add zibi test by challenger1024 · Pull Request #1060 · riscv/riscv-arch-test

challenger1024 · 2026-03-09T08:55:47Z

Add tests for the experimental extension Zibi.
Link: https://github.com/riscv/zibi/issues/2

Copilot

Pull request overview

This PR adds test generation support for the experimental Zibi RISC-V extension, which defines two new branch instructions (beqi and bnei) that compare a register against a compact 5-bit immediate.

Changes:

Adds a new BI instruction type formatter, a new coverpoint generator for cross-product of rs1/cimm edges, and extends cp_offset and cp_imm_edges to support the BI type.
Adds the Zibi.csv testplan, new coverage templates (sample_BI.sv, cr_rs1_cimm_edges_offset.sv, cp_imm_edges_5bit_u_n0.sv), and new trace/disassembler support for BEQI/BNEI.
Adds imm_5bit edge values to IMMEDIATE_EDGES and new cimm field to RISCV_trace_data and RISCV_instruction_base.

Reviewed changes

Copilot reviewed 7 out of 12 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
`testplans/Zibi.csv`	Testplan defining coverpoints for `beqi` and `bnei`
`generators/testgen/src/testgen/formatters/types/bi_type.py`	New BI instruction type formatter
`generators/testgen/src/testgen/data/edges.py`	Adds 5-bit unsigned immediate edge values
`generators/testgen/src/testgen/coverpoints/special/cr_rs1_cimm_edges_offset.py`	New cross-product coverpoint generator for rs1 edges × cimm edges × branch direction
`generators/testgen/src/testgen/coverpoints/special/cp_offset.py`	Extends cp_offset to support BI type; has a bug with missing `immval` in backward branch emission
`generators/testgen/src/testgen/coverpoints/cp_imm_edges.py`	Adds 5-bit unsigned immediate edge support; misleading comment
`generators/coverage/templates/sample_BI.sv`	Coverage sampling template for BI type
`generators/coverage/templates/cr_rs1_cimm_edges_offset.sv`	Cross-coverage template
`generators/coverage/templates/cp_imm_edges_5bit_u_n0.sv`	Empty — missing coverpoint definition
`framework/src/act/fcov/disassemble.svh`	Adds BEQI/BNEI disassembly; uses undefined constants `BEQI`/`BNEI`
`framework/src/act/fcov/coverage/RISCV_trace_data.svh`	Adds `cimm` field to trace data
`framework/src/act/fcov/coverage/RISCV_instruction_base.svh`	Adds `add_cimm()` method

Copilot · 2026-03-10T12:00:38Z

generators/testgen/src/testgen/coverpoints/special/cr_rs1_cimm_edges_offset.py

+##################################
+# cr_rs1_cimm_edges_offset.py
+#
+# tjc.challenger1024@jmail.com Mar 2026


The email address tjc.challenger1024@jmail.com appears to be a typo — it says jmail.com instead of gmail.com.

Copilot · 2026-03-10T12:00:38Z

framework/src/act/fcov/disassemble.svh

+    BEQI: $sformat(decoded, "beqi %s, %0d, %0d", rs1, cimm5, immBType);
+    BNEI: $sformat(decoded, "bnei %s, %0d, %0d", rs1, cimm5, immBType);


The BEQI and BNEI constants used in the casez statement are not defined anywhere in the decode packages (RISCV_decode_pkg.svh or RISCV_imported_decode_pkg.svh). All other instructions in the casez (e.g., BEQ, BNE, EBREAK, etc.) have corresponding localparam entries. Without these definitions, the SystemVerilog will fail to compile because BEQI and BNEI are undefined identifiers. The opcode bit patterns for beqi and bnei need to be added to RISCV_decode_pkg.svh or RISCV_imported_decode_pkg.svh.

Copilot · 2026-03-10T12:00:39Z

generators/testgen/src/testgen/coverpoints/special/cp_offset.py

+        elif instr_type == "BI":
+            assert params.rs1 is not None and params.immval is not None
+            if instr_name == "beqi":
+                test_lines.append(f"LI(x{params.rs1}, {params.immval}) # setup for taken branch")
+            else:  # bnei
+                val = params.immval + 1 if params.immval != 31 else params.immval - 1
+                test_lines.append(f"LI(x{params.rs1}, {val}) # setup for taken branch")


The backward branch instruction generated for BI type is missing the immediate operand. Line 55 (unchanged context) uses f"2: {instr_name} x{params.rs1}, {f'x{params.rs2},' if params.rs2 is not None else ''} 1b # backward branch". For BI type, params.rs2 is None, so this generates beqi x{rs1}, 1b rather than the correct beqi x{rs1}, {params.immval}, 1b. The elif instr_type == "BI" branch added here must also handle the branch instruction emission on line 55, or line 55 needs to be updated to include params.immval for the BI type.

