How to work with an existing Hog-handled repository¶
This section is intended for a developer that starts working with an existing HDL project that is managed with Hog.
All the instructions below can be executed both on a Linux shell or on git bash on a Windows machine.
Tip
To open a git bash session navigate to the directory where you want to open the bash. Right-click on the folder and select open git bash here.
For all of the following to work, Vivado (or Quartus or PlanAhead) executable must be in your PATH environment variable: i.e. if you type vivado the program must run. If you intend to use Modelsim or Questasim, also the vsim executable must be in the PATH: i.e. if you type vsim the simulator should start.
Requirements¶
This is a list of the requirements:
Have git (version 2.7.2 or greater) installed and know git basics
Have Vivado, PlanAhead or Quartus installed and in the PATH
Optionally have Questasim installed and vsim in the PATH
Cloning the repository¶
First of all, you have to clone the repository, let’s call it repo from now on. Go to the website of the repository, choose the protocol (ssh, git, https, krb5) and copy the clone the link.
git clone --recursive <protocol>://gitlab.cern.ch/repo.git
Now you have all the source code and scripts you need in the repo folder.
Note
You will have to choose the protocol that works for you: ssh, https, or krb5. We used the --recursive option to automatically clone all the submodules included. In general, an HDL repository may or may not include other submodules, but the Hog scripts are always included as submodules. If you have cloned the repository without the recursive options (or if that option does not work, we heard that it happens on Windows), you will have to go inside it and initialise the submodules git submodule init and update them git submodule update.
Create Vivado/PlanAhead/Quartus projects¶
To start working you can now create the Vivado/PlanAhead/Quartus project you are interested in, say it’s called project1.
To do that, go into the repository (cd repo) and type:
./Hog/CreateProject.sh project1
This will create the Vivado/PlanAhead/Quartus project in the directory Projects/project1.
Inside this directory, you will find the Vivado .xpr file (or the Quartus .qpf or the PlanAhead .xise files).
You can now open the project file with the GUI and synthesise/implement the project.
Note
Inside the Quartus GUI, you must always run a full compile process after creating a new project, otherwise the synthesis process will fail. Due to the Quartus structure, only running the full compilation flow will ensure the Hog generics are properly set.
If you do not know the project name, just run ./Hog/CreateProject.sh and you will get a list of the existing projects present on the repository.
Tip
Alternatively, you can type cd Top (the Top folder is always present in a Hog handled HDL repository) and type ls: each directory in this path corresponds to a Vivado/PlanAhead/Quartus project in the repository.
To create all the projects in the repository, you can run the Hog initialisation script, like this:
./Hog/Init.sh
This script will also, if you wish, compile Modelsim/Questasim libraries.
Adding a new file to the project¶
When you open project1, you can work with the compiler IDE (almost) normally.
The CreateProject.sh script, that you have just run, has integrated Hog’s Tcl scripts in the Vivado/PlanAhead/Quartus project. From now on, Hog scripts will run automatically, every time you start the compilation workflow or any step in the workflow. In particular, the pre-synthesis script will interact with your local git repository and integrate its version and git commit SHA into your HDL project.
We said almost normally because there is one exception: adding a new file (HDL code, constraint, IP, etc.) to the project using the GUI is not enough. You must also add the file name in one of Hog’s list files as explained in the following.
Let’s now suppose that you want to add a new file to the project and that this file is located in repo/dir1/ and is called file1.hdl.
First of all, the new file (that is unknown to git) must be added to the repository:
cd repo
git add ./dir1/file1.hdl
git commit -m "add new file file1.hdl"
git push
Now that the file is safely on git, we have to add it to the Hog project, so that if another developer clones the repository, as you did at the beginning of this guide, the file will appear in the project.
Note
Not all the files stored in the git repository are part of a project: there can be non-HDL files, obsolete files that are stored just in case, new files that are not ready to be used. Moreover, some files could be part of a project and not of another. In our example, the repository could contain project2 and project3 that use different subsets of files in the repository.
This is a new file, unknown to Hog for now, and we want it to be included in the project the next time that we run the CreateProject.sh script described above. To do this, you must add the file name and path of file1.vhd into a Hog list file. The list files are located in repo/Top/project1/list/.
Manually updating the list file¶
Let’s assume that the list file you want to add your file to is lib1.src.
Open the file with a text editor and add the file name and path in a new line.
Now that the new file is included in a list file, you can close the project and re-create it by typing ./Hog/CreateProject.sh <project name> again.
