core/vm, cmd/eofdump: implement eof validation #30418

holiman · 2024-09-11T08:02:05Z

The bulk of this PR is authored by @lightclient , in the original EOF-work.
More recently, the code has been picked up and reworked for the new EOF specification, by @MariusVanDerWijden , in #29518, and also @shemnon has contributed with fixes.

They are the main authors to be credited for the code in this PR.

This PR is an attempt to start eating the elephant one small bite at a time, by selecting only the eof-validation as a standalone piece which can be merged without interfering too much in the core stuff.

In this PR:

Validation of eof containers, lifted from WIP mega eof #29518, along with test-vectors from consensus-tests and fuzzing, to ensure that the move did not lose any functionality.
Definition of eof opcodes, which is a prerequisite for validation
Addition of undefined to a jumptable entry item. I'm not super-happy with this, but for the moment it seems the least invasive way to do it. A better way might be to go back and allowing nil-items or nil execute-functions to denote "undefined".
Gas-calculators. These do not strictly need to be here, I can remove them if we want to reduce the diff even further.
benchmarks of eof validation speed

It would have been nice to break it up further, core/vm is growing very very large. However, it's not easy to put eof in a submodule.

eof validation requires at least the opcodes, and certain opcode-specific info (e.g. stack usage)
interpreter requires eof to tell it to validate

So what we might do is move the opcodes, and the jumptable definition to a new low-level module. And in core/vm keep the jumptable instantiations, so we can pass the actual jumptables to eof. Then eof doesn't have to import core/vm, I think.

Even with this smaller piece cherry-picked, there's quite a bit of work to do to improve it, better documentation and test coverage. I'll push commits here and rebase quite a bit.

To do in follow-up PR(s) is to modify the interpreter a bit more in-depth, to operate differently in eof-mode versus non-eof-mode.

Co-authored-by: @lightclient
Co-authored-by: @MariusVanDerWijden
Co-authored-by: @shemnon

holiman · 2024-09-11T08:09:41Z

core/vm/jump_table.go

+		INVALID: {
+			execute:  opUndefined,
+			minStack: minStack(0, 0),
+			maxStack: maxStack(0, 0),
+		},


This is IMO very wonky, see discussion here: https://github.com/ethereum/go-ethereum/pull/29518/files#r1753395744 .

holiman · 2024-09-11T08:10:44Z

core/vm/validate.go

+	ErrUndefinedInstruction     = errors.New("undefined instruction")
+	ErrTruncatedImmediate       = errors.New("truncated immediate")
+	ErrInvalidSectionArgument   = errors.New("invalid section argument")
+	ErrInvalidCallArgument      = errors.New("callf into non-returning section")
+	ErrInvalidDataloadNArgument = errors.New("invalid dataloadN argument")
+	ErrInvalidJumpDest          = errors.New("invalid jump destination")
+	ErrInvalidBackwardJump      = errors.New("invalid backward jump")
+	//ErrConflictingStack              = errors.New("conflicting stack height")
+	//ErrInvalidBranchCount            = errors.New("invalid number of branches in jump table")
+	ErrInvalidOutputs                = errors.New("invalid number of outputs")
+	ErrInvalidMaxStackHeight         = errors.New("invalid max stack height")
+	ErrInvalidCodeTermination        = errors.New("invalid code termination")
+	ErrEOFCreateWithTruncatedSection = errors.New("eofcreate with truncated section")
+	ErrOrphanedSubcontainer          = errors.New("subcontainer not referenced at all")
+	ErrIncompatibleContainerKind     = errors.New("incompatible container kind")
+	ErrStopAndReturnContract         = errors.New("Stop/Return and Returncontract in the same code section")
+	ErrStopInInitCode                = errors.New("initcode contains a RETURN or STOP opcode")
+	ErrTruncatedTopLevelContainer    = errors.New("truncated top level container")
+	ErrUnreachableCode               = errors.New("unreachable code")
+	ErrInvalidNonReturningFlag       = errors.New("invalid non-returning flag, bad RETF")


Not all of these methods are mapped in the eofparser error-to-number mapping function. Should they be?

