nonce_offset_i inferred in async reset always block, but is not included in reset

[nstewart-amd]

Lint violations results from the following:

"./pcrvault/rtl/pv_gen_hash.sv"
198 always_ff @(posedge clk or negedge rst_b) begin : api_regs
199 if (~rst_b) begin
200 gen_hash_fsm_ps <= GEN_HASH_IDLE;
201 block_offset_i <= '0;
202 read_entry <= '0;
203 read_offset <= '0;
204 end
205 else begin
206 gen_hash_fsm_ps <= gen_hash_fsm_ns;
207 block_offset_i <= block_offset_nxt;
208 nonce_offset_i <= nonce_offset_nxt;
209 read_entry <= read_entry_nxt;
210 read_offset <= read_offset_nxt;
211 end
212 end

nonce_offset_i is inferred in an async reset always block, but is not included in the reset condition.

Correction:

1. Assuming nonce_offset_i is don't care until after it's initialized by the combination assignment of nonce_offset_nxt, nonce_offset_i should be moved to a new (simple) sequential always block with no reset inference.
   198 always_ff @(posedge clk or negedge rst_b) begin : api_regs
   199 if (~rst_b) begin
   200 gen_hash_fsm_ps <= GEN_HASH_IDLE;
   201 block_offset_i <= '0;
   202 read_entry <= '0;
   203 read_offset <= '0;
   204 end
   205 else begin
   206 gen_hash_fsm_ps <= gen_hash_fsm_ns;
   207 block_offset_i <= block_offset_nxt;
   209 read_entry <= read_entry_nxt;
   210 read_offset <= read_offset_nxt;
   211 end
   212 end

198 always_ff @(posedge clk) begin : api_regs
208 nonce_offset_i <= nonce_offset_nxt;
212 end

2. If nonce_offset_i is NOT don't care, it should be added to the reset begin/end group.
   198 always_ff @(posedge clk or negedge rst_b) begin : api_regs
   199 if (~rst_b) begin
   200 gen_hash_fsm_ps <= GEN_HASH_IDLE;
   201 block_offset_i <= '0;
   **208 nonce_offset_i <= '0;
   202 read_entry <= '0;
   203 read_offset <= '0;
   204 end
   205 else begin
   206 gen_hash_fsm_ps <= gen_hash_fsm_ns;
   207 block_offset_i <= block_offset_nxt;
   208 nonce_offset_i <= nonce_offset_nxt;
   209 read_entry <= read_entry_nxt;
   210 read_offset <= read_offset_nxt;
   211 end
   212 end

[Nitsirks]

Looking into why we aren't seeing this on our end. I thought we had this rule.
@nstewart-amd can you send me the rule that caught this?

Looks like nonce offset was added later and just missed in the reset block. I don't think it was explicitly meant to be a non-reset flop. We should add this to the reset block.

[nstewart-amd]

Michael - This is caught by UnInitializedReset-ML, which was not included in 1p0RC LINT rule set recommendations.

[bharatpillilli]

@Nitsirks - close this?

[calebofearth]

This is addressed by commit 020cdc6 from #483