Jongkuk Lim

Using fixed tty device path on Ubuntu

2023-06-28T00:00:00+09:00

Why fix the tty device path?

Usually, you don’t need this but if you are working on a project that uses an embedded device with some peripherals, it’s nice to have a fixed device path since /dev/ttyUSB0 or /dev/ttyUSB1 can be any device.

Wouldn’t it be better if we have /dev/ttyMyPort0 always be the device 0 and /dev/ttyMyPort1 be the device 1?

How do I do this

Finding the unique attribute

We need to tell Linux that a specific device must go to /dev/ttyMyPort0. And the specific device can be distinguished by the following command.

udevadm info -a -n /dev/ttyUSB0

user@ubuntu:~$ udevadm info -a -n /dev/ttyUSB0

Udevadm info starts with the device specified by the devpath and then
walks up the chain of parent devices. It prints for every device
found, all possible attributes in the udev rules key format.
A rule to match, can be composed by the attributes of the device
and the attributes from one single parent device.

  looking at device '/devices/3610000.xhci/usb1/1-2/1-2.1/1-2.1:1.0/ttyUSB0/tty/ttyUSB0':
    KERNEL=="ttyUSB0"
    SUBSYSTEM=="tty"
    DRIVER==""

  looking at parent device '/devices/3610000.xhci/usb1/1-2/1-2.1/1-2.1:1.0/ttyUSB0':
    KERNELS=="ttyUSB0"
    SUBSYSTEMS=="usb-serial"
    DRIVERS=="ftdi_sio"
    ATTRS{latency_timer}=="16"
    ATTRS{port_number}=="0"

  looking at parent device '/devices/3610000.xhci/usb1/1-2/1-2.1/1-2.1:1.0':
    KERNELS=="1-2.1:1.0"
    SUBSYSTEMS=="usb"
    DRIVERS=="ftdi_sio"
    ATTRS{authorized}=="1"
    ATTRS{bAlternateSetting}==" 0"
    ATTRS{bInterfaceClass}=="ff"
    ATTRS{bInterfaceNumber}=="00"
    ATTRS{bInterfaceProtocol}=="ff"
    ATTRS{bInterfaceSubClass}=="ff"
    ATTRS{bNumEndpoints}=="02"
    ATTRS{interface}=="Quad RS232-HS"
    ATTRS{supports_autosuspend}=="1"

  ...
  ...
  ...

This will tell us device information such as which port is connected to the device. Try this command with different device paths. You will see some parameters are different from the others. In my case, I usually use KERNELS==1-2.1:1.0 which tells me that this device is connected to 1-2.1:1.0 physical port.

Tell Linux to use the fixed path that I want

Once we have a unique attribute of the device, it’s time to tell Linux that I want to use this device to be this path. Open /etc/udev/rules.d/99-usb-serial.rules with root permission and add something like the below.

KERNEL=="ttyUSB*",KERNELS=="1-2.1:1.0",SYMLINK+="ttyTelem"

KERNEL and KERNELS are the unique attributes of /dev/ttyUSB0 device. And SYMLINK tells Linux to make a symbolic link of /dev/ttyUSB0 to /dev/ttyTelem. You can add more attributes prior to SYMLINK to make a better separation.

Reboot your system or use below the commands to apply these changes immediately.

sudo udevadm control --reload-rules
sudo udevadm trigger

See if /dev/ttyTelem exists and if it is linked to /dev/ttyUSB0. If it doesn’t show up, the unique attributes might not be unique.

FFMPEG cheat sheets

2022-03-07T00:00:00+09:00

1. Video Conversion

1.1 Basic

1.1.1. MOV to MP4

ffmpeg -i /source/path /output/path

1.1.2. Change FPS

ffmpeg -i <input> -filter:v fps=fps=30 <output>

1.1.3. Scale Change

ffmpeg -i <input> -vf scale=320:240 <output>

ffmpeg -i <input> -vf scale=320:-1 <output>

1.1.4. No Audio

ffmpeg -i <input> -an <output>

1.1.5 Crop Video

# ffmpeg -i in.mp4 -filter:v "crop=out_w:out_h:x:y" out.mp4

# Specific Location
ffmpeg -i <input> -filter:v "crop=200:100:300:100" <output>

# Crop Only Top of specific pixel
ffmpeg -i <input> -filter:v "crop=iw:ih-20:0:ih" <output>

1.1.6 Trim Video

# Trim with duration
ffmpeg -i input.mp4 -ss 00:05:20 -t 00:10:00 output.mp4

# Trim with specific time
ffmpeg -i input.mp4 -ss 00:05:10 -to 00:15:30 output.mp4

# Trim last 10 minutes
ffmpeg -sseof -00:10:00 -i input.mp4 output.mp4

1.2 Advanced

1.2.1 비디오 재생 속도 조절하기

ffmpeg -i input.mkv -filter:v "setpts=0.5*PTS" output.mk

(0.5 - 조절하는 속도 (2배속))