core/vm/validate.go

core/vm/memory_table.go

holiman · 2024-09-11T09:13:30Z

core/vm/validate.go

+	ErrTooManyInputs                 = errors.New("invalid type content, too many inputs")
+	ErrTooManyOutputs                = errors.New("invalid type content, too many outputs")


Why are these needed, when we already have ErrInvalidInputs/Outputs?

core/vm/validate.go

holiman · 2024-09-11T11:03:03Z

Added benchmarks, based on some of the worst-cases from the consensus-tests + fuzzing vectors.

goos: linux
goarch: amd64
pkg: github.com/ethereum/go-ethereum/cmd/eofdump
cpu: 12th Gen Intel(R) Core(TM) i7-1270P
BenchmarkEofParse
BenchmarkEofParse/test-1
BenchmarkEofParse/test-1-8               1043824              1145 ns/op          17.47 MB/s         352 B/op          9 allocs/op
BenchmarkEofParse/test-2
BenchmarkEofParse/test-2-8                717020              1526 ns/op          13.77 MB/s         416 B/op         12 allocs/op
BenchmarkEofParse/test-3
BenchmarkEofParse/test-3-8                645865              1830 ns/op          12.56 MB/s         464 B/op         16 allocs/op
BenchmarkEofParse/test-4
BenchmarkEofParse/test-4-8                666870              1875 ns/op          12.27 MB/s         464 B/op         16 allocs/op
BenchmarkEofParse/test-5
BenchmarkEofParse/test-5-8                614787              2180 ns/op          11.01 MB/s         488 B/op         18 allocs/op
BenchmarkEofParse/test-6
BenchmarkEofParse/test-6-8                337408              3773 ns/op          21.20 MB/s        1064 B/op         35 allocs/op
BenchmarkEofParse/test-7
BenchmarkEofParse/test-7-8                   282           4462357 ns/op          11.00 MB/s     1058652 B/op      38282 allocs/op
BenchmarkEofParse/test-8
BenchmarkEofParse/test-8-8                342518              4078 ns/op          15.69 MB/s         934 B/op         35 allocs/op
BenchmarkEofParse/test-9
BenchmarkEofParse/test-9-8                301110              3716 ns/op          17.22 MB/s         934 B/op         35 allocs/op
BenchmarkEofParse/test-10
BenchmarkEofParse/test-10-8               288266              4832 ns/op           6.83 MB/s        1031 B/op         37 allocs/op
BenchmarkEofParse/test-11
BenchmarkEofParse/test-11-8               277754              4582 ns/op           7.20 MB/s        1031 B/op         37 allocs/op
BenchmarkEofParse/test-12
BenchmarkEofParse/test-12-8                  632           1874380 ns/op          13.11 MB/s      453325 B/op      18881 allocs/op
BenchmarkEofParse/test-13
BenchmarkEofParse/test-13-8               185851              6186 ns/op           5.50 MB/s        1696 B/op         40 allocs/op
BenchmarkEofParse/test-14
BenchmarkEofParse/test-14-8                97939             13215 ns/op           4.09 MB/s        3557 B/op         80 allocs/op
BenchmarkEofParse/test-15
BenchmarkEofParse/test-15-8               104028             13065 ns/op          11.48 MB/s        3557 B/op         80 allocs/op
BenchmarkEofParse/test-16
BenchmarkEofParse/test-16-8                36811             36740 ns/op          14.59 MB/s       12777 B/op        279 allocs/op
BenchmarkEofParse/test-17
BenchmarkEofParse/test-17-8                 9140            117813 ns/op           6.25 MB/s       30370 B/op        739 allocs/op
BenchmarkEofParse/test-18
BenchmarkEofParse/test-18-8                 3262            368707 ns/op           2.83 MB/s      110391 B/op       2091 allocs/op
BenchmarkEofParse/test-19
BenchmarkEofParse/test-19-8               720025              1415 ns/op          22.62 MB/s         408 B/op         12 allocs/op
PASS
ok      github.com/ethereum/go-ethereum/cmd/eofdump     31.290s

holiman · 2024-09-11T13:04:50Z

Sped it up quite a bit, the slowest vector is now improved by a factor of 10 :)

