nimbus-eth1/nimbus/vm/evmc_host.nim

262 lines
9.5 KiB
Nim
Raw Normal View History

2020-01-16 04:20:13 +00:00
# Nimbus
# Copyright (c) 2019 Status Research & Development GmbH
# Licensed under either of
# * Apache License, version 2.0, ([LICENSE-APACHE](LICENSE-APACHE) or http://www.apache.org/licenses/LICENSE-2.0)
# * MIT license ([LICENSE-MIT](LICENSE-MIT) or http://opensource.org/licenses/MIT)
# at your option. This file may not be copied, modified, or distributed except according to those terms.
2020-01-31 01:08:44 +00:00
proc hostReleaseResultImpl(res: var nimbus_result) {.cdecl, gcsafe.} =
dealloc(res.output_data)
2020-01-16 04:20:13 +00:00
proc hostGetTxContextImpl(ctx: Computation): nimbus_tx_context {.cdecl.} =
2020-01-16 04:20:13 +00:00
let vmstate = ctx.vmState
result.tx_gas_price = toEvmc(vmstate.txGasPrice.u256)
result.tx_origin = vmstate.txOrigin
result.block_coinbase = vmstate.coinbase
2020-01-16 04:20:13 +00:00
result.block_number = vmstate.blockNumber.truncate(int64)
result.block_timestamp = vmstate.timestamp.toUnix()
result.block_gas_limit = int64(vmstate.blockHeader.gasLimit)
result.block_difficulty = toEvmc(vmstate.difficulty)
result.chain_id = toEvmc(vmstate.chaindb.config.chainId.uint.u256)
2020-01-16 04:20:13 +00:00
proc hostGetBlockHashImpl(ctx: Computation, number: int64): Hash256 {.cdecl.} =
ctx.vmState.getAncestorHash(number.u256)
2020-01-16 04:20:13 +00:00
proc hostAccountExistsImpl(ctx: Computation, address: EthAddress): bool {.cdecl.} =
2020-01-17 11:48:14 +00:00
let db = ctx.vmState.readOnlyStateDB
if ctx.fork >= FkSpurious:
not db.isDeadAccount(address)
2020-01-17 11:48:14 +00:00
else:
db.accountExists(address)
2020-01-16 04:20:13 +00:00
proc hostGetStorageImpl(ctx: Computation, address: EthAddress, key: var evmc_bytes32): evmc_bytes32 {.cdecl.} =
2020-07-21 06:15:06 +00:00
ctx.vmState.accountDB.getStorage(address, Uint256.fromEvmc(key)).toEvmc()
2020-01-16 04:20:13 +00:00
2020-07-21 12:58:17 +00:00
proc sstoreNetGasMetering(ctx: Computation): bool {.inline.} =
ctx.fork in {FkConstantinople, FkIstanbul, FkBerlin, FkLondon}
2020-07-21 12:58:17 +00:00
proc hostSetStorageImpl(ctx: Computation, address: EthAddress,
2020-01-16 04:20:13 +00:00
key, value: var evmc_bytes32): evmc_storage_status {.cdecl.} =
let
slot = Uint256.fromEvmc(key)
newValue = Uint256.fromEvmc(value)
statedb = ctx.vmState.readOnlyStateDb
2020-07-21 06:15:06 +00:00
currValue = statedb.getStorage(address, slot)
2020-01-16 04:20:13 +00:00
assert address == ctx.msg.contractAddress
2020-01-16 04:20:13 +00:00
var
status = EVMC_STORAGE_MODIFIED
2020-01-18 00:32:39 +00:00
gasRefund = 0.GasInt
origValue = 0.u256
2020-01-16 04:20:13 +00:00
if newValue == currValue:
status = EVMC_STORAGE_UNCHANGED
else:
origValue = statedb.getCommittedStorage(address, slot)
2020-07-21 12:58:17 +00:00
if origValue == currValue or not ctx.sstoreNetGasMetering():
2020-01-17 14:49:22 +00:00
if currValue == 0:
2020-01-18 00:32:39 +00:00
status = EVMC_STORAGE_ADDED
elif newValue == 0:
status = EVMC_STORAGE_DELETED
else:
status = EVMC_STORAGE_MODIFIED_AGAIN
ctx.vmState.mutateStateDB:
db.setStorage(address, slot, newValue)
2020-01-18 00:32:39 +00:00
let gasParam = GasParams(kind: Op.Sstore,
s_status: status,
s_currentValue: currValue,
s_originalValue: origValue
)
gasRefund = ctx.gasCosts[Sstore].c_handler(newValue, gasParam)[1]
if gasRefund != 0:
2020-01-17 14:49:22 +00:00
ctx.gasMeter.refundGas(gasRefund)
2020-01-16 04:20:13 +00:00
result = status
proc hostGetBalanceImpl(ctx: Computation, address: EthAddress): evmc_bytes32 {.cdecl.} =
ctx.vmState.readOnlyStateDB.getBalance(address).toEvmc()
2020-01-16 04:20:13 +00:00
proc hostGetCodeSizeImpl(ctx: Computation, address: EthAddress): uint {.cdecl.} =
ctx.vmState.readOnlyStateDB.getCode(address).len.uint
2020-01-16 15:48:22 +00:00
proc hostGetCodeHashImpl(ctx: Computation, address: EthAddress): Hash256 {.cdecl.} =
let db = ctx.vmstate.readOnlyStateDB
2020-01-16 15:48:22 +00:00
if not db.accountExists(address):
return
if db.isEmptyAccount(address):
return
db.getCodeHash(address)
2020-01-16 15:48:22 +00:00
proc hostCopyCodeImpl(ctx: Computation, address: EthAddress,
codeOffset: int, bufferData: ptr byte,
bufferSize: int): int {.cdecl.} =
2020-01-16 04:20:13 +00:00
var code = ctx.vmState.readOnlyStateDB.getCode(address)
