refactor(tests): bind the EIP-8141 calldata floor settlement tests by the minimum data by spencer-tb · Pull Request #3784 · ethereum/execution-specs

spencer-tb · 2026-10-12T20:34:33Z

Description

Follow-up to this review comment on #3756: the calldata floor settlement tests now bind the floor by the minimum data instead of a fixed margin.

floor_binding_padding in the 8141 helpers, the binary search feat(spec-specs,test-forks,tests): charge EIP-8250 keyed nonce access in intrinsic gas and the calldata floor #3756 added for 8250, finds the fewest bytes of frame data that bind the floor.
test_calldata_floor_with_state_gas drops the 30,000-byte FLOOR_PADDING_DATA and adds below_floor, one byte short of binding. It now runs from a contract sender, which carries no signature bytes, so the floor starts below the standard cost, and each frame budget is exactly what it uses.
test_storage_refund_settlement replaces the 16-byte linear search with byte-exact boundaries against the post-refund and pre-refund gas, and adds floor_just_below_pre_refund.

Counting one floor byte too many in the spec fails all eight cases, and one too few fails the five where the floor binds.

Related Issues or PRs

feat(spec-specs,test-forks,tests): charge EIP-8250 keyed nonce access in intrinsic gas and the calldata floor #3756

Checklist

Ran fast static checks to avoid CI fails, see Code Standards & Verifying Changes: just static
PR title has the form <type>(<area>): <title>, where <type> and <area> come from an appropriate C-<type>, respectively A-<area>, label. The title should match the target squash commit message.

… the minimum data

codecov · 2026-10-12T20:39:32Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (eips/bogota/eip-8141@55182c9). Learn more about missing BASE report.

Additional details and impacted files

@@                   Coverage Diff                   @@
##             eips/bogota/eip-8141    #3784   +/-   ##
=======================================================
  Coverage                        ?   94.94%           
=======================================================
  Files                           ?      676           
  Lines                           ?    41858           
  Branches                        ?     3942           
=======================================================
  Hits                            ?    39741           
  Misses                          ?     1465           
  Partials                        ?      652

Flag	Coverage Δ
unittests	`94.94% <ø> (?)`

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

…act sender

LouisTsai-Csie

I create a PR for some refactor, please let me know what you think about it:

spencer-tb#77

LouisTsai-Csie · 2026-10-13T04:06:53Z

+    )
    worker_code = (write if with_state_write else Bytecode()) + Op.STOP
    worker = pre.deploy_contract(code=worker_code)
    state_used = write.state_cost(fork) if with_state_write else 0


I think this is the same

Suggested change

state_used = write.state_cost(fork) if with_state_write else 0

state_used = worker_code.state_cost(fork)

LouisTsai-Csie · 2026-10-13T05:17:09Z

+        data_length -= 1
+    tx, gas_used_before_refund, calldata_floor = build(data_length)
+    # The premise of the test: the refund applies uncapped.
+    assert 0 < refund <= gas_used_before_refund // 5


I suggest not hardcode the number here, use max_quotient() function instead.

gurukamath · 2026-10-13T08:14:20Z

+    gas_used = max(
+        execution_used(padding), floor_calculator(frames=frames(padding))
+    )


Suggested change

gas_used = max(

execution_used(padding), floor_calculator(frames=frames(padding))

)

execution_gas = execution_used(padding)

calldata_floor = floor_calculator(frames=frames(padding))

# The premise of the test: the floor binds, or in `below_floor`

# falls short by the last byte.

assert (calldata_floor > execution_gas) == floor_bound

gas_used = calldata_floor if floor_bound else execution_gas

I know the premise of this test is already asserted by the helper but it is still worth making it explicit here imo

gurukamath · 2026-10-13T08:18:14Z

+    tx, gas_used_before_refund, calldata_floor = build(data_length)
+    # The premise of the test: the refund applies uncapped.
+    assert 0 < refund <= gas_used_before_refund // 5
+    gas_used_after_refund = gas_used_before_refund - refund


Suggested change

gas_used_after_refund = gas_used_before_refund - refund

gas_used_after_refund = gas_used_before_refund - refund

# Assert the premise of each case

if against_post_refund and one_byte_short:

assert calldata_floor < gas_used_after_refund

elif not against_post_refund and not one_byte_short:

assert gas_used_before_refund < calldata_floor

else:

assert gas_used_after_refund < calldata_floor < gas_used_before_refund

The premise of this test is a bit more complex and asserting it in full at this stage prevents silent drift. Also serves as a documentation of the premise here.

gurukamath · 2026-10-13T08:26:26Z

 @pytest.mark.parametrize(
-    "floor_case",
+    "against_post_refund,one_byte_short",
    [
-        pytest.param("below_post_refund", id="post_refund_above_floor"),
-        pytest.param("between", id="floor_between_pre_and_post_refund"),
-        pytest.param("above_pre_refund", id="floor_above_pre_refund"),
+        pytest.param(True, True, id="post_refund_above_floor"),
+        pytest.param(True, False, id="floor_between_pre_and_post_refund"),
+        pytest.param(False, True, id="floor_just_below_pre_refund"),
+        pytest.param(False, False, id="floor_above_pre_refund"),
    ],
 )


Suggested change

@pytest.mark.parametrize(

"floor_case",

"against_post_refund,one_byte_short",

[

pytest.param("below_post_refund", id="post_refund_above_floor"),

pytest.param("between", id="floor_between_pre_and_post_refund"),

pytest.param("above_pre_refund", id="floor_above_pre_refund"),

pytest.param(True, True, id="post_refund_above_floor"),

pytest.param(True, False, id="floor_between_pre_and_post_refund"),

pytest.param(False, True, id="floor_just_below_pre_refund"),

pytest.param(False, False, id="floor_above_pre_refund"),

],

)

@pytest.mark.parametrize(

"boundary,floor_side",

[

pytest.param("post_refund", "below", id="floor_just_below_post_refund"),

pytest.param("post_refund", "above", id="floor_just_above_post_refund"),

pytest.param("pre_refund", "below", id="floor_just_below_pre_refund"),

pytest.param("pre_refund", "above", id="floor_just_above_pre_refund"),

],

)

Not sure the parameter names here are self-explanatory. Perhaps we should name the boundary (pre vs post) and the side (above vs below) more directly?

gurukamath · 2026-10-13T08:42:01Z

-    sender = pre.fund_eoa()
-    write = Op.SSTORE(SLOT_CREATED, 1)
+    approve = Op.APPROVE(0, 0, Spec.APPROVE_EXECUTION_AND_PAYMENT)
+    # A contract sender carries no signature entries, so the charged data


This is not a blocker but more of a nice to have.

Rather than choosing between an EOA and a contract sender here, could the test take sender_kind as a parameter? The floor is composed differently in the two cases, so might be valuable to test both. The EOA case needs a little worker padding so the search precondition holds in no_state_write

PS: for the EOA case, the floor only sees signature bytes after signing.

gurukamath · 2026-10-13T08:54:58Z

+            padding
+        )
+
+    padding = floor_binding_padding(floor_binds)


Nit: the first test calls the data length padding while the refund test uses padding for bytecode and data_length for the data. data_length in both would avoid the overlap

…rker code

… quotient

…or side

…t case

gurukamath

LGTM!

… the minimum data (ethereum#3784) Co-authored-by: Guruprasad Kamath <48196632+gurukamath@users.noreply.github.com> Co-authored-by: 蔡佳誠 Louis Tsai <72684086+LouisTsai-Csie@users.noreply.github.com>

… the minimum data (#3784) Co-authored-by: Guruprasad Kamath <48196632+gurukamath@users.noreply.github.com> Co-authored-by: 蔡佳誠 Louis Tsai <72684086+LouisTsai-Csie@users.noreply.github.com>