goos: linux
goarch: amd64
pkg: github.com/ethereum/go-ethereum/cmd/eofdump
cpu: 12th Gen Intel(R) Core(TM) i7-1270P
                   │  eof.bench.1  │             eof.bench.2             │
                   │    sec/op     │    sec/op     vs base               │
EofParse/test-1-8     1.062µ ± ∞ ¹   1.105µ ± ∞ ¹        ~ (p=0.151 n=5)
EofParse/test-2-8     1.451µ ± ∞ ¹   1.442µ ± ∞ ¹        ~ (p=0.151 n=5)
EofParse/test-3-8     1.770µ ± ∞ ¹   1.533µ ± ∞ ¹  -13.39% (p=0.008 n=5)
EofParse/test-4-8     1.775µ ± ∞ ¹   1.550µ ± ∞ ¹  -12.68% (p=0.008 n=5)
EofParse/test-5-8     2.035µ ± ∞ ¹   1.631µ ± ∞ ¹  -19.85% (p=0.008 n=5)
EofParse/test-6-8     3.484µ ± ∞ ¹   3.039µ ± ∞ ¹  -12.77% (p=0.008 n=5)
EofParse/test-7-8     3.990m ± ∞ ¹   2.896m ± ∞ ¹  -27.43% (p=0.008 n=5)
EofParse/test-8-8     3.868µ ± ∞ ¹   1.225µ ± ∞ ¹  -68.33% (p=0.008 n=5)
EofParse/test-9-8     3.765µ ± ∞ ¹   1.170µ ± ∞ ¹  -68.92% (p=0.008 n=5)
EofParse/test-10-8    4.252µ ± ∞ ¹   1.838µ ± ∞ ¹  -56.77% (p=0.008 n=5)
EofParse/test-11-8    4.727µ ± ∞ ¹   1.829µ ± ∞ ¹  -61.31% (p=0.008 n=5)
EofParse/test-12-8    1.745m ± ∞ ¹   1.106m ± ∞ ¹  -36.63% (p=0.008 n=5)
EofParse/test-13-8    5.778µ ± ∞ ¹   1.836µ ± ∞ ¹  -68.22% (p=0.008 n=5)
EofParse/test-14-8   12.637µ ± ∞ ¹   2.335µ ± ∞ ¹  -81.52% (p=0.008 n=5)
EofParse/test-15-8   12.224µ ± ∞ ¹   2.856µ ± ∞ ¹  -76.64% (p=0.008 n=5)
EofParse/test-16-8    33.30µ ± ∞ ¹   15.61µ ± ∞ ¹  -53.12% (p=0.008 n=5)
EofParse/test-17-8   106.47µ ± ∞ ¹   11.42µ ± ∞ ¹  -89.27% (p=0.008 n=5)
EofParse/test-18-8   337.85µ ± ∞ ¹   27.20µ ± ∞ ¹  -91.95% (p=0.008 n=5)
EofParse/test-19-8    1.368µ ± ∞ ¹   1.365µ ± ∞ ¹        ~ (p=0.889 n=5)
geomean               11.30µ         5.055µ        -55.27%
¹ need >= 6 samples for confidence interval at level 0.95