Audio 가 있는 경우

ffmpeg -i input.mkv -filter_complex "[0:v]setpts=0.5*PTS[v];[0:a]atempo=2.0[a]" -map "[v]" -map "[a]" output.mkv

(0.5 - 영상 속도 (2배속)) (2.0 - 오디오 속도 (2배속))

1.2.2 Change speed, resolution, and FPS at the same time

ffmpeg -i input.MOV -filter_complex "fps=60,setpts=0.1*PTS,scale=640:-1;atempo=10.0" output.mp4

1.2.3 Video stack vertically

ffmpeg -i input0 -i input1 -filter_complex vstack=inputs=2 output

1.2.4 Video stack horizontally

ffmpeg -i input0 -i input1 -filter_complex hstack=inputs=2 output

1.2.5 Video stack with a border

ffmpeg -i input0 -i input1 -filter_complex "[0]pad=iw+5:color=black[left];[left][1]hstack=inputs=2" output

Further details can be found here https://stackoverflow.com/questions/11552565/vertically-or-horizontally-stack-mosaic-several-videos-using-ffmpeg

1.2.6 Video 2x2 grid with text

ffmpeg -i 2019_1231_face_test.mov -i test_result_noone.mp4 -i test_result_light_DSFD.mp4 -i test_result_DSFD.mp4 -filter_complex "[0]drawtext=text='(Original)':borderw=5:bordercolor='WhiteSmoke':fontsize=100:x=w-text_w-10:y=h-text_h-20[v0]; [1]drawtext=text='(Noone video)':borderw=5:bordercolor='WhiteSmoke':fontsize=100:x=10:y=h-text_h-30[v1]; [2]drawtext=text='(lightDSFD)':borderw=5:bordercolor='WhiteSmoke':fontsize=100:x=w-text_w-10:y=20[v2]; [3]drawtext=text='(DSFD)':borderw=5:bordercolor='WhiteSmoke':fontsize=100:x=10:y=20[v3]; [v0][v1][v2][v3]xstack=inputs=4:layout=0_0|w0_0|0_h0|w0_h0[v]" -map "[v]" output_grid.mp4

1.2.7 Create Images from Video

ffmpeg -i <input> -filter:v fps=1 <output%03d.jpg>

1.2.8. Create Video from Images

ffmpeg -framerate 30 -pattern_type glob -i '*.png' -c:v libx264 -pix_fmt yuv420p out.mp4

1.2.9. Download m3u8 playlist video

ffmpeg -protocol_whitelist file,http,https,tcp,tls,crypto -i my_movie.m3u8 -c copy my_movie.mp4

1.2.10. Using GPU

ffmpeg -hwaccel cuvid -c:v h264_cuvid -i input.MOV -c:v h264_nvenc -b:v 10240k output.mp4

Use ffmpeg -hwaccels to check supported hardware

2. GIF

2.1 Basic

ffmpeg -ss 0 -t 12 -i input.mov -vf "fps=2,scale=2048:-1:flags=lanczos,split[s0][s1];[s0]palettegen[p];[s1][p]paletteuse" -loop 0 output.gif

ss - 시작 지점

t 끝나는 지점

fps - 프레임 속도

scale = 가로 해상도:세로 해상도

2.2 Multiple Images to gif

ffmpeg -f image2 -i %02d.png -vf scale=900:-1:sws_dither=ed,palettegen palette.png
ffmpeg -f image2 -framerate 2 -i %02d.png video.flv
ffmpeg -i video.flv -i palette.png -filter_complex "fps=2,scale=430:-1:flags=lanczos[x];[x][1:v]paletteuse" out.gif

rm video.flv palette.png

2.3 Lossy GIF

https://github.com/kohler/gifsicle

brew install gifsicle

gifsicle -O9 --lossy=200 -o output.gif input.gif
# 200 is highest

How to build IPC communication using Protobuf and gRPC

2022-03-07T00:00:00+09:00

Some projects aren’t enough using a single programming language. When in comes to the project that uses multiple languages, communicating between different languages become a problem. There are (of course) many ways to implement IPC communication but Google’s protobuf seems to be a reasonable choice unleess I have a special purpose in mind.

Basic workflow (How does it work?)

Protobuf is a language-neutral serialization solution. Language-neutral means that you need to define messages first no matter which language you use. It looks like we are doing another work but it has a certain advantages. Be patient.

