Actions
Forum Thread Details
  • State Not Answered
  • Replies 2 replies
  • Subscribers 328 subscribers
  • Views 316 views
  • Users 0 members are here
Related

PHY set up, and interacting with the "gem"

Former Member

I am trying to configure the PHY in an ethernet packet manipulating program that I am writing.
According to the technical reference manual (Zynq-7000 AP SoC Technical Reference Manual), before I begin configuring bits on the PHY maintenance register I need to wait for it to become idle ("gem.net_status[phy_mgmt_idle]=1").

However, I cannot simply do "while(gem.net_status[phy_mgmt_idle]!=1){sleep(1);print("waiting for idle
");}
because gem is not an object in my program.

Where do I go to find the documentation about how to interact gem?

  • 0

    As discussed in an earlier thread you take a look at the example code for the xemacps there are functions in this code used to access the PHY registers: http://zedboard.org/content/peripheralstest-xemacps

     

    These routines are written at a higher level of abstraction and are taking care of the low level 'bit banging' of the physical interface you are referring to from the TRM so that you can just call these functions and read or write the PHY registers. You can follow the code down level by level to see this by looking at function calls like 'XEmacPs_PhyRead' and 'XEmacPs_PhyWrite'  within the functions like XEmacPsDetectPHY used in the example code. These functions use lower level function calls like XEmacPs_WriteReg and XEmacPs_ReadReg which are directly accessing the registers referred to in the TRM.

     

    -Gary

  • 0

    Alright, I've figured out how to read from, and write to the gem.
    You can use 3 methods:

    1st method - XEmacPs_WriteReg. This is really just a macro that expands to call the function Xil_Out32.

    2nd method - Xil_Out32. You could directly call the function itself. However you lose the ability to easily change which function you're calling, by simply changing the macro.

    3rd method - *(volatile u32 *) address. Just go to the (address + offset) and change it yourself.


    Now, I have opted for the 3rd method, and wrote my own functions to replace Xil_Out32. There are many parts in the "Programming Guide" of the Technical Reference Manual where it tells you to write to specific bits, and other parts where it say to write to a whole register. By writing my own functions I can handle that... Well, to my knowledge, correctly. Xil_Out32 didn't handle changing only a single bit, within a u32.

    So now I have:
    static void write_bit(u32 address, u32 bit, u32 value){
    tif(value == 1){*(volatile u32 *) address |= (1u << bit);}
    telse{*(volatile u32 *) address &= ~(1u << bit);}
    }
    static void write_reg(u32 address, u32 value){
    t*(volatile u32 *) address = value;
    }
    static u32 read_bit(u32 address, u32 bit){
    t//TODO, I'm about to leave.
    }
    static u32 read_reg(u32 address){
    treturn *(volatile u32 *) address;
    }