                   │  eof.bench.1  │               eof.bench.2               │
                   │      B/s      │      B/s        vs base                 │
EofParse/test-1-8    17.96Mi ± ∞ ¹    17.27Mi ± ∞ ¹          ~ (p=0.151 n=5)
EofParse/test-2-8    13.80Mi ± ∞ ¹    13.89Mi ± ∞ ¹          ~ (p=0.151 n=5)
EofParse/test-3-8    12.39Mi ± ∞ ¹    14.31Mi ± ∞ ¹    +15.47% (p=0.008 n=5)
EofParse/test-4-8    12.36Mi ± ∞ ¹    14.15Mi ± ∞ ¹    +14.51% (p=0.008 n=5)
EofParse/test-5-8    11.24Mi ± ∞ ¹    14.04Mi ± ∞ ¹    +24.85% (p=0.008 n=5)
EofParse/test-6-8    21.90Mi ± ∞ ¹    25.11Mi ± ∞ ¹    +14.68% (p=0.008 n=5)
EofParse/test-7-8    11.74Mi ± ∞ ¹    16.17Mi ± ∞ ¹    +37.77% (p=0.008 n=5)
EofParse/test-8-8    15.78Mi ± ∞ ¹    49.84Mi ± ∞ ¹   +215.77% (p=0.008 n=5)
EofParse/test-9-8    16.21Mi ± ∞ ¹    52.18Mi ± ∞ ¹   +221.82% (p=0.008 n=5)
EofParse/test-10-8   7.401Mi ± ∞ ¹   17.118Mi ± ∞ ¹   +131.31% (p=0.008 n=5)
EofParse/test-11-8   6.657Mi ± ∞ ¹   17.204Mi ± ∞ ¹   +158.45% (p=0.008 n=5)
EofParse/test-12-8   13.43Mi ± ∞ ¹    21.20Mi ± ∞ ¹    +57.88% (p=0.008 n=5)
EofParse/test-13-8   5.608Mi ± ∞ ¹   17.662Mi ± ∞ ¹   +214.97% (p=0.008 n=5)
EofParse/test-14-8   4.072Mi ± ∞ ¹   22.058Mi ± ∞ ¹   +441.69% (p=0.008 n=5)
EofParse/test-15-8   11.70Mi ± ∞ ¹    50.09Mi ± ∞ ¹   +328.04% (p=0.008 n=5)
EofParse/test-16-8   15.35Mi ± ∞ ¹    32.74Mi ± ∞ ¹   +113.23% (p=0.008 n=5)
EofParse/test-17-8   6.590Mi ± ∞ ¹   61.455Mi ± ∞ ¹   +832.56% (p=0.008 n=5)
EofParse/test-18-8   2.947Mi ± ∞ ¹   36.602Mi ± ∞ ¹  +1142.07% (p=0.008 n=5)
EofParse/test-19-8   22.31Mi ± ∞ ¹    22.36Mi ± ∞ ¹          ~ (p=0.841 n=5)
geomean              10.65Mi          23.81Mi         +123.64%
¹ need >= 6 samples for confidence interval at level 0.95

                   │   eof.bench.1   │              eof.bench.2               │
                   │      B/op       │     B/op       vs base                 │
EofParse/test-1-8        352.0 ± ∞ ¹     352.0 ± ∞ ¹        ~ (p=1.000 n=5) ²
EofParse/test-2-8        416.0 ± ∞ ¹     416.0 ± ∞ ¹        ~ (p=1.000 n=5) ²
EofParse/test-3-8        464.0 ± ∞ ¹     424.0 ± ∞ ¹   -8.62% (p=0.008 n=5)
EofParse/test-4-8        464.0 ± ∞ ¹     424.0 ± ∞ ¹   -8.62% (p=0.008 n=5)
EofParse/test-5-8        488.0 ± ∞ ¹     440.0 ± ∞ ¹   -9.84% (p=0.008 n=5)
EofParse/test-6-8       1064.0 ± ∞ ¹    1016.0 ± ∞ ¹   -4.51% (p=0.008 n=5)
EofParse/test-7-8     1033.8Ki ± ∞ ¹   910.3Ki ± ∞ ¹  -11.95% (p=0.008 n=5)
EofParse/test-8-8        934.0 ± ∞ ¹     368.0 ± ∞ ¹  -60.60% (p=0.008 n=5)
EofParse/test-9-8        934.0 ± ∞ ¹     368.0 ± ∞ ¹  -60.60% (p=0.008 n=5)
EofParse/test-10-8      1031.0 ± ∞ ¹     472.0 ± ∞ ¹  -54.22% (p=0.008 n=5)
EofParse/test-11-8      1031.0 ± ∞ ¹     472.0 ± ∞ ¹  -54.22% (p=0.008 n=5)
EofParse/test-12-8     442.7Ki ± ∞ ¹   369.0Ki ± ∞ ¹  -16.64% (p=0.008 n=5)
EofParse/test-13-8      1697.0 ± ∞ ¹     472.0 ± ∞ ¹  -72.19% (p=0.008 n=5)
EofParse/test-14-8      3557.0 ± ∞ ¹     568.0 ± ∞ ¹  -84.03% (p=0.008 n=5)
EofParse/test-15-8      3557.0 ± ∞ ¹     984.0 ± ∞ ¹  -72.34% (p=0.008 n=5)
EofParse/test-16-8     12.48Ki ± ∞ ¹   10.69Ki ± ∞ ¹  -14.34% (p=0.008 n=5)
EofParse/test-17-8    29.664Ki ± ∞ ¹   3.297Ki ± ∞ ¹  -88.89% (p=0.008 n=5)
EofParse/test-18-8   107.798Ki ± ∞ ¹   4.789Ki ± ∞ ¹  -95.56% (p=0.008 n=5)
EofParse/test-19-8       408.0 ± ∞ ¹     408.0 ± ∞ ¹        ~ (p=1.000 n=5) ²
geomean                3.065Ki         1.472Ki        -51.96%
¹ need >= 6 samples for confidence interval at level 0.95
² all samples are equal