Copilot · 2026-03-10T12:00:39Z

generators/testgen/src/testgen/coverpoints/cp_imm_edges.py

+    elif coverpoint.endswith("_5bit"):
+        edges_imm = IMMEDIATE_EDGES.imm_5bit
+    elif coverpoint.endswith("_5bit_u_n0"):
+        edges_imm = IMMEDIATE_EDGES.imm_5bit[1:]  # exclude imm=0, add imm=-1 for Zibi extension


The comment on line 33 says "exclude imm=0, add imm=-1 for Zibi extension", but the code only slices IMMEDIATE_EDGES.imm_5bit[1:] which simply excludes 0. No -1 value is ever added. If cimm=0 is supposed to encode -1 per the Zibi spec (as reflected in disassemble.svh line 59 where cimm5 = (instr[24:20] == 0) ? -1 : instr[24:20]), the comment is misleading. Either the comment should be corrected to just say "exclude imm=0", or a -1 entry should actually be added to test the special cimm=0 encoding.

davidharrishmc

Exciting to see this first PR for a new extension in ACT4. Thank you for contributing!

A few things I'm trying to follow, and I'm sure @jordancarlin will have feedback too.

davidharrishmc · 2026-03-10T12:03:15Z

generators/coverage/templates/cr_rs1_cimm_edges_offset.sv

@@ -0,0 +1,3 @@
+    cr_rs1_cimm_edges_offset : cross cp_rs1_edges,cp_imm_edges_5bit_u_n0,cp_offset  iff (ins.trap == 0 )  {


The cp_imm_edges_5bit_u_n0.sv file is showing up empty for me during review. Either it wasn't committed properly or Github is doing something wonky on my viewer.

I'd like to check the case of 0 becoming -1 is in the coverpoint.

davidharrishmc · 2026-03-10T12:14:31Z

generators/testgen/src/testgen/coverpoints/special/cr_rs1_cimm_edges_offset.py

+                    "0: # destination for backwards branch that is never taken",
+                    f"{instr_name} x{params.rs1}, {params.immval}, 3f # forward branch, if taken",
+                    "1: # goes here if not taken",
+                    f"{instr_name} x{params.rs1}, {params.immval}, 0b # backward branch, never taken",


If there is a bug in the implementation and this branch is taken, it looks like the test may infinite loop. Can this be modified to set a signature (which will mismatch the reference model) and keep going?

davidharrishmc · 2026-03-10T12:16:12Z

generators/testgen/src/testgen/coverpoints/special/cr_rs1_cimm_edges_offset.py

+                    "j 4f # done with test",
+                    "3: # goes here during forward branch if taken",
+                    f"{instr_name} x{params.rs1}, {params.immval}, 2b # backward branch, definitely taken",
+                    "4: # done with test",


Put another sigupd here in case the branch that should have been taken didn't get taken.

davidharrishmc · 2026-03-10T12:22:09Z

generators/testgen/src/testgen/coverpoints/cp_imm_edges.py

+    elif coverpoint.endswith("_5bit"):
+        edges_imm = IMMEDIATE_EDGES.imm_5bit
+    elif coverpoint.endswith("_5bit_u_n0"):
+        edges_imm = IMMEDIATE_EDGES.imm_5bit[1:]  # exclude imm=0, add imm=-1 for Zibi extension


I'm having trouble following how the effective immediate of -1 is generated. There is a draft spec below, which I think is backward (should it say immediate negative one is represented as zero)? Is this code excluding that case?

Zibi extension encodes all 5-bit immediate as positive number, with the exception of immediate zero, which is represented as negative one
https://github.com/riscv/zibi/blob/main/src/zibi.adoc

add zibi test

aeac480

davidharrishmc requested a review from Copilot March 10, 2026 11:49

Copilot started reviewing on behalf of davidharrishmc March 10, 2026 11:54 View session

Copilot AI reviewed Mar 10, 2026

View reviewed changes

davidharrishmc requested changes Mar 10, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

add zibi test#1060

add zibi test#1060
challenger1024 wants to merge 1 commit intoriscv:act4from
challenger1024:zibi-test

challenger1024 commented Mar 9, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 10, 2026

Uh oh!

Copilot AI Mar 10, 2026

Uh oh!

Copilot AI Mar 10, 2026

Uh oh!

Copilot AI Mar 10, 2026

Uh oh!

davidharrishmc left a comment

Uh oh!

davidharrishmc Mar 10, 2026

Uh oh!

davidharrishmc Mar 10, 2026

Uh oh!

davidharrishmc Mar 10, 2026

Uh oh!

davidharrishmc Mar 10, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		BEQI: $sformat(decoded, "beqi %s, %0d, %0d", rs1, cimm5, immBType);
		BNEI: $sformat(decoded, "bnei %s, %0d, %0d", rs1, cimm5, immBType);

		@@ -0,0 +1,3 @@
		cr_rs1_cimm_edges_offset : cross cp_rs1_edges,cp_imm_edges_5bit_u_n0,cp_offset iff (ins.trap == 0 ) {

Conversation

challenger1024 commented Mar 9, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

davidharrishmc left a comment

Choose a reason for hiding this comment

Uh oh!

davidharrishmc Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

davidharrishmc Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

davidharrishmc Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

davidharrishmc Mar 10, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants