Allow choosing between `uvm_analysis_tlm_fifo` and `uvm_analysis_imp`

[antmarzam]

Using uvm_analysis_fifo is fine, but requires calling get whenever an item is available.

uvm_analysis_imp or uvm_analysis_imp_decl(port_identifier) are helpful because they have the TLM connections in it, you don’t have to call get. Instead, there is a write method (write_port_identifier for uvm_analysis_imp_decl) which gets called every time there’s a new item available.

Uvm_analysis_imp example:

Example for uvm_analysis_imp_decl:
https://verificationguide.com/uvm/tlm-analysis-port-multi-analysis-imp-port/

[marnovandermaas]

What's the specific suggestion here?

[antmarzam]

Currently the DV guidelines say to use uvm_tlm_analysis_fifo.
I think it would be nice to also allow using uvm_analysis_imp. Which as you can see in the example above comes with all the built-in infrastructure regarding receiving the sequence items. You just have to write a function which gets called every time an item is available. Instead of having to explicitly call the get method in uvm_tlm_analysis_fifo

[rswarbrick]

I think that's probably a good idea, but maybe a low-ish priority. Why don't we try porting some uses in OpenTitan to the _imp style? That might make it obvious if there's a reason to avoid it.