refactor(tests): bind the EIP-8141 calldata floor settlement tests by…

b481105

… the minimum data

refactor(tests): bind the EIP-8141 floor settlement test from a contr…

21ab1cc

…act sender

spencer-tb requested review from gurukamath and marioevz October 12, 2026 20:57

spencer-tb mentioned this pull request Oct 12, 2026

Test Release Tracker: bogota-devnet@v1.1.0 #3773

Open

14 of 16 tasks

LouisTsai-Csie self-requested a review October 13, 2026 03:34

LouisTsai-Csie requested changes Oct 13, 2026

View reviewed changes

gurukamath requested changes Oct 13, 2026

View reviewed changes

spencer-tb added 7 commits October 13, 2026 11:25

refactor(tests): derive the EIP-8141 floor test state gas from the wo…

ace1c6b

…rker code

refactor(tests): bound the EIP-8141 refund premise by the fork refund…

6dcf3e9

… quotient

refactor(tests): name the EIP-8141 refund settlement boundary and flo…

338a0b2

…or side

refactor(tests): assert the premise of each EIP-8141 refund settlemen…

32f4632

…t case

refactor(tests): rename the EIP-8141 floor test padding to data_length

810f912

refactor(tests): assert the EIP-8141 calldata floor test premise

a7f19e7

feat(tests): bind the EIP-8141 floor from an EOA and a contract sender

c9b1c5a

gurukamath approved these changes Oct 13, 2026

View reviewed changes

spencer-tb merged commit e7adfc3 into ethereum:eips/bogota/eip-8141 Oct 13, 2026
20 checks passed

Navigation Menu

Repository navigation

refactor(tests): bind the EIP-8141 calldata floor settlement tests by the minimum data - #3784

refactor(tests): bind the EIP-8141 calldata floor settlement tests by the minimum data#3784
spencer-tb merged 9 commits into
ethereum:eips/bogota/eip-8141from
spencer-tb:eip-8141/byte-exact-floor-boundaries

spencer-tb commented Oct 12, 2026 •

edited

Loading

Uh oh!

codecov Bot commented Oct 12, 2026 •

edited

Loading

Uh oh!

LouisTsai-Csie left a comment

Uh oh!

LouisTsai-Csie Oct 13, 2026

Uh oh!

LouisTsai-Csie Oct 13, 2026

Uh oh!

gurukamath Oct 13, 2026

Uh oh!

gurukamath Oct 13, 2026

Uh oh!

gurukamath Oct 13, 2026

Uh oh!

gurukamath Oct 13, 2026

Uh oh!

gurukamath Oct 13, 2026

Uh oh!

gurukamath left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

	state_used = write.state_cost(fork) if with_state_write else 0
	state_used = worker_code.state_cost(fork)

-    gas_used = max(
-        execution_used(padding), floor_calculator(frames=frames(padding))
-    )
+    execution_gas = execution_used(padding)
+    calldata_floor = floor_calculator(frames=frames(padding))
+    # The premise of the test: the floor binds, or in `below_floor`
+    # falls short by the last byte.
+    assert (calldata_floor > execution_gas) == floor_bound
+    gas_used = calldata_floor if floor_bound else execution_gas

-    gas_used_after_refund = gas_used_before_refund - refund
+    gas_used_after_refund = gas_used_before_refund - refund
+    # Assert the premise of each case
+    if against_post_refund and one_byte_short:
+        assert calldata_floor < gas_used_after_refund
+    elif not against_post_refund and not one_byte_short:
+        assert gas_used_before_refund < calldata_floor
+    else:
+        assert gas_used_after_refund < calldata_floor < gas_used_before_refund

Conversation

spencer-tb commented Oct 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Related Issues or PRs

Checklist

Uh oh!

codecov Bot commented Oct 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

LouisTsai-Csie left a comment

Choose a reason for hiding this comment

Uh oh!

LouisTsai-Csie Oct 13, 2026

Choose a reason for hiding this comment

Uh oh!

LouisTsai-Csie Oct 13, 2026

Choose a reason for hiding this comment

Uh oh!

gurukamath Oct 13, 2026

Choose a reason for hiding this comment

Uh oh!

gurukamath Oct 13, 2026

Choose a reason for hiding this comment

Uh oh!

gurukamath Oct 13, 2026

Choose a reason for hiding this comment

Uh oh!

gurukamath Oct 13, 2026

Choose a reason for hiding this comment

Uh oh!

gurukamath Oct 13, 2026

Choose a reason for hiding this comment

Uh oh!

gurukamath left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

spencer-tb commented Oct 12, 2026 •

edited

Loading

codecov Bot commented Oct 12, 2026 •

edited

Loading