I still would prefer if there was an Ethernet interface already on the board.
Yes I did have a quick look at a few Zynq boards which I knew of, but none had a PL connected ethernet PHY. I guess development board manufacturers expect you to do all your networking through the PS
That is the intended use: boot Linux first, bring up the FPGA later when the relevant drivers load.
I am not sure about the Zynq architecture, but maybe it's possible to reach the interconnect bus from FPGA side without initializing the CPU?
I'm fairly sure that that is not possible on the Zynq. It's quite easy to do the reverse (use a PL pin for a PS peripheral) in most cases, but there is no access to dedicated PS pins from the PL.
There are some rarely-seen Zynq variants that boots FPGA first, but in most cases you have to boot ARM first, and the FPGA in intended to serve solely as a peripheral to the ARM.
@technix,
In my case I don't want to use Linux for Ethernet. I wan't to use the cores from "FPGA-Cores.com". Why?
* It is much easier to access the FPGA logic using the AXI4-streams instead of going through an OS.
* I only add the netlist and I have all Ethernet stuff I need. Very easy.
* I can update/program the flash from remote.
* I can debug with the internal logic analyzer from remote.
In the ARM I'm not going to have Linux at all. It will only have some code running without os. Doing some algorithms that is more suitable to have in c++.
The cores does only support standard interfaces like RMII so I don't think it is possible to access the AXI-Master of the PS.
If this is your use case, maybe Zynq is NOT the correct choice for you. A better option would be choose a chip with built-in ARM Cortex-M hard core, burn some gates to include a soft ARM Cortex-M or RISC-V core, or use an external microcontroller hooked to some kind of communication interconnect (SPI, for example.)