                   │  eof.bench.1  │              eof.bench.2              │
                   │   allocs/op   │  allocs/op    vs base                 │
EofParse/test-1-8      9.000 ± ∞ ¹    9.000 ± ∞ ¹        ~ (p=1.000 n=5) ²
EofParse/test-2-8      12.00 ± ∞ ¹    14.00 ± ∞ ¹  +16.67% (p=0.008 n=5)
EofParse/test-3-8      16.00 ± ∞ ¹    14.00 ± ∞ ¹  -12.50% (p=0.008 n=5)
EofParse/test-4-8      16.00 ± ∞ ¹    14.00 ± ∞ ¹  -12.50% (p=0.008 n=5)
EofParse/test-5-8      18.00 ± ∞ ¹    14.00 ± ∞ ¹  -22.22% (p=0.008 n=5)
EofParse/test-6-8      35.00 ± ∞ ¹    31.00 ± ∞ ¹  -11.43% (p=0.008 n=5)
EofParse/test-7-8     38.28k ± ∞ ¹   28.30k ± ∞ ¹  -26.08% (p=0.008 n=5)
EofParse/test-8-8      35.00 ± ∞ ¹    11.00 ± ∞ ¹  -68.57% (p=0.008 n=5)
EofParse/test-9-8      35.00 ± ∞ ¹    11.00 ± ∞ ¹  -68.57% (p=0.008 n=5)
EofParse/test-10-8     37.00 ± ∞ ¹    14.00 ± ∞ ¹  -62.16% (p=0.008 n=5)
EofParse/test-11-8     37.00 ± ∞ ¹    14.00 ± ∞ ¹  -62.16% (p=0.008 n=5)
EofParse/test-12-8    18.88k ± ∞ ¹   10.71k ± ∞ ¹  -43.29% (p=0.008 n=5)
EofParse/test-13-8     40.00 ± ∞ ¹    14.00 ± ∞ ¹  -65.00% (p=0.008 n=5)
EofParse/test-14-8     80.00 ± ∞ ¹    14.00 ± ∞ ¹  -82.50% (p=0.008 n=5)
EofParse/test-15-8     80.00 ± ∞ ¹    14.00 ± ∞ ¹  -82.50% (p=0.008 n=5)
EofParse/test-16-8    279.00 ± ∞ ¹    22.00 ± ∞ ¹  -92.11% (p=0.008 n=5)
EofParse/test-17-8    739.00 ± ∞ ¹    11.00 ± ∞ ¹  -98.51% (p=0.008 n=5)
EofParse/test-18-8   2091.00 ± ∞ ¹    11.00 ± ∞ ¹  -99.47% (p=0.008 n=5)
EofParse/test-19-8     12.00 ± ∞ ¹    12.00 ± ∞ ¹        ~ (p=1.000 n=5) ²
geomean                93.51          29.15        -68.82%
¹ need >= 6 samples for confidence interval at level 0.95
² all samples are equal

cmd/eofdump: benchmarks of eof validation speeds core/vm: move eof instructions to separate file core/vm: unexport fields Co-authored-by: lightclient <lightclient@protonmail.com> Co-authored-by: Marius van der Wijden <m.vanderwijden@live.de> Co-authored-by: Danno Ferrin <danno.ferrin@shemnon.com>

holiman · 2024-09-12T08:11:39Z

core/vm/validate_test.go

+	}
+}
+
+func BenchmarkEOFValidation2(b *testing.B) {


All the benchmark-function in this file appears to be non-functional, should we nuke them or salvage them, @MariusVanDerWijden ?

