These are all practical things, and I feel they are closer to practicality. I will add cross-compilation and partitioning, io operations later when I have time. I am also familiar with them. I wish you all a happy time! \com/downloads?cat=5 Download the PhoeniSuit tool, which has a Drivers directory with ADB drivers. Plug the board into the PC, and the device manager will pop up an exclamation mark, then you can install it yourself. Download "ADB (tab auto-complete version)" as well, and use it to interact with the board later. Summary That's all I'll write today, and it's also a summary of my two weeks of playing with T113. I hope you can avoid detours. In fact, I've played with a lot of boards, including my current work, which also runs Linux on Qualcomm and Realtek platforms, including stm32, FreeRTOS and lwip. I've been playing with them for more than ten years. Why did I choose T113? I think there are several reasons: 1. I need a board to help me verify small Linux applications sometimes. I think the Raspberry Pi is too big. 2. T113 is not a BGA package. If you want to do some open source projects yourself, the pressure on the PCB is relatively small. 3. To be honest, AI and machine learning are far away from me. These things are only affordable by big companies. For our personal learning, unless it is in this industry, there is no need to join in the fun. It is exhausting to build a learning model. After it is done, I don’t know what I can do. The board is left to gather dust. My learning focus is still on doing some open source Internet of Things applications, watering flowers remotely, counting the reported environmental data, and building some remote servers. These are all practical things, and I feel they are closer to practicality. I will add cross-compilation and partitioning, io operations later when I have time. I am also familiar with them. I wish you all a happy time! \

damiaa

The OP is thoughtful!