Basically, you define message .proto file.

message Person {
  optional string name = 1;
  optional int32 id = 2;
  optional string email = 3;
}

And turn this .proto file into your language API using protoc

Person john = Person.newBuilder()
    .setId(1234)
    .setName("John Doe")
    .setEmail("jdoe@example.com")
    .build();
output = new FileOutputStream(args[0]);
john.writeTo(output);

TA-DA. Serialized Person instance has been written to output stream! It looks simple enough. Let’s take a closer look how we can apply Protobuf into the project.

Install protobuf

Protobuf requires to be installed prior to use it. You can go to https://developers.google.com/protocol-buffers/docs/downloads to find an installation package. If you are planning to use C++, you will need to build from the source. Check C++ installation page in protobuf GitHub page.

It turns out I didn’t need to install protobuf separately. gRPC installs protobuf since it’s based on protobuf serialization.

Install gRPC

Wait, what is gRPC? My original plan was to serialize messages via protobuf and send the serialized data into socket communication which will work as IPC communication. But why reinvent wheels? gRPC was designed to call a method on a server from the client directly which its purpose exactly matches with what IPC does.

You can follow instructions on how to install gRPC in https://grpc.io/docs/languages/cpp/quickstart/

Write .proto for gRPC ready.

I have introduced simple Person message definition above. Let’s expand it to IPC communication ready. Here is .proto message definition.

syntax = "proto3";

package protobuf_example;

service WhoIsIt {
    rpc SendMe (PersonRequest) returns (Person) {}
}

message Person {
  string name = 1;
  optional int32 id = 2;
  optional string email = 3;
}

message PersonRequest {
    string name = 1;
}

In this definition, I have defined WhoIsIt service which takes control of server and client side communication via RPC(Remote Procedure Call). SendMe method receieves PersonRequest message and responses with Person message. The structures of messages are defined as we have written in .proto file.

Generate `.cc` and `.h` file (And `.py` file)

Once we have written the .proto file, we are ready to write language-dependent files. We assume that our message definition .proto file is msg.proto.

protoc --cpp_out ./cxx --grpc_out ./cxx --plugin=protoc-gen-grpc="/home/user/.local/bin/grpc_cpp_plugin" msg.proto

You will need to change proto-gen-grpc path along with your installation path. Once you run this command, you will have 4 files. msg.grpc.pb.cc, msg.grpc.pb.h, msg.pb.cc, msg.pb.h As you already might have noticed that protoc generates msg.pb.cc and msg.pb.h which defines how protobuf generates message serialization. And the protoc-gen-grpc plugin generates msg.grpc.pb.cc and msg.grpc.pb.h which contains communication services.

What if I want to use python language for IPC between C++ and python? You will need to generate python file as well.

protoc --python_out ./python --grpc_out ./python --plugin=protoc-gen-grpc="/home/user/.local/bin/grpc_python_plugin" msg.proto

And you will see msg_pb2_grpc.py, msg_pb2.py files has been generated. Now we are ready to write our own IPC communication program!

Write main.cpp for IPC server

I want to write an application that receieves name of the person request and send reply with email information in C++ and send request message from python language. Let start with C++ first.

#include <grpcpp/ext/proto_server_reflection_plugin.h>
#include <grpcpp/grpcpp.h>
#include <grpcpp/health_check_service_interface.h>

#include <iostream>
#include <memory>
#include <string>

#include "msg.grpc.pb.h"

using grpc::Server;
using grpc::ServerBuilder;
using grpc::ServerContext;
using grpc::Status;

using protobuf_example::Person;
using protobuf_example::PersonRequest;
using protobuf_example::WhoIsIt;

class PersonServiceImpl final : public WhoIsIt::Service {
  Status SendMe(ServerContext* context, const PersonRequest* request,
                Person* person) override {
    person->set_name(request->name());
    person->set_email("lim.jeikei@gmail.com");
    return Status::OK;
  }
};

int main(int argc, char** argv) {
  std::string server_address("0.0.0.0:50051");
  grpc::EnableDefaultHealthCheckService(true);
  grpc::reflection::InitProtoReflectionServerBuilderPlugin();

  PersonServiceImpl service;

  ServerBuilder builder;
  builder.AddListeningPort(server_address, grpc::InsecureServerCredentials());
  builder.RegisterService(&service);
  std::unique_ptr<Server> server(builder.BuildAndStart());
  std::cout << "Server listening on " << server_address << std::endl;

  server->Wait();

  return 0;
}

I have made a PersonServiceImpl class that inherits WhoIsIt::Service which was generated by protoc. Let take a look in python code now.

