Quartus 20.1: dmi/cpu_s/rstn_ext assigned a value but never read

[andkae]

Hi Stephan,

at the moment i do some synthesis and checking logs. There i found following messages:

Warning (10036): Verilog HDL or VHDL warning at neorv32_top.vhd(339): object "dmi" assigned a value but never read
Warning (10036): Verilog HDL or VHDL warning at neorv32_top.vhd(286): object "cpu_s" assigned a value but never read
Warning (10036): Verilog HDL or VHDL warning at neorv32_top.vhd(258): object "rstn_ext" assigned a value but never read

My related core config is:

ON_CHIP_DEBUGGER_EN => false,
IO_WDT_EN => false

#DMI
DMI could the soultuion to remove this assignments:

neorv32/rtl/core/neorv32_top.vhd

Lines 1725 to 1728 in 278ab39

	dmi.req_ready <= '0';
	dmi.resp_valid <= '0';
	dmi.resp_data <= (others => '0');
	dmi.resp_err <= '0';

#rstn_ext
For rstn_ext could the assignement done in a conditional assignment:

neorv32/rtl/core/neorv32_top.vhd

Line 476 in 278ab39

rstn_ext <= '0';

if (true = (IO_WDT_EN or ON_CHIP_DEBUGGER_EN)) then
  rstn_ext      <= '0';
end if;

neorv32/rtl/core/neorv32_top.vhd

Line 488 in 278ab39

rstn_ext <= and_reduce_f(rstn_ext_sreg); -- external reset (via reset pin)

if (true = (IO_WDT_EN or ON_CHIP_DEBUGGER_EN)) then
  rstn_ext <= and_reduce_f(rstn_ext_sreg); -- external reset (via reset pin)
end if;

For cpu_s it's more complicated because the signal is assigned inside neorv32_cpu, there i have no idea now.

What do you think could this be a change in the NEORV32. If yes, i could create a PR.

BR,
Andreas

[stnolting]

Hey Andreas!

There i found following messages: [...]

I know those messages/warnings are annoying, but they do not really hurt. Unconnected signals are reported, but they won't be synthesized in any way. So there is no drawback for the final hardware when having such "dead" signals.

For cpu_s it's more complicated because the signal is assigned inside neorv32_cpu, there i have no idea now.

That is the reason I am not sure about this. I like to be consequent here - if one unused signal is explicitly terminated then this should be done for all unused signals.

However, I am not sure about this and I'm open for a discussion. 🤔