- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi all,
I was able to successfully configure uClinux for my NEEK board, and run some demo applications on it. Now, I have to make all of this work on a board that we designed in the office. I believe there should be some file that we need to modify or create and that I should select it as my platform, is this right? if yes, how should I create it? based on what rules ... ? Plus, don't I have to tell uClinux about the pin mapping ? Any pointers would be helpful. ThanksLink Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
The Linux build process uses a file that Quartus outputs to assign addresses to the peripherals/resources that are defined in the FPGA. To allow for this the names assigned to them in Quartus need to match the build script. (AFAIK, there are different files used for the current scripts for the MMU and the no-MMU flavor of Linux.)
AFAIK, There are some articles about this in the Linux-Wiki. Linux does not know about pins. -Michael- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thanks mschnell for your help,
Actually I am having trouble finding this topic in the nios wiki, can you please paste some links. Thanks regards- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
You may want to start here to investigate the use of the ptf file. (the NoMMU case is covered more decently in the Wiki):
http://www.nioswiki.com/operatingsystems/uclinux/uclinuxdist -Michael- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Micheal,
I might have been not so clear in explaining what I was searching for. On my NEEK board, I was able to develop my own applications and run them. Now it's time to do the same tasks on our design (developed at our company). I need pointers to config linux to run on our platform (different than NEEK). Regards, Dany- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Best you use a directory under "vendors" for your hardware and software design to allow menuconfig to manage the specific properties of your project. You can use "nios2mmu" as a template by copying the nios2nommu directory below vendors to your own one besites same. In menuconfig -> Vendor/Product selection, you now can select this to manage the configuration files in there and to use the make file it contains, which you might want to manipulate.
Regarding your not NEEK compatible hardware: either the appropriate driver is provided with the distribution, now you need to do the Quartus design (and thus the ptf file) so that the make process recognizes it. That is why the names of the elements need to be chosen appropriately. Or there is no compatible (i.e. besites other things: correct names) Kernel driver for that hardware yet: Now you need to create it yourself and take the appropriate steps to integrate it in the make script hierarchy. You need to study the Wiki and the existing script on learning how to do this. -Michael- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Micheal,
Our design is similar to the NEEK (though obviously different pin mapings), we will be using the same hardware chips except for the LCD controller, we will be using frame reader instead of lcd sgdma and external video output. Additional hardware modules will be on our board as well, I will have to access these from userland. This means that most drivers are there and ready to use, once I can point linux to our designs pins (I think). One of my concerns will be able to display graphics on the LCD (using fb.h) with the frame reader instead of the lcd sgdma. From what you explained, creating our SOPC design with names that linux know will enable my board to run linux, right ? Once again, thanks for taking the time to answer my beginner's questions. Regards, Dany- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Sorry. I have no knowledge at all about "Frame Reader" :(. What is this ? I did some initial test with VNC and was able to transfer some picture data via TCP/IP to the PC, but I did not try yet to link this picture buffer to Nano-X.
But when modifying the function of a device usually means that the Kernel driver needs to be modified. Here the frame Buffer driver used by the NEEK Linux distribution uses the SGDMA's programmer's interface. If you replace that, the driver will not work any more. -Michael- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Ok, than I should be able to modify the frame buffer to use frame reader instead of frame buffer, right ? Any pointers on how to do that ?

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page