2020-01-16 04:20:13 +00:00
# Handle "big offset" edge case.
if codeOffset > code.len:
2020-01-16 04:20:13 +00:00
return 0
let maxToCopy = code.len - codeOffset
let numToCopy = min(maxToCopy, bufferSize)
2020-01-16 04:20:13 +00:00
if numToCopy > 0:
2020-07-21 06:15:06 +00:00
copyMem(bufferData, code[codeOffset].addr, numToCopy)
result = numToCopy
2020-01-16 04:20:13 +00:00
proc hostSelfdestructImpl(ctx: Computation, address, beneficiary: EthAddress) {.cdecl.} =
assert address == ctx.msg.contractAddress
ctx.execSelfDestruct(beneficiary)
2020-01-16 04:20:13 +00:00
proc hostEmitLogImpl(ctx: Computation, address: EthAddress,
2020-01-17 13:11:02 +00:00
data: ptr byte, dataSize: int,
topics: UncheckedArray[evmc_bytes32], topicsCount: int) {.cdecl.} =
2020-01-16 04:20:13 +00:00
var log: Log
2020-01-17 13:11:02 +00:00
if topicsCount > 0:
log.topics = newSeq[Topic](topicsCount)
for i in 0 ..< topicsCount:
log.topics[i] = topics[i].bytes
2020-03-09 10:20:08 +00:00
log.data = @(makeOpenArray(data, dataSize))
log.address = address
2020-01-16 04:20:13 +00:00
ctx.addLogEntry(log)
EVMC: Small stacks when using EVMC, closes #575 (segfaults) This patch reduces stack space used with EVM in ENABLE_EVMC=1 mode, from 13 MB worst case to 550 kB, a 24x reduction. This completes fixing the "stack problem" and closes #575 (`EVM: Different segmentation faults when running the test suite with EVMC`). It also closes #256 (`recursive EVM call trigger unrecoverable stack overflow`). After this patch, it is possible to re-enable the CI targets which had to be disabled due to #575. This change is also a required precursor for switching over to "nearly EVMC" as the clean and focused Nimbus-internal API between EVM and sync/database processes, and is also key to the use of Chronos `async` in those processes when calling the EVM. (The motivation is the internal interface has to be substantially changed _anyway_ for the parallel sync and database processes, and EVMC turns out to be well-designed and well-suited for this. It provides good separation between modules, and suits our needs better than our other current interface. Might as well use a good one designed by someone else. EVMC is 98% done in Nimbus thanks to great work done before by @jangko, and we can use Nimbus-specific extensions where we need flexibility, including for performance. Being aligned with the ecosystem is a useful bonus feature.) All tests below were run on Ubuntu 20.04 LTS server, x86-64. This matches one of the targets that has been disabled for a while in CI in EVMC mode due to stack overflow crashing the tests, so it's a good choice. Measurements before =================== Testing commit `e76e0144 2021-04-22 11:29:42 +0700 add submodules: graphql and toml-serialization`. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default to avoid crash. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 38416 depthHigh 3 ... Stack range 13074720 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json These tests use 13.07 MB of stack to run, and so crash with the default stack limit on Ubuntu Server 20.04 (8MB). Exactly 12768 bytes per EVM call stack frame. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 14384 depthHigh 2 ... Stack range 3495456 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 3709600 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 7831600 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stCreate2/Create2OnDepth1024.json These tests use 7.83MB of stack to run. About 7648 bytes per EVM call stack frame. It _only just_ avoids crashing with the default Ubuntu Server stack limit of 8 MB. However, it still crashes on Windows x86-64, which is why the Windows CI EVMC target is currently disabled. On Linux where this passes, this is so borderline that it affects work and testing of the complex storage code, because that's called from the EVM. Also, this greatly exceeds the default thread stack size. Measurements after ================== $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 1936 depthHigh 3 ... Stack range 556272 depthHigh 1022 Stack range 556512 depthHigh 1023 Stack range 556816 depthHigh 1023 Stack range 557056 depthHigh 1024 Stack range 557360 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 1392 depthHigh 2 ... Stack range 248912 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 264144 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 557360 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stStaticCall/static_CallRecursiveBombPreCall.json For both tests, a satisfying *544 bytes* per EVM call stack frame, and EVM takes less than 600 kB total. With other overheads, both tests run in 600 kB stack total at maximum EVM depth. We must add some headroom on this for database activity called from the EVM, and different compile targets. But it means the EVM itself is no longer a stack burden. This is much smaller than the default thread stack size on Linux (2MB), with plenty of margin. (Just fyi, it isn't smaller than a _small_ thread stack on Linux from a long time ago (128kB), and some small embedded C targets.) This size is well suited to running EVMs in threads. Further reduction ================= This patch solves the stack problem. Windows and Linux 64-bit EVMC CI targets can be re-enabled, and there is no longer a problem with stack usage. We can reduce further to ~340 bytes per frame and 350 kB total, while still complying with EVMC. But as this involves changing how errors are handled to comply fully with EVMC, and removing `dispose` calls, it's not worth doing now while there are other EVMC changes in progress that will have the same effect. A Nimbus-specific extension will allow us to avoid recursion with EVMC anyway, bringing bytes per frame to zero. We need the extension anyway, to support Chronos `async` which parallel transaction processing is built around. Interop with non-Nimbus over EVMC won't let us avoid recursion, but then we can't control the stack frame size either. To prevent stack overflow in interop I anticipate using (this method in Aleth) [https://github.com/ethereum/aleth/blob/6e96ce34e3f131e2d42f3cb00741b54e05ab029d/libethereum/ExtVM.cpp#L61]. Smoke test other versions of GCC and Clang/LLVM =============================================== As all builds including Windows use GCC or Apple's Clang/LLVM, this is just to verify we're in the right ballpark on all targets. I've only checked `x86_64` though, not 32-bit, and not ARM. It's interesting to see GCC 10 uses less stack. This is because it optimises `struct` returns better, sometimes skipping an intermediate copy. Here it benefits the EVMC API, but I found GCC 10 also improves the larger stack usage of the rest of `nimbus-eth1` as well. Apple clang 12.0.0 (clang-1200.0.26.2) on MacOS 10.15: - 544 bytes per EVM call stack frame GCC 10.3.0 (Ubuntu 10.3.0-1ubuntu1) on Ubuntu 21.04: - 464 bytes per EVM call stack frame GCC 10.2.0 (Ubuntu 10.2.0-5ubuntu1~20.04) on Ubuntu 20.04 LTS: - 464 bytes per EVM call stack frame GCC 11.0.1 20210417 (experimental; Ubuntu 11-20210417-1ubuntu1) on Ubuntu 21.04: - 8 bytes per EVM call stack frame GCC 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 8.4.0 (Ubuntu 8.4.0-3ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 7.5.0 (Ubuntu 7.5.0-6ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 9.2.1 20191008 (Ubuntu 9.2.1-9ubuntu2) on Ubuntu 19.10: - 528 bytes per EVM call stack frame Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-04-26 15:01:00 +00:00
proc enterCreateImpl(c: Computation, m: nimbus_message): Computation =
2020-01-31 01:08:44 +00:00
# TODO: use evmc_message to evoid copy
2020-02-06 04:37:44 +00:00
let childMsg = Message(
2020-02-03 05:37:33 +00:00
kind: CallKind(m.kind),
2020-01-31 01:08:44 +00:00
depth: m.depth,
gas: m.gas,
sender: m.sender,
2020-02-06 04:37:44 +00:00
value: Uint256.fromEvmc(m.value),
2020-03-09 10:20:08 +00:00
data: @(makeOpenArray(m.inputData, m.inputSize.int))
2020-01-31 01:08:44 +00:00
)
EVMC: Small stacks when using EVMC, closes #575 (segfaults) This patch reduces stack space used with EVM in ENABLE_EVMC=1 mode, from 13 MB worst case to 550 kB, a 24x reduction. This completes fixing the "stack problem" and closes #575 (`EVM: Different segmentation faults when running the test suite with EVMC`). It also closes #256 (`recursive EVM call trigger unrecoverable stack overflow`). After this patch, it is possible to re-enable the CI targets which had to be disabled due to #575. This change is also a required precursor for switching over to "nearly EVMC" as the clean and focused Nimbus-internal API between EVM and sync/database processes, and is also key to the use of Chronos `async` in those processes when calling the EVM. (The motivation is the internal interface has to be substantially changed _anyway_ for the parallel sync and database processes, and EVMC turns out to be well-designed and well-suited for this. It provides good separation between modules, and suits our needs better than our other current interface. Might as well use a good one designed by someone else. EVMC is 98% done in Nimbus thanks to great work done before by @jangko, and we can use Nimbus-specific extensions where we need flexibility, including for performance. Being aligned with the ecosystem is a useful bonus feature.) All tests below were run on Ubuntu 20.04 LTS server, x86-64. This matches one of the targets that has been disabled for a while in CI in EVMC mode due to stack overflow crashing the tests, so it's a good choice. Measurements before =================== Testing commit `e76e0144 2021-04-22 11:29:42 +0700 add submodules: graphql and toml-serialization`. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default to avoid crash. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 38416 depthHigh 3 ... Stack range 13074720 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json These tests use 13.07 MB of stack to run, and so crash with the default stack limit on Ubuntu Server 20.04 (8MB). Exactly 12768 bytes per EVM call stack frame. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 14384 depthHigh 2 ... Stack range 3495456 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 3709600 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 7831600 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stCreate2/Create2OnDepth1024.json These tests use 7.83MB of stack to run. About 7648 bytes per EVM call stack frame. It _only just_ avoids crashing with the default Ubuntu Server stack limit of 8 MB. However, it still crashes on Windows x86-64, which is why the Windows CI EVMC target is currently disabled. On Linux where this passes, this is so borderline that it affects work and testing of the complex storage code, because that's called from the EVM. Also, this greatly exceeds the default thread stack size. Measurements after ================== $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 1936 depthHigh 3 ... Stack range 556272 depthHigh 1022 Stack range 556512 depthHigh 1023 Stack range 556816 depthHigh 1023 Stack range 557056 depthHigh 1024 Stack range 557360 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 1392 depthHigh 2 ... Stack range 248912 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 264144 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 557360 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stStaticCall/static_CallRecursiveBombPreCall.json For both tests, a satisfying *544 bytes* per EVM call stack frame, and EVM takes less than 600 kB total. With other overheads, both tests run in 600 kB stack total at maximum EVM depth. We must add some headroom on this for database activity called from the EVM, and different compile targets. But it means the EVM itself is no longer a stack burden. This is much smaller than the default thread stack size on Linux (2MB), with plenty of margin. (Just fyi, it isn't smaller than a _small_ thread stack on Linux from a long time ago (128kB), and some small embedded C targets.) This size is well suited to running EVMs in threads. Further reduction ================= This patch solves the stack problem. Windows and Linux 64-bit EVMC CI targets can be re-enabled, and there is no longer a problem with stack usage. We can reduce further to ~340 bytes per frame and 350 kB total, while still complying with EVMC. But as this involves changing how errors are handled to comply fully with EVMC, and removing `dispose` calls, it's not worth doing now while there are other EVMC changes in progress that will have the same effect. A Nimbus-specific extension will allow us to avoid recursion with EVMC anyway, bringing bytes per frame to zero. We need the extension anyway, to support Chronos `async` which parallel transaction processing is built around. Interop with non-Nimbus over EVMC won't let us avoid recursion, but then we can't control the stack frame size either. To prevent stack overflow in interop I anticipate using (this method in Aleth) [https://github.com/ethereum/aleth/blob/6e96ce34e3f131e2d42f3cb00741b54e05ab029d/libethereum/ExtVM.cpp#L61]. Smoke test other versions of GCC and Clang/LLVM =============================================== As all builds including Windows use GCC or Apple's Clang/LLVM, this is just to verify we're in the right ballpark on all targets. I've only checked `x86_64` though, not 32-bit, and not ARM. It's interesting to see GCC 10 uses less stack. This is because it optimises `struct` returns better, sometimes skipping an intermediate copy. Here it benefits the EVMC API, but I found GCC 10 also improves the larger stack usage of the rest of `nimbus-eth1` as well. Apple clang 12.0.0 (clang-1200.0.26.2) on MacOS 10.15: - 544 bytes per EVM call stack frame GCC 10.3.0 (Ubuntu 10.3.0-1ubuntu1) on Ubuntu 21.04: - 464 bytes per EVM call stack frame GCC 10.2.0 (Ubuntu 10.2.0-5ubuntu1~20.04) on Ubuntu 20.04 LTS: - 464 bytes per EVM call stack frame GCC 11.0.1 20210417 (experimental; Ubuntu 11-20210417-1ubuntu1) on Ubuntu 21.04: - 8 bytes per EVM call stack frame GCC 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 8.4.0 (Ubuntu 8.4.0-3ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 7.5.0 (Ubuntu 7.5.0-6ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 9.2.1 20191008 (Ubuntu 9.2.1-9ubuntu2) on Ubuntu 19.10: - 528 bytes per EVM call stack frame Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-04-26 15:01:00 +00:00
return newComputation(c.vmState, childMsg, Uint256.fromEvmc(m.create2_salt))
2020-01-31 01:08:44 +00:00
EVMC: Small stacks when using EVMC, closes #575 (segfaults) This patch reduces stack space used with EVM in ENABLE_EVMC=1 mode, from 13 MB worst case to 550 kB, a 24x reduction. This completes fixing the "stack problem" and closes #575 (`EVM: Different segmentation faults when running the test suite with EVMC`). It also closes #256 (`recursive EVM call trigger unrecoverable stack overflow`). After this patch, it is possible to re-enable the CI targets which had to be disabled due to #575. This change is also a required precursor for switching over to "nearly EVMC" as the clean and focused Nimbus-internal API between EVM and sync/database processes, and is also key to the use of Chronos `async` in those processes when calling the EVM. (The motivation is the internal interface has to be substantially changed _anyway_ for the parallel sync and database processes, and EVMC turns out to be well-designed and well-suited for this. It provides good separation between modules, and suits our needs better than our other current interface. Might as well use a good one designed by someone else. EVMC is 98% done in Nimbus thanks to great work done before by @jangko, and we can use Nimbus-specific extensions where we need flexibility, including for performance. Being aligned with the ecosystem is a useful bonus feature.) All tests below were run on Ubuntu 20.04 LTS server, x86-64. This matches one of the targets that has been disabled for a while in CI in EVMC mode due to stack overflow crashing the tests, so it's a good choice. Measurements before =================== Testing commit `e76e0144 2021-04-22 11:29:42 +0700 add submodules: graphql and toml-serialization`. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default to avoid crash. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 38416 depthHigh 3 ... Stack range 13074720 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json These tests use 13.07 MB of stack to run, and so crash with the default stack limit on Ubuntu Server 20.04 (8MB). Exactly 12768 bytes per EVM call stack frame. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 14384 depthHigh 2 ... Stack range 3495456 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 3709600 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 7831600 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stCreate2/Create2OnDepth1024.json These tests use 7.83MB of stack to run. About 7648 bytes per EVM call stack frame. It _only just_ avoids crashing with the default Ubuntu Server stack limit of 8 MB. However, it still crashes on Windows x86-64, which is why the Windows CI EVMC target is currently disabled. On Linux where this passes, this is so borderline that it affects work and testing of the complex storage code, because that's called from the EVM. Also, this greatly exceeds the default thread stack size. Measurements after ================== $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 1936 depthHigh 3 ... Stack range 556272 depthHigh 1022 Stack range 556512 depthHigh 1023 Stack range 556816 depthHigh 1023 Stack range 557056 depthHigh 1024 Stack range 557360 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 1392 depthHigh 2 ... Stack range 248912 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 264144 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 557360 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stStaticCall/static_CallRecursiveBombPreCall.json For both tests, a satisfying *544 bytes* per EVM call stack frame, and EVM takes less than 600 kB total. With other overheads, both tests run in 600 kB stack total at maximum EVM depth. We must add some headroom on this for database activity called from the EVM, and different compile targets. But it means the EVM itself is no longer a stack burden. This is much smaller than the default thread stack size on Linux (2MB), with plenty of margin. (Just fyi, it isn't smaller than a _small_ thread stack on Linux from a long time ago (128kB), and some small embedded C targets.) This size is well suited to running EVMs in threads. Further reduction ================= This patch solves the stack problem. Windows and Linux 64-bit EVMC CI targets can be re-enabled, and there is no longer a problem with stack usage. We can reduce further to ~340 bytes per frame and 350 kB total, while still complying with EVMC. But as this involves changing how errors are handled to comply fully with EVMC, and removing `dispose` calls, it's not worth doing now while there are other EVMC changes in progress that will have the same effect. A Nimbus-specific extension will allow us to avoid recursion with EVMC anyway, bringing bytes per frame to zero. We need the extension anyway, to support Chronos `async` which parallel transaction processing is built around. Interop with non-Nimbus over EVMC won't let us avoid recursion, but then we can't control the stack frame size either. To prevent stack overflow in interop I anticipate using (this method in Aleth) [https://github.com/ethereum/aleth/blob/6e96ce34e3f131e2d42f3cb00741b54e05ab029d/libethereum/ExtVM.cpp#L61]. Smoke test other versions of GCC and Clang/LLVM =============================================== As all builds including Windows use GCC or Apple's Clang/LLVM, this is just to verify we're in the right ballpark on all targets. I've only checked `x86_64` though, not 32-bit, and not ARM. It's interesting to see GCC 10 uses less stack. This is because it optimises `struct` returns better, sometimes skipping an intermediate copy. Here it benefits the EVMC API, but I found GCC 10 also improves the larger stack usage of the rest of `nimbus-eth1` as well. Apple clang 12.0.0 (clang-1200.0.26.2) on MacOS 10.15: - 544 bytes per EVM call stack frame GCC 10.3.0 (Ubuntu 10.3.0-1ubuntu1) on Ubuntu 21.04: - 464 bytes per EVM call stack frame GCC 10.2.0 (Ubuntu 10.2.0-5ubuntu1~20.04) on Ubuntu 20.04 LTS: - 464 bytes per EVM call stack frame GCC 11.0.1 20210417 (experimental; Ubuntu 11-20210417-1ubuntu1) on Ubuntu 21.04: - 8 bytes per EVM call stack frame GCC 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 8.4.0 (Ubuntu 8.4.0-3ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 7.5.0 (Ubuntu 7.5.0-6ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 9.2.1 20191008 (Ubuntu 9.2.1-9ubuntu2) on Ubuntu 19.10: - 528 bytes per EVM call stack frame Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-04-26 15:01:00 +00:00
template leaveCreateImpl(c, child: Computation, res: nimbus_result) =
2020-01-31 01:08:44 +00:00
if not child.shouldBurnGas:
res.gas_left = child.gasMeter.gasRemaining
if child.isSuccess:
c.merge(child)
res.status_code = EVMC_SUCCESS
res.create_address = child.msg.contractAddress
else:
res.status_code = if child.shouldBurnGas: EVMC_FAILURE else: EVMC_REVERT
if child.output.len > 0:
2020-02-06 04:37:44 +00:00
# TODO: can we move the ownership of seq to raw pointer?
2020-01-31 01:08:44 +00:00
res.output_size = child.output.len.uint
res.output_data = cast[ptr byte](alloc(child.output.len))
copyMem(res.output_data, child.output[0].addr, child.output.len)
res.release = hostReleaseResultImpl
EVMC: Small stacks when using EVMC, closes #575 (segfaults) This patch reduces stack space used with EVM in ENABLE_EVMC=1 mode, from 13 MB worst case to 550 kB, a 24x reduction. This completes fixing the "stack problem" and closes #575 (`EVM: Different segmentation faults when running the test suite with EVMC`). It also closes #256 (`recursive EVM call trigger unrecoverable stack overflow`). After this patch, it is possible to re-enable the CI targets which had to be disabled due to #575. This change is also a required precursor for switching over to "nearly EVMC" as the clean and focused Nimbus-internal API between EVM and sync/database processes, and is also key to the use of Chronos `async` in those processes when calling the EVM. (The motivation is the internal interface has to be substantially changed _anyway_ for the parallel sync and database processes, and EVMC turns out to be well-designed and well-suited for this. It provides good separation between modules, and suits our needs better than our other current interface. Might as well use a good one designed by someone else. EVMC is 98% done in Nimbus thanks to great work done before by @jangko, and we can use Nimbus-specific extensions where we need flexibility, including for performance. Being aligned with the ecosystem is a useful bonus feature.) All tests below were run on Ubuntu 20.04 LTS server, x86-64. This matches one of the targets that has been disabled for a while in CI in EVMC mode due to stack overflow crashing the tests, so it's a good choice. Measurements before =================== Testing commit `e76e0144 2021-04-22 11:29:42 +0700 add submodules: graphql and toml-serialization`. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default to avoid crash. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 38416 depthHigh 3 ... Stack range 13074720 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json These tests use 13.07 MB of stack to run, and so crash with the default stack limit on Ubuntu Server 20.04 (8MB). Exactly 12768 bytes per EVM call stack frame. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 14384 depthHigh 2 ... Stack range 3495456 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 3709600 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 7831600 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stCreate2/Create2OnDepth1024.json These tests use 7.83MB of stack to run. About 7648 bytes per EVM call stack frame. It _only just_ avoids crashing with the default Ubuntu Server stack limit of 8 MB. However, it still crashes on Windows x86-64, which is why the Windows CI EVMC target is currently disabled. On Linux where this passes, this is so borderline that it affects work and testing of the complex storage code, because that's called from the EVM. Also, this greatly exceeds the default thread stack size. Measurements after ================== $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 1936 depthHigh 3 ... Stack range 556272 depthHigh 1022 Stack range 556512 depthHigh 1023 Stack range 556816 depthHigh 1023 Stack range 557056 depthHigh 1024 Stack range 557360 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 1392 depthHigh 2 ... Stack range 248912 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 264144 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 557360 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stStaticCall/static_CallRecursiveBombPreCall.json For both tests, a satisfying *544 bytes* per EVM call stack frame, and EVM takes less than 600 kB total. With other overheads, both tests run in 600 kB stack total at maximum EVM depth. We must add some headroom on this for database activity called from the EVM, and different compile targets. But it means the EVM itself is no longer a stack burden. This is much smaller than the default thread stack size on Linux (2MB), with plenty of margin. (Just fyi, it isn't smaller than a _small_ thread stack on Linux from a long time ago (128kB), and some small embedded C targets.) This size is well suited to running EVMs in threads. Further reduction ================= This patch solves the stack problem. Windows and Linux 64-bit EVMC CI targets can be re-enabled, and there is no longer a problem with stack usage. We can reduce further to ~340 bytes per frame and 350 kB total, while still complying with EVMC. But as this involves changing how errors are handled to comply fully with EVMC, and removing `dispose` calls, it's not worth doing now while there are other EVMC changes in progress that will have the same effect. A Nimbus-specific extension will allow us to avoid recursion with EVMC anyway, bringing bytes per frame to zero. We need the extension anyway, to support Chronos `async` which parallel transaction processing is built around. Interop with non-Nimbus over EVMC won't let us avoid recursion, but then we can't control the stack frame size either. To prevent stack overflow in interop I anticipate using (this method in Aleth) [https://github.com/ethereum/aleth/blob/6e96ce34e3f131e2d42f3cb00741b54e05ab029d/libethereum/ExtVM.cpp#L61]. Smoke test other versions of GCC and Clang/LLVM =============================================== As all builds including Windows use GCC or Apple's Clang/LLVM, this is just to verify we're in the right ballpark on all targets. I've only checked `x86_64` though, not 32-bit, and not ARM. It's interesting to see GCC 10 uses less stack. This is because it optimises `struct` returns better, sometimes skipping an intermediate copy. Here it benefits the EVMC API, but I found GCC 10 also improves the larger stack usage of the rest of `nimbus-eth1` as well. Apple clang 12.0.0 (clang-1200.0.26.2) on MacOS 10.15: - 544 bytes per EVM call stack frame GCC 10.3.0 (Ubuntu 10.3.0-1ubuntu1) on Ubuntu 21.04: - 464 bytes per EVM call stack frame GCC 10.2.0 (Ubuntu 10.2.0-5ubuntu1~20.04) on Ubuntu 20.04 LTS: - 464 bytes per EVM call stack frame GCC 11.0.1 20210417 (experimental; Ubuntu 11-20210417-1ubuntu1) on Ubuntu 21.04: - 8 bytes per EVM call stack frame GCC 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 8.4.0 (Ubuntu 8.4.0-3ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 7.5.0 (Ubuntu 7.5.0-6ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 9.2.1 20191008 (Ubuntu 9.2.1-9ubuntu2) on Ubuntu 19.10: - 528 bytes per EVM call stack frame Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-04-26 15:01:00 +00:00
template enterCallImpl(c: Computation, m: nimbus_message): Computation =
2020-02-06 04:37:44 +00:00
let childMsg = Message(
2020-02-03 05:37:33 +00:00
kind: CallKind(m.kind),
depth: m.depth,
gas: m.gas,
sender: m.sender,
codeAddress: m.destination,
contractAddress: if m.kind == EVMC_CALL: m.destination else: c.msg.contractAddress,
value: Uint256.fromEvmc(m.value),
2020-03-09 10:20:08 +00:00
data: @(makeOpenArray(m.inputData, m.inputSize.int)),
2020-02-03 05:37:33 +00:00
flags: MsgFlags(m.flags)
)
EVMC: Small stacks when using EVMC, closes #575 (segfaults) This patch reduces stack space used with EVM in ENABLE_EVMC=1 mode, from 13 MB worst case to 550 kB, a 24x reduction. This completes fixing the "stack problem" and closes #575 (`EVM: Different segmentation faults when running the test suite with EVMC`). It also closes #256 (`recursive EVM call trigger unrecoverable stack overflow`). After this patch, it is possible to re-enable the CI targets which had to be disabled due to #575. This change is also a required precursor for switching over to "nearly EVMC" as the clean and focused Nimbus-internal API between EVM and sync/database processes, and is also key to the use of Chronos `async` in those processes when calling the EVM. (The motivation is the internal interface has to be substantially changed _anyway_ for the parallel sync and database processes, and EVMC turns out to be well-designed and well-suited for this. It provides good separation between modules, and suits our needs better than our other current interface. Might as well use a good one designed by someone else. EVMC is 98% done in Nimbus thanks to great work done before by @jangko, and we can use Nimbus-specific extensions where we need flexibility, including for performance. Being aligned with the ecosystem is a useful bonus feature.) All tests below were run on Ubuntu 20.04 LTS server, x86-64. This matches one of the targets that has been disabled for a while in CI in EVMC mode due to stack overflow crashing the tests, so it's a good choice. Measurements before =================== Testing commit `e76e0144 2021-04-22 11:29:42 +0700 add submodules: graphql and toml-serialization`. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default to avoid crash. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 38416 depthHigh 3 ... Stack range 13074720 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json These tests use 13.07 MB of stack to run, and so crash with the default stack limit on Ubuntu Server 20.04 (8MB). Exactly 12768 bytes per EVM call stack frame. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 14384 depthHigh 2 ... Stack range 3495456 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 3709600 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 7831600 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stCreate2/Create2OnDepth1024.json These tests use 7.83MB of stack to run. About 7648 bytes per EVM call stack frame. It _only just_ avoids crashing with the default Ubuntu Server stack limit of 8 MB. However, it still crashes on Windows x86-64, which is why the Windows CI EVMC target is currently disabled. On Linux where this passes, this is so borderline that it affects work and testing of the complex storage code, because that's called from the EVM. Also, this greatly exceeds the default thread stack size. Measurements after ================== $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 1936 depthHigh 3 ... Stack range 556272 depthHigh 1022 Stack range 556512 depthHigh 1023 Stack range 556816 depthHigh 1023 Stack range 557056 depthHigh 1024 Stack range 557360 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 1392 depthHigh 2 ... Stack range 248912 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 264144 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 557360 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stStaticCall/static_CallRecursiveBombPreCall.json For both tests, a satisfying *544 bytes* per EVM call stack frame, and EVM takes less than 600 kB total. With other overheads, both tests run in 600 kB stack total at maximum EVM depth. We must add some headroom on this for database activity called from the EVM, and different compile targets. But it means the EVM itself is no longer a stack burden. This is much smaller than the default thread stack size on Linux (2MB), with plenty of margin. (Just fyi, it isn't smaller than a _small_ thread stack on Linux from a long time ago (128kB), and some small embedded C targets.) This size is well suited to running EVMs in threads. Further reduction ================= This patch solves the stack problem. Windows and Linux 64-bit EVMC CI targets can be re-enabled, and there is no longer a problem with stack usage. We can reduce further to ~340 bytes per frame and 350 kB total, while still complying with EVMC. But as this involves changing how errors are handled to comply fully with EVMC, and removing `dispose` calls, it's not worth doing now while there are other EVMC changes in progress that will have the same effect. A Nimbus-specific extension will allow us to avoid recursion with EVMC anyway, bringing bytes per frame to zero. We need the extension anyway, to support Chronos `async` which parallel transaction processing is built around. Interop with non-Nimbus over EVMC won't let us avoid recursion, but then we can't control the stack frame size either. To prevent stack overflow in interop I anticipate using (this method in Aleth) [https://github.com/ethereum/aleth/blob/6e96ce34e3f131e2d42f3cb00741b54e05ab029d/libethereum/ExtVM.cpp#L61]. Smoke test other versions of GCC and Clang/LLVM =============================================== As all builds including Windows use GCC or Apple's Clang/LLVM, this is just to verify we're in the right ballpark on all targets. I've only checked `x86_64` though, not 32-bit, and not ARM. It's interesting to see GCC 10 uses less stack. This is because it optimises `struct` returns better, sometimes skipping an intermediate copy. Here it benefits the EVMC API, but I found GCC 10 also improves the larger stack usage of the rest of `nimbus-eth1` as well. Apple clang 12.0.0 (clang-1200.0.26.2) on MacOS 10.15: - 544 bytes per EVM call stack frame GCC 10.3.0 (Ubuntu 10.3.0-1ubuntu1) on Ubuntu 21.04: - 464 bytes per EVM call stack frame GCC 10.2.0 (Ubuntu 10.2.0-5ubuntu1~20.04) on Ubuntu 20.04 LTS: - 464 bytes per EVM call stack frame GCC 11.0.1 20210417 (experimental; Ubuntu 11-20210417-1ubuntu1) on Ubuntu 21.04: - 8 bytes per EVM call stack frame GCC 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 8.4.0 (Ubuntu 8.4.0-3ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 7.5.0 (Ubuntu 7.5.0-6ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 9.2.1 20191008 (Ubuntu 9.2.1-9ubuntu2) on Ubuntu 19.10: - 528 bytes per EVM call stack frame Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-04-26 15:01:00 +00:00
newComputation(c.vmState, childMsg)
2020-02-03 05:37:33 +00:00
EVMC: Small stacks when using EVMC, closes #575 (segfaults) This patch reduces stack space used with EVM in ENABLE_EVMC=1 mode, from 13 MB worst case to 550 kB, a 24x reduction. This completes fixing the "stack problem" and closes #575 (`EVM: Different segmentation faults when running the test suite with EVMC`). It also closes #256 (`recursive EVM call trigger unrecoverable stack overflow`). After this patch, it is possible to re-enable the CI targets which had to be disabled due to #575. This change is also a required precursor for switching over to "nearly EVMC" as the clean and focused Nimbus-internal API between EVM and sync/database processes, and is also key to the use of Chronos `async` in those processes when calling the EVM. (The motivation is the internal interface has to be substantially changed _anyway_ for the parallel sync and database processes, and EVMC turns out to be well-designed and well-suited for this. It provides good separation between modules, and suits our needs better than our other current interface. Might as well use a good one designed by someone else. EVMC is 98% done in Nimbus thanks to great work done before by @jangko, and we can use Nimbus-specific extensions where we need flexibility, including for performance. Being aligned with the ecosystem is a useful bonus feature.) All tests below were run on Ubuntu 20.04 LTS server, x86-64. This matches one of the targets that has been disabled for a while in CI in EVMC mode due to stack overflow crashing the tests, so it's a good choice. Measurements before =================== Testing commit `e76e0144 2021-04-22 11:29:42 +0700 add submodules: graphql and toml-serialization`. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default to avoid crash. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 38416 depthHigh 3 ... Stack range 13074720 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json These tests use 13.07 MB of stack to run, and so crash with the default stack limit on Ubuntu Server 20.04 (8MB). Exactly 12768 bytes per EVM call stack frame. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 14384 depthHigh 2 ... Stack range 3495456 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 3709600 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 7831600 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stCreate2/Create2OnDepth1024.json These tests use 7.83MB of stack to run. About 7648 bytes per EVM call stack frame. It _only just_ avoids crashing with the default Ubuntu Server stack limit of 8 MB. However, it still crashes on Windows x86-64, which is why the Windows CI EVMC target is currently disabled. On Linux where this passes, this is so borderline that it affects work and testing of the complex storage code, because that's called from the EVM. Also, this greatly exceeds the default thread stack size. Measurements after ================== $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 1936 depthHigh 3 ... Stack range 556272 depthHigh 1022 Stack range 556512 depthHigh 1023 Stack range 556816 depthHigh 1023 Stack range 557056 depthHigh 1024 Stack range 557360 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 1392 depthHigh 2 ... Stack range 248912 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 264144 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 557360 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stStaticCall/static_CallRecursiveBombPreCall.json For both tests, a satisfying *544 bytes* per EVM call stack frame, and EVM takes less than 600 kB total. With other overheads, both tests run in 600 kB stack total at maximum EVM depth. We must add some headroom on this for database activity called from the EVM, and different compile targets. But it means the EVM itself is no longer a stack burden. This is much smaller than the default thread stack size on Linux (2MB), with plenty of margin. (Just fyi, it isn't smaller than a _small_ thread stack on Linux from a long time ago (128kB), and some small embedded C targets.) This size is well suited to running EVMs in threads. Further reduction ================= This patch solves the stack problem. Windows and Linux 64-bit EVMC CI targets can be re-enabled, and there is no longer a problem with stack usage. We can reduce further to ~340 bytes per frame and 350 kB total, while still complying with EVMC. But as this involves changing how errors are handled to comply fully with EVMC, and removing `dispose` calls, it's not worth doing now while there are other EVMC changes in progress that will have the same effect. A Nimbus-specific extension will allow us to avoid recursion with EVMC anyway, bringing bytes per frame to zero. We need the extension anyway, to support Chronos `async` which parallel transaction processing is built around. Interop with non-Nimbus over EVMC won't let us avoid recursion, but then we can't control the stack frame size either. To prevent stack overflow in interop I anticipate using (this method in Aleth) [https://github.com/ethereum/aleth/blob/6e96ce34e3f131e2d42f3cb00741b54e05ab029d/libethereum/ExtVM.cpp#L61]. Smoke test other versions of GCC and Clang/LLVM =============================================== As all builds including Windows use GCC or Apple's Clang/LLVM, this is just to verify we're in the right ballpark on all targets. I've only checked `x86_64` though, not 32-bit, and not ARM. It's interesting to see GCC 10 uses less stack. This is because it optimises `struct` returns better, sometimes skipping an intermediate copy. Here it benefits the EVMC API, but I found GCC 10 also improves the larger stack usage of the rest of `nimbus-eth1` as well. Apple clang 12.0.0 (clang-1200.0.26.2) on MacOS 10.15: - 544 bytes per EVM call stack frame GCC 10.3.0 (Ubuntu 10.3.0-1ubuntu1) on Ubuntu 21.04: - 464 bytes per EVM call stack frame GCC 10.2.0 (Ubuntu 10.2.0-5ubuntu1~20.04) on Ubuntu 20.04 LTS: - 464 bytes per EVM call stack frame GCC 11.0.1 20210417 (experimental; Ubuntu 11-20210417-1ubuntu1) on Ubuntu 21.04: - 8 bytes per EVM call stack frame GCC 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 8.4.0 (Ubuntu 8.4.0-3ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 7.5.0 (Ubuntu 7.5.0-6ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 9.2.1 20191008 (Ubuntu 9.2.1-9ubuntu2) on Ubuntu 19.10: - 528 bytes per EVM call stack frame Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-04-26 15:01:00 +00:00
template leaveCallImpl(c, child: Computation, res: nimbus_result) =
2020-02-03 05:37:33 +00:00
if not child.shouldBurnGas:
res.gas_left = child.gasMeter.gasRemaining
if child.isSuccess:
c.merge(child)
res.status_code = EVMC_SUCCESS
else:
res.status_code = if child.shouldBurnGas: EVMC_FAILURE else: EVMC_REVERT
if child.output.len > 0:
2020-02-06 04:37:44 +00:00
# TODO: can we move the ownership of seq to raw pointer?
2020-02-03 05:37:33 +00:00
res.output_size = child.output.len.uint
res.output_data = cast[ptr byte](alloc(child.output.len))
copyMem(res.output_data, child.output[0].addr, child.output.len)
res.release = hostReleaseResultImpl
2020-01-16 04:20:13 +00:00
EVMC: Small stacks when using EVMC, closes #575 (segfaults) This patch reduces stack space used with EVM in ENABLE_EVMC=1 mode, from 13 MB worst case to 550 kB, a 24x reduction. This completes fixing the "stack problem" and closes #575 (`EVM: Different segmentation faults when running the test suite with EVMC`). It also closes #256 (`recursive EVM call trigger unrecoverable stack overflow`). After this patch, it is possible to re-enable the CI targets which had to be disabled due to #575. This change is also a required precursor for switching over to "nearly EVMC" as the clean and focused Nimbus-internal API between EVM and sync/database processes, and is also key to the use of Chronos `async` in those processes when calling the EVM. (The motivation is the internal interface has to be substantially changed _anyway_ for the parallel sync and database processes, and EVMC turns out to be well-designed and well-suited for this. It provides good separation between modules, and suits our needs better than our other current interface. Might as well use a good one designed by someone else. EVMC is 98% done in Nimbus thanks to great work done before by @jangko, and we can use Nimbus-specific extensions where we need flexibility, including for performance. Being aligned with the ecosystem is a useful bonus feature.) All tests below were run on Ubuntu 20.04 LTS server, x86-64. This matches one of the targets that has been disabled for a while in CI in EVMC mode due to stack overflow crashing the tests, so it's a good choice. Measurements before =================== Testing commit `e76e0144 2021-04-22 11:29:42 +0700 add submodules: graphql and toml-serialization`. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default to avoid crash. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 38416 depthHigh 3 ... Stack range 13074720 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json These tests use 13.07 MB of stack to run, and so crash with the default stack limit on Ubuntu Server 20.04 (8MB). Exactly 12768 bytes per EVM call stack frame. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 14384 depthHigh 2 ... Stack range 3495456 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 3709600 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 7831600 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stCreate2/Create2OnDepth1024.json These tests use 7.83MB of stack to run. About 7648 bytes per EVM call stack frame. It _only just_ avoids crashing with the default Ubuntu Server stack limit of 8 MB. However, it still crashes on Windows x86-64, which is why the Windows CI EVMC target is currently disabled. On Linux where this passes, this is so borderline that it affects work and testing of the complex storage code, because that's called from the EVM. Also, this greatly exceeds the default thread stack size. Measurements after ================== $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 1936 depthHigh 3 ... Stack range 556272 depthHigh 1022 Stack range 556512 depthHigh 1023 Stack range 556816 depthHigh 1023 Stack range 557056 depthHigh 1024 Stack range 557360 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 1392 depthHigh 2 ... Stack range 248912 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 264144 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 557360 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stStaticCall/static_CallRecursiveBombPreCall.json For both tests, a satisfying *544 bytes* per EVM call stack frame, and EVM takes less than 600 kB total. With other overheads, both tests run in 600 kB stack total at maximum EVM depth. We must add some headroom on this for database activity called from the EVM, and different compile targets. But it means the EVM itself is no longer a stack burden. This is much smaller than the default thread stack size on Linux (2MB), with plenty of margin. (Just fyi, it isn't smaller than a _small_ thread stack on Linux from a long time ago (128kB), and some small embedded C targets.) This size is well suited to running EVMs in threads. Further reduction ================= This patch solves the stack problem. Windows and Linux 64-bit EVMC CI targets can be re-enabled, and there is no longer a problem with stack usage. We can reduce further to ~340 bytes per frame and 350 kB total, while still complying with EVMC. But as this involves changing how errors are handled to comply fully with EVMC, and removing `dispose` calls, it's not worth doing now while there are other EVMC changes in progress that will have the same effect. A Nimbus-specific extension will allow us to avoid recursion with EVMC anyway, bringing bytes per frame to zero. We need the extension anyway, to support Chronos `async` which parallel transaction processing is built around. Interop with non-Nimbus over EVMC won't let us avoid recursion, but then we can't control the stack frame size either. To prevent stack overflow in interop I anticipate using (this method in Aleth) [https://github.com/ethereum/aleth/blob/6e96ce34e3f131e2d42f3cb00741b54e05ab029d/libethereum/ExtVM.cpp#L61]. Smoke test other versions of GCC and Clang/LLVM =============================================== As all builds including Windows use GCC or Apple's Clang/LLVM, this is just to verify we're in the right ballpark on all targets. I've only checked `x86_64` though, not 32-bit, and not ARM. It's interesting to see GCC 10 uses less stack. This is because it optimises `struct` returns better, sometimes skipping an intermediate copy. Here it benefits the EVMC API, but I found GCC 10 also improves the larger stack usage of the rest of `nimbus-eth1` as well. Apple clang 12.0.0 (clang-1200.0.26.2) on MacOS 10.15: - 544 bytes per EVM call stack frame GCC 10.3.0 (Ubuntu 10.3.0-1ubuntu1) on Ubuntu 21.04: - 464 bytes per EVM call stack frame GCC 10.2.0 (Ubuntu 10.2.0-5ubuntu1~20.04) on Ubuntu 20.04 LTS: - 464 bytes per EVM call stack frame GCC 11.0.1 20210417 (experimental; Ubuntu 11-20210417-1ubuntu1) on Ubuntu 21.04: - 8 bytes per EVM call stack frame GCC 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 8.4.0 (Ubuntu 8.4.0-3ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 7.5.0 (Ubuntu 7.5.0-6ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 9.2.1 20191008 (Ubuntu 9.2.1-9ubuntu2) on Ubuntu 19.10: - 528 bytes per EVM call stack frame Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-04-26 15:01:00 +00:00