There is also a quicker option to add a new file without recreating the project. You can add the file with the GUI and also add the file to a .src list file. If you choose to do this, in Vivado, you have to choose the correct library when adding the file. The library must have the same name as the .src file to which you added the source file. In our example, the HDL file was added to a list file called lib1.src, so the library that you have to choose is lib1. You can select the library in the Vivado GUI from a drop-down menu when you add the file.
This procedure is valid for any kind of source file. If your file is a constraint file, just add it to a .con list file in the list directory, e.g. repo/Top/project1/list/lib1.con.
Renaming a file already in the project¶
If you need to rename or move a file, say from path1/f1.hdl to path2/f2.hdl do so and change the name in the proper list file accordingly.
Do not forget to rename the file on git as well:
git mv path1/f1.hdl path2/f2.hdl
git commit -m "Renamed f1 into f2"
git push
What can go wrong?¶
If you do something wrong (e.g. you add a name of a non-existing file, create a list file with an invalid extension, etc.) you will get an error when you run the CreateProject.sh script. In this case, read Hog’s error message and try to fix it.
If you do something wrong with Vivado library, or if you forgot to update the list files, the error will show at synthesis time because Vivado will not be able to find the component.
Automatically updating the list file¶
Warning
This feature is currently only supported for Vivado.
Hog provides a tcl script (check_list_files.tcl) to facilitate the handling of the list files. Users can work normally within the Vivado GUI, adding new files to the projects (and selecting the proper library); afterwards, they can automatically update the list files by either:
clicking on the
Hog_listFilescustom button on the Vivado toolbar;
Tip
To have the Hog_listFiles button on the Vivado toolbar, the Hog Tcl script add_hog_custom_button.tcl must be run.
running the command (e.g. from bash console)
vivado -mode batch -notrace -source <repository_path>/Hog/Tcl/utils/check_list_files.tcl -tclargs -project <project name> -recreate -force
Info
New vhdl files will be added to repo/Top/<project>/list/<library>.src.
New simulation files will be added to repo/Top/<project>/list/<simulation_set>.sim.
New constraint files wil be added to repo/Top/<project>/list/default.con.
Other files (verilog, IPs, etc.) wil be added to repo/Top/<project>/list/default.src.
Adding a new IP¶
If you want to add a new IP core, say it is called my_ip1, you must create it in out-of-context mode and save the .xci file (and only that) in the repository in a subfolder of your repository.
Similarly in for Quartus, you can add the .ip file to your repository. If your target FPGA only supports .qip files (e.g. Intel Max10 FPGAs) you must add the folder in which the IP was generated to your project.
Info
You can keep IPs wherever you want in your repository.
For example: repo/some_folder/my_ip1/my_ip1.xci, but particular attention should be paid with the .gitignore file.
When Vivado synthesises the IPs, it creates plenty of additional files where the .xci file is located. To avoid to commit those file to the repository, a .gitignore file is used.
Info
Every file that is not a .xci file inside the directory containing the IP must be ignored by git.
This can be specified in a global .gitignore file or in several local .gitignore files copied in each directory that you use to store IPs.
The .gitignore file, provided in Hog templates, assumes you are storing your IPs in a single directory called IP and located in the root of your repository.
Once the IP is created and added to the repository, you can add the .xci normally to any source list file in the list folder of your project.
IP initialisation coefficient files (.coe, .mif)¶
If you have a .coe or .mif file for RAM initialisation, you can store it anywhere in the repository.
Pay attention to specify the path to this file as a relative path.
This can be done in the text box in GUI when you customise the IP.
Please note that as .gitignore file is to ignore files other than .xci, you cannot save the coefficient file in the same directory as your .xci.
Note
Note that browsing to the file location using the GUI will cause the absolute path to be loaded in the textbox. Be careful to replace this with a relative path.
Please note that as .gitignore file is to ignore files other than .xci, you cannot save the coefficient file in the same directory as your .xci.
A couple of things before getting to work¶
Here you can find a couple of details and suggestions that can be useful when working with Hog-handled repository.
Commit before starting the workflow¶
All the Hog scripts handling version control are automatically added to your project: this means that you can create a reproducible and traceable binary file, even when you run locally. This will happen only if you commit your local changes before running synthesis. You don’t have to push! Just commit locally, then you can push when you are sure that your work is good enough. If you don’t commit, Hog will alert you with a Critical Warning at the beginning of the synthesis.
Different list files¶
As we have explained above, source files taken from different list files will be added to your project in different “libraries”: the name of each library is the name of the list file. When working with components coming from different list files, you will need to formally include the libraries and call the component from the library it belongs to. For example, in VHDL:
library lib1
use lib1.all
...
u_1 : entity lib1.a_component_in_lib1
port map(
clk => clk,
din => din,
dout => dout
);
If working within the same library, you can normally use the “work” library.