example

[user@work vm]$ go test . -run - -bench EOFV --- FAIL: BenchmarkEOFValidation validate_test.go:353: callf into non-returning section: section 1 --- FAIL: BenchmarkEOFValidation2 validate_test.go:403: callf into non-returning section: section 1 --- FAIL: BenchmarkEOFValidation3 validate_test.go:452: callf into non-returning section: section 1

I will try to salvage

core/vm: some clarifications in the eof code core/vm: clarifications + minor speedup core/vm: clarifications + lint + minor speedup core/vm, core/asm: support eof in asm instruction iteration core/vm: comment out unused core/vm: remove gasfunctions

holiman requested review from karalabe and rjl493456442 as code owners September 11, 2024 08:02

holiman force-pushed the eof-validation branch from e1feb9c to 6ebbbd9 Compare September 11, 2024 08:07

holiman commented Sep 11, 2024

View reviewed changes

core/vm/validate.go Outdated Show resolved Hide resolved

holiman force-pushed the eof-validation branch from 6ebbbd9 to 666abfe Compare September 11, 2024 08:26

holiman commented Sep 11, 2024

View reviewed changes

core/vm/memory_table.go Outdated Show resolved Hide resolved

holiman force-pushed the eof-validation branch from 666abfe to 6cfe117 Compare September 11, 2024 08:31

holiman commented Sep 11, 2024

View reviewed changes

core/vm/validate.go Outdated Show resolved Hide resolved

holiman force-pushed the eof-validation branch from a80b89a to 5099626 Compare September 11, 2024 11:20

holiman force-pushed the eof-validation branch from 33cc41e to 8196b46 Compare September 12, 2024 06:59

holiman commented Sep 12, 2024

View reviewed changes

holiman added the prague label Sep 16, 2024

holiman force-pushed the eof-validation branch from e262a9f to 50dcc67 Compare September 16, 2024 06:44

holiman force-pushed the eof-validation branch from 50dcc67 to 8783059 Compare September 16, 2024 08:45

MariusVanDerWijden added 2 commits September 16, 2024 11:41

core/vm: remove unnecessary check

34dff6a

core/vm: fix benchmarks

84591bf

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

core/vm, cmd/eofdump: implement eof validation #30418

core/vm, cmd/eofdump: implement eof validation #30418

holiman commented Sep 11, 2024 •

edited

Loading

holiman Sep 11, 2024

holiman Sep 11, 2024

holiman Sep 11, 2024

holiman commented Sep 11, 2024

holiman commented Sep 11, 2024 •

edited

Loading

holiman Sep 12, 2024

MariusVanDerWijden Sep 16, 2024

		ErrTooManyInputs = errors.New("invalid type content, too many inputs")
		ErrTooManyOutputs = errors.New("invalid type content, too many outputs")

core/vm, cmd/eofdump: implement eof validation #30418

Are you sure you want to change the base?

core/vm, cmd/eofdump: implement eof validation #30418

Conversation

holiman commented Sep 11, 2024 • edited Loading

holiman Sep 11, 2024

Choose a reason for hiding this comment

holiman Sep 11, 2024

Choose a reason for hiding this comment

holiman Sep 11, 2024

Choose a reason for hiding this comment

holiman commented Sep 11, 2024

holiman commented Sep 11, 2024 • edited Loading

holiman Sep 12, 2024

Choose a reason for hiding this comment

MariusVanDerWijden Sep 16, 2024

Choose a reason for hiding this comment

holiman commented Sep 11, 2024 •

edited

Loading

holiman commented Sep 11, 2024 •

edited

Loading