import grpc

from protobuf.python.msg_pb2 import Person
from protobuf.python.msg_pb2_grpc import WhoIsItStub

if __name__ == "__main__":
    with grpc.insecure_channel('localhost:50051') as channel:
        stub = WhoIsItStub(channel)
        response = stub.SendMe(Person(name="Jongkuk Lim"))

    print(response)

You will find that python prints message like below. Once you run binary file built from C++ and run python script.

name: "Jongkuk Lim"
email: "lim.jeikei@gmail.com"

You can find demo repository in https://github.com/JeiKeiLim/protobuf-ipc-example

Jekyll Cheat Sheet 모음

2022-03-02T00:00:00+09:00

Jekyll 로 블로그 글을 작성하면서 자주 쓰게 되는 Markdown 문법을 여기에 정리할 계획 입니다. 해당 페이지는 수시로 업데이트 될 예정입니다. Jekyll의 minimal-mistakes 테마를 기준으로 작성되었습니다.

Image

Image with caption

{% include figure 
    image_path="assets/images/kindle.jpg" 
    alt="Kindle" 
    caption="This is a caption of the fire image"
%}

This is a caption of the fire image

Image width adjustment

![Stack vertical](/assets/images/kindle.jpg){: width="10%"}
![Stack vertical](/assets/images/kindle.jpg){: width="25%"}
![Stack vertical](/assets/images/kindle.jpg){: width="50%"}

Text

Raw text with liquid

마지막 % } 의 공백은 붙여줘야 한다.

{% raw %} {% some syntax %} {% endraw % }

{% some syntax %}

Comments

{% comment %}
This is a markdown comment that will not be shown in rendered HTML.
{% endcomment %}

Hyperlink with _blank target

[limjk.ai](https://limjk.ai){:target='_blank'}

limjk.ai

블로깅을 다시 한번 시작해보려고 합니다.

2022-02-14T00:00:00+09:00

그동안 미뤄왔던 GitHub Pages와 Jekyll을 연동하여 블로깅을 시작하려고 합니다. 사실 블로깅 서비스야 워낙 많아서 굳이 Jekyll로 안해도 될 것 같지만 Jekyll을 선택한 이유가 있습니다.

1. Vim이 너무 편해져버렸습니다.

저도 누군가가 Vim을 쓴다고 하면 이상한 눈으로 쳐다보던 때가 있었습니다. 세상에 편한 툴이 얼마나 많은데 Vim을 쓰다니 뭐야 변태인가? 라고 생각했었죠. 그런데 어쩌다보니 Vim을 많이 써야되는 환경에 노출되고 점점 익숙해지니까 이제는 Vim이 아니면 타이핑하기 싫은 상황이 되버렸습니다. 왜 그런지 이야기하자면 너무 길어지니 더 자세한 이야기는 별도 포스팅으로 한번 작성해야겠네요.

2. Markdown은 강력 합니다.

요즘 문서를 작성할때 일반 문서보다 Markdown 으로 작성하는 일이 더 많아진 것 같네요. GitHub도 그렇고 Notion도 그렇구요. 그러다보니 아무리 WYSIWIG이 좋다고 해도 코드가 익숙한 사람 입장에서는 Markdown을 선호하게 되었습니다.

3. 한번에 글을 작성하지 말고 틈틈히 글을 쓰고 싶었습니다.

이건 사실 다른 블로깅 플랫폼에서도 할 수 있는거지요. 역설적이게도 저 같은 경우는 블로깅 서비스가 사용하기 쉽기 때문에 글을 쓰다 멈추고 다시 이어쓰고 싶지 않았던 경우가 꽤 되었습니다. 왜일까요? 이 이유에 대해서는 저도 한번 고민해봐야겠네요. 몇 가지 떠오르는 이유가 있긴한데 아직 명확하게 설명하기는 어렵습니다.

글 앞에 번호를 붙이면 이상하게 5가지 이상의 이유를 적어야 할 것 같은 생각이 들지만 사실 1번의 이유가 가장 크기 때문에 Jekyll을 선택한 이유에 대해서는 이쯤 마무리를 하겠습니다.

이외에도 처음 생각했던 것은 Jekyll 테마중 적당한걸 하나 골라서 구축하려고 했는데 기왕이면 Notion 페이지와 유사한 랜딩 페이지를 만들고 싶더군요. 그래서 최종적으로는 Jekyll 테마중에서 minimal-mistakes의 splash 페이지를 조금 수정한 버전으로 구축하였습니다. 생각외로 테마 수정이 유연해서 어렵지 않았네요.

지금 가장 큰 걱정은 꾸준하게 블로깅을 할 것인가인데 너무 부담은 갖지 않고 천천히 해보려고 합니다. 그럼 다음 포스팅에서 뵙겠습니다.