proc enterHostCall(c: Computation, msg: var nimbus_message): Computation {.noinline.} =
2020-01-31 01:08:44 +00:00
if msg.kind == EVMC_CREATE or msg.kind == EVMC_CREATE2:
EVMC: Small stacks when using EVMC, closes #575 (segfaults) This patch reduces stack space used with EVM in ENABLE_EVMC=1 mode, from 13 MB worst case to 550 kB, a 24x reduction. This completes fixing the "stack problem" and closes #575 (`EVM: Different segmentation faults when running the test suite with EVMC`). It also closes #256 (`recursive EVM call trigger unrecoverable stack overflow`). After this patch, it is possible to re-enable the CI targets which had to be disabled due to #575. This change is also a required precursor for switching over to "nearly EVMC" as the clean and focused Nimbus-internal API between EVM and sync/database processes, and is also key to the use of Chronos `async` in those processes when calling the EVM. (The motivation is the internal interface has to be substantially changed _anyway_ for the parallel sync and database processes, and EVMC turns out to be well-designed and well-suited for this. It provides good separation between modules, and suits our needs better than our other current interface. Might as well use a good one designed by someone else. EVMC is 98% done in Nimbus thanks to great work done before by @jangko, and we can use Nimbus-specific extensions where we need flexibility, including for performance. Being aligned with the ecosystem is a useful bonus feature.) All tests below were run on Ubuntu 20.04 LTS server, x86-64. This matches one of the targets that has been disabled for a while in CI in EVMC mode due to stack overflow crashing the tests, so it's a good choice. Measurements before =================== Testing commit `e76e0144 2021-04-22 11:29:42 +0700 add submodules: graphql and toml-serialization`. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default to avoid crash. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 38416 depthHigh 3 ... Stack range 13074720 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json These tests use 13.07 MB of stack to run, and so crash with the default stack limit on Ubuntu Server 20.04 (8MB). Exactly 12768 bytes per EVM call stack frame. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 14384 depthHigh 2 ... Stack range 3495456 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 3709600 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 7831600 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stCreate2/Create2OnDepth1024.json These tests use 7.83MB of stack to run. About 7648 bytes per EVM call stack frame. It _only just_ avoids crashing with the default Ubuntu Server stack limit of 8 MB. However, it still crashes on Windows x86-64, which is why the Windows CI EVMC target is currently disabled. On Linux where this passes, this is so borderline that it affects work and testing of the complex storage code, because that's called from the EVM. Also, this greatly exceeds the default thread stack size. Measurements after ================== $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 1936 depthHigh 3 ... Stack range 556272 depthHigh 1022 Stack range 556512 depthHigh 1023 Stack range 556816 depthHigh 1023 Stack range 557056 depthHigh 1024 Stack range 557360 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 1392 depthHigh 2 ... Stack range 248912 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 264144 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 557360 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stStaticCall/static_CallRecursiveBombPreCall.json For both tests, a satisfying *544 bytes* per EVM call stack frame, and EVM takes less than 600 kB total. With other overheads, both tests run in 600 kB stack total at maximum EVM depth. We must add some headroom on this for database activity called from the EVM, and different compile targets. But it means the EVM itself is no longer a stack burden. This is much smaller than the default thread stack size on Linux (2MB), with plenty of margin. (Just fyi, it isn't smaller than a _small_ thread stack on Linux from a long time ago (128kB), and some small embedded C targets.) This size is well suited to running EVMs in threads. Further reduction ================= This patch solves the stack problem. Windows and Linux 64-bit EVMC CI targets can be re-enabled, and there is no longer a problem with stack usage. We can reduce further to ~340 bytes per frame and 350 kB total, while still complying with EVMC. But as this involves changing how errors are handled to comply fully with EVMC, and removing `dispose` calls, it's not worth doing now while there are other EVMC changes in progress that will have the same effect. A Nimbus-specific extension will allow us to avoid recursion with EVMC anyway, bringing bytes per frame to zero. We need the extension anyway, to support Chronos `async` which parallel transaction processing is built around. Interop with non-Nimbus over EVMC won't let us avoid recursion, but then we can't control the stack frame size either. To prevent stack overflow in interop I anticipate using (this method in Aleth) [https://github.com/ethereum/aleth/blob/6e96ce34e3f131e2d42f3cb00741b54e05ab029d/libethereum/ExtVM.cpp#L61]. Smoke test other versions of GCC and Clang/LLVM =============================================== As all builds including Windows use GCC or Apple's Clang/LLVM, this is just to verify we're in the right ballpark on all targets. I've only checked `x86_64` though, not 32-bit, and not ARM. It's interesting to see GCC 10 uses less stack. This is because it optimises `struct` returns better, sometimes skipping an intermediate copy. Here it benefits the EVMC API, but I found GCC 10 also improves the larger stack usage of the rest of `nimbus-eth1` as well. Apple clang 12.0.0 (clang-1200.0.26.2) on MacOS 10.15: - 544 bytes per EVM call stack frame GCC 10.3.0 (Ubuntu 10.3.0-1ubuntu1) on Ubuntu 21.04: - 464 bytes per EVM call stack frame GCC 10.2.0 (Ubuntu 10.2.0-5ubuntu1~20.04) on Ubuntu 20.04 LTS: - 464 bytes per EVM call stack frame GCC 11.0.1 20210417 (experimental; Ubuntu 11-20210417-1ubuntu1) on Ubuntu 21.04: - 8 bytes per EVM call stack frame GCC 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 8.4.0 (Ubuntu 8.4.0-3ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 7.5.0 (Ubuntu 7.5.0-6ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 9.2.1 20191008 (Ubuntu 9.2.1-9ubuntu2) on Ubuntu 19.10: - 528 bytes per EVM call stack frame Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-04-26 15:01:00 +00:00
enterCreateImpl(c, msg)
2020-01-31 01:08:44 +00:00
else:
EVMC: Small stacks when using EVMC, closes #575 (segfaults) This patch reduces stack space used with EVM in ENABLE_EVMC=1 mode, from 13 MB worst case to 550 kB, a 24x reduction. This completes fixing the "stack problem" and closes #575 (`EVM: Different segmentation faults when running the test suite with EVMC`). It also closes #256 (`recursive EVM call trigger unrecoverable stack overflow`). After this patch, it is possible to re-enable the CI targets which had to be disabled due to #575. This change is also a required precursor for switching over to "nearly EVMC" as the clean and focused Nimbus-internal API between EVM and sync/database processes, and is also key to the use of Chronos `async` in those processes when calling the EVM. (The motivation is the internal interface has to be substantially changed _anyway_ for the parallel sync and database processes, and EVMC turns out to be well-designed and well-suited for this. It provides good separation between modules, and suits our needs better than our other current interface. Might as well use a good one designed by someone else. EVMC is 98% done in Nimbus thanks to great work done before by @jangko, and we can use Nimbus-specific extensions where we need flexibility, including for performance. Being aligned with the ecosystem is a useful bonus feature.) All tests below were run on Ubuntu 20.04 LTS server, x86-64. This matches one of the targets that has been disabled for a while in CI in EVMC mode due to stack overflow crashing the tests, so it's a good choice. Measurements before =================== Testing commit `e76e0144 2021-04-22 11:29:42 +0700 add submodules: graphql and toml-serialization`. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default to avoid crash. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 38416 depthHigh 3 ... Stack range 13074720 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json These tests use 13.07 MB of stack to run, and so crash with the default stack limit on Ubuntu Server 20.04 (8MB). Exactly 12768 bytes per EVM call stack frame. $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 16384 # Requires larger stack than default. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 14384 depthHigh 2 ... Stack range 3495456 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 3709600 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 7831600 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stCreate2/Create2OnDepth1024.json These tests use 7.83MB of stack to run. About 7648 bytes per EVM call stack frame. It _only just_ avoids crashing with the default Ubuntu Server stack limit of 8 MB. However, it still crashes on Windows x86-64, which is why the Windows CI EVMC target is currently disabled. On Linux where this passes, this is so borderline that it affects work and testing of the complex storage code, because that's called from the EVM. Also, this greatly exceeds the default thread stack size. Measurements after ================== $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 9 | tee tlog [Suite] persist block json tests ... Stack range 1936 depthHigh 3 ... Stack range 556272 depthHigh 1022 Stack range 556512 depthHigh 1023 Stack range 556816 depthHigh 1023 Stack range 557056 depthHigh 1024 Stack range 557360 depthHigh 1024 [OK] tests/fixtures/PersistBlockTests/block1431916.json $ rm -f build/all_tests && make ENABLE_EVMC=1 test $ ulimit -S -s 600 # Because we can! 600k stack. $ ./build/all_tests 7 | tee tlog [Suite] new generalstate json tests ... Stack range 1392 depthHigh 2 ... Stack range 248912 depthHigh 457 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest639.json ... Stack range 264144 depthHigh 485 [OK] tests/fixtures/eth_tests/GeneralStateTests/stRandom2/randomStatetest458.json ... Stack range 557360 depthHigh 1024 [OK] tests/fixtures/eth_tests/GeneralStateTests/stStaticCall/static_CallRecursiveBombPreCall.json For both tests, a satisfying *544 bytes* per EVM call stack frame, and EVM takes less than 600 kB total. With other overheads, both tests run in 600 kB stack total at maximum EVM depth. We must add some headroom on this for database activity called from the EVM, and different compile targets. But it means the EVM itself is no longer a stack burden. This is much smaller than the default thread stack size on Linux (2MB), with plenty of margin. (Just fyi, it isn't smaller than a _small_ thread stack on Linux from a long time ago (128kB), and some small embedded C targets.) This size is well suited to running EVMs in threads. Further reduction ================= This patch solves the stack problem. Windows and Linux 64-bit EVMC CI targets can be re-enabled, and there is no longer a problem with stack usage. We can reduce further to ~340 bytes per frame and 350 kB total, while still complying with EVMC. But as this involves changing how errors are handled to comply fully with EVMC, and removing `dispose` calls, it's not worth doing now while there are other EVMC changes in progress that will have the same effect. A Nimbus-specific extension will allow us to avoid recursion with EVMC anyway, bringing bytes per frame to zero. We need the extension anyway, to support Chronos `async` which parallel transaction processing is built around. Interop with non-Nimbus over EVMC won't let us avoid recursion, but then we can't control the stack frame size either. To prevent stack overflow in interop I anticipate using (this method in Aleth) [https://github.com/ethereum/aleth/blob/6e96ce34e3f131e2d42f3cb00741b54e05ab029d/libethereum/ExtVM.cpp#L61]. Smoke test other versions of GCC and Clang/LLVM =============================================== As all builds including Windows use GCC or Apple's Clang/LLVM, this is just to verify we're in the right ballpark on all targets. I've only checked `x86_64` though, not 32-bit, and not ARM. It's interesting to see GCC 10 uses less stack. This is because it optimises `struct` returns better, sometimes skipping an intermediate copy. Here it benefits the EVMC API, but I found GCC 10 also improves the larger stack usage of the rest of `nimbus-eth1` as well. Apple clang 12.0.0 (clang-1200.0.26.2) on MacOS 10.15: - 544 bytes per EVM call stack frame GCC 10.3.0 (Ubuntu 10.3.0-1ubuntu1) on Ubuntu 21.04: - 464 bytes per EVM call stack frame GCC 10.2.0 (Ubuntu 10.2.0-5ubuntu1~20.04) on Ubuntu 20.04 LTS: - 464 bytes per EVM call stack frame GCC 11.0.1 20210417 (experimental; Ubuntu 11-20210417-1ubuntu1) on Ubuntu 21.04: - 8 bytes per EVM call stack frame GCC 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 8.4.0 (Ubuntu 8.4.0-3ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 7.5.0 (Ubuntu 7.5.0-6ubuntu2) on Ubuntu 20.04 LTS: - 544 bytes per EVM call stack frame GCC 9.2.1 20191008 (Ubuntu 9.2.1-9ubuntu2) on Ubuntu 19.10: - 528 bytes per EVM call stack frame Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-04-26 15:01:00 +00:00
enterCallImpl(c, msg)
proc leaveHostCall(c, child: Computation, kind: evmc_call_kind): nimbus_result {.noinline.} =
if kind == EVMC_CREATE or kind == EVMC_CREATE2:
leaveCreateImpl(c, child, result)
else:
leaveCallImpl(c, child, result)
proc hostCallImpl(ctx: Computation, msg: var nimbus_message): nimbus_result {.cdecl.} =
let child = enterHostCall(ctx, msg)
child.execCallOrCreate()
leaveHostCall(ctx, child, msg.kind)
2020-01-31 01:08:44 +00:00
2020-01-16 04:20:13 +00:00
proc initHostInterface(): evmc_host_interface =
result.account_exists = cast[evmc_account_exists_fn](hostAccountExistsImpl)
result.get_storage = cast[evmc_get_storage_fn](hostGetStorageImpl)
result.set_storage = cast[evmc_set_storage_fn](hostSetStorageImpl)
result.get_balance = cast[evmc_get_balance_fn](hostGetBalanceImpl)
result.get_code_size = cast[evmc_get_code_size_fn](hostGetCodeSizeImpl)
result.get_code_hash = cast[evmc_get_code_hash_fn](hostGetCodeHashImpl)
result.copy_code = cast[evmc_copy_code_fn](hostCopyCodeImpl)
result.selfdestruct = cast[evmc_selfdestruct_fn](hostSelfdestructImpl)
result.call = cast[evmc_call_fn](hostCallImpl)
result.get_tx_context = cast[evmc_get_tx_context_fn](hostGetTxContextImpl)
result.get_block_hash = cast[evmc_get_block_hash_fn](hostGetBlockHashImpl)
result.emit_log = cast[evmc_emit_log_fn](hostEmitLogImpl)
proc vmSetOptionImpl(vm: ptr evmc_vm, name, value: cstring): evmc_set_option_result {.cdecl.} =
return EVMC_SET_OPTION_INVALID_NAME
proc vmExecuteImpl(vm: ptr evmc_vm, host: ptr evmc_host_interface,
ctx: Computation, rev: evmc_revision,
msg: evmc_message, code: ptr byte, code_size: uint): evmc_result {.cdecl.} =
discard
proc vmGetCapabilitiesImpl(vm: ptr evmc_vm): evmc_capabilities {.cdecl.} =
result.incl(EVMC_CAPABILITY_EVM1)
proc vmDestroyImpl(vm: ptr evmc_vm) {.cdecl.} =
dealloc(vm)
const
EVMC_HOST_NAME = "nimbus_vm"
EVMC_VM_VERSION = "0.0.1"
proc init(vm: var evmc_vm) =
vm.abi_version = EVMC_ABI_VERSION
vm.name = EVMC_HOST_NAME
vm.version = EVMC_VM_VERSION
vm.destroy = vmDestroyImpl
vm.execute = cast[evmc_execute_fn](vmExecuteImpl)
vm.get_capabilities = vmGetCapabilitiesImpl
vm.set_option = vmSetOptionImpl
let gHost = initHostInterface()
proc nim_host_get_interface(): ptr nimbus_host_interface {.exportc, cdecl.} =
result = cast[ptr nimbus_host_interface](gHost.unsafeAddr)
2020-01-16 04:20:13 +00:00
proc nim_host_create_context(vmstate: BaseVmState, msg: ptr evmc_message): Computation {.exportc, cdecl.} =
#result = HostContext(
# vmState: vmstate,
# gasPrice: GasInt(gasPrice),
# origin: fromEvmc(origin)
#)
GC_ref(result)
proc nim_host_destroy_context(ctx: Computation) {.exportc, cdecl.} =
GC_unref(ctx)
proc nim_create_nimbus_vm(): ptr evmc_vm {.exportc, cdecl.} =
result = create(evmc_vm)
init(result[])