This PR is a more advanced form of the dirty-to-clean cacher (#18995),
where we reuse previous database write batches as datasets to uncache,
saving a dirty-trie-iteration and a dirty-trie-rlp-reencoding per block.
This change
- implements concurrent LES request serving even for a single peer.
- replaces the request cost estimation method with a cost table based on
benchmarks which gives much more consistent results. Until now the
allowed number of light peers was just a guess which probably contributed
a lot to the fluctuating quality of available service. Everything related
to request cost is implemented in a single object, the 'cost tracker'. It
uses a fixed cost table with a global 'correction factor'. Benchmark code
is included and can be run at any time to adapt costs to low-level
implementation changes.
- reimplements flowcontrol.ClientManager in a cleaner and more efficient
way, with added capabilities: There is now control over bandwidth, which
allows using the flow control parameters for client prioritization.
Target utilization over 100 percent is now supported to model concurrent
request processing. Total serving bandwidth is reduced during block
processing to prevent database contention.
- implements an RPC API for the LES servers allowing server operators to
assign priority bandwidth to certain clients and change prioritized
status even while the client is connected. The new API is meant for
cases where server operators charge for LES using an off-protocol mechanism.
- adds a unit test for the new client manager.
- adds an end-to-end test using the network simulator that tests bandwidth
control functions through the new API.
This change unbreaks the build and removes racy access to
disableCheckFreq. Even though the field is set while holding
the lock, it was read outside of the protected section.
TryUpdate does not call t.trie.TryUpdate(key, value) and calls t.trie.TryDelete
instead. The update operation simply deletes the corresponding entry, though
it could retrieve later by odr. However, it adds further network overhead.
This PR enables the indexers to work in light client mode by
downloading a part of these tries (the Merkle proofs of the last
values of the last known section) in order to be able to add new
values and recalculate subsequent hashes. It also adds CHT data to
NodeInfo.
* les, light: fix CHT trie retrievals
* les, light: minor polishes, test remote CHT retrievals
* les, light: deterministic nodeset rlp, bloombits test skeleton
* les: add an event emission to the les bloombits test
* les: drop dead tester code
This PR implements the new LES protocol version extensions:
* new and more efficient Merkle proofs reply format (when replying to
a multiple Merkle proofs request, we just send a single set of trie
nodes containing all necessary nodes)
* BBT (BloomBitsTrie) works similarly to the existing CHT and contains
the bloombits search data to speed up log searches
* GetTxStatusMsg returns the inclusion position or the
pending/queued/unknown state of a transaction referenced by hash
* an optional signature of new block data (number/hash/td) can be
included in AnnounceMsg to provide an option for "very light
clients" (mobile/embedded devices) to skip expensive Ethash check
and accept multiple signatures of somewhat trusted servers (still a
lot better than trusting a single server completely and retrieving
everything through RPC). The new client mode is not implemented in
this PR, just the protocol extension.
* ethdb: add Putter interface and Has method
* ethdb: improve docs and add IdealBatchSize
* ethdb: remove memory batch lock
Batches are not safe for concurrent use.
* core: use ethdb.Putter for Write* functions
This covers the easy cases.
* core/state: simplify StateSync
* trie: optimize local node check
* ethdb: add ValueSize to Batch
* core: optimize HasHeader check
This avoids one random database read get the block number. For many uses
of HasHeader, the expectation is that it's actually there. Using Has
avoids a load + decode of the value.
* core: write fast sync block data in batches
Collect writes into batches up to the ideal size instead of issuing many
small, concurrent writes.
* eth/downloader: commit larger state batches
Collect nodes into a batch up to the ideal size instead of committing
whenever a node is received.
* core: optimize HasBlock check
This avoids a random database read to get the number.
* core: use numberCache in HasHeader
numberCache has higher capacity, increasing the odds of finding the
header without a database lookup.
* core: write imported block data using a batch
Restore batch writes of state and add blocks, tx entries, receipts to
the same batch. The change also simplifies the miner.
This commit also removes posting of logs when a forked block is imported.
* core: fix DB write error handling
* ethdb: use RLock for Has
* core: fix HasBlock comment
* core: remove redundant storage of transactions and receipts
* core, eth, internal: new transaction schema usage polishes
* eth: implement upgrade mechanism for db deduplication
* core, eth: drop old sequential key db upgrader
* eth: close last iterator on successful db upgrage
* core: prefix the lookup entries to make their purpose clearer
With this commit, core/state's access to the underlying key/value database is
mediated through an interface. Database errors are tracked in StateDB and
returned by CommitTo or the new Error method.
Motivation for this change: We can remove the light client's duplicated copy of
core/state. The light client now supports node iteration, so tracing and storage
enumeration can work with the light client (not implemented in this commit).
The key was constructed from nibbles, which isn't possible for all
nodes. Remove the only use of Key in LightTrie by always retrying with
the original key that was looked up.
This commit adds pluggable consensus engines to go-ethereum. In short, it
introduces a generic consensus interface, and refactors the entire codebase to
use this interface.
This commit solves several issues concerning the genesis block:
* Genesis/ChainConfig loading was handled by cmd/geth code. This left
library users in the cold. They could specify a JSON-encoded
string and overwrite the config, but didn't get any of the additional
checks performed by geth.
* Decoding and writing of genesis JSON was conflated in
WriteGenesisBlock. This made it a lot harder to embed the genesis
block into the forthcoming config file loader. This commit changes
things so there is a single Genesis type that represents genesis
blocks. All uses of Write*Genesis* are changed to use the new type
instead.
* If the chain config supplied by the user was incompatible with the
current chain (i.e. the chain had already advanced beyond a scheduled
fork), it got overwritten. This is not an issue in practice because
previous forks have always had the highest total difficulty. It might
matter in the future though. The new code reverts the local chain to
the point of the fork when upgrading configuration.
The change to genesis block data removes compression library
dependencies from package core.
There is no need to depend on the old context package now that the
minimum Go version is 1.7. The move to "context" eliminates our weird
vendoring setup. Some vendored code still uses golang.org/x/net/context
and it is now vendored in the normal way.
This change triggered new vet checks around context.WithTimeout which
didn't fire with golang.org/x/net/context.
* common/math: optimize PaddedBigBytes, use it more
name old time/op new time/op delta
PaddedBigBytes-8 71.1ns ± 5% 46.1ns ± 1% -35.15% (p=0.000 n=20+19)
name old alloc/op new alloc/op delta
PaddedBigBytes-8 48.0B ± 0% 32.0B ± 0% -33.33% (p=0.000 n=20+20)
* all: unify big.Int zero checks
Various checks were in use. This commit replaces them all with Int.Sign,
which is cheaper and less code.
eg templates:
func before(x *big.Int) bool { return x.BitLen() == 0 }
func after(x *big.Int) bool { return x.Sign() == 0 }
func before(x *big.Int) bool { return x.BitLen() > 0 }
func after(x *big.Int) bool { return x.Sign() != 0 }
func before(x *big.Int) int { return x.Cmp(common.Big0) }
func after(x *big.Int) int { return x.Sign() }
* common/math, crypto/secp256k1: make ReadBits public in package math
* common: remove CurrencyToString
Move denomination values to params instead.
* common: delete dead code
* common: move big integer operations to common/math
This commit consolidates all big integer operations into common/math and
adds tests and documentation.
There should be no change in semantics for BigPow, BigMin, BigMax, S256,
U256, Exp and their behaviour is now locked in by tests.
The BigD, BytesToBig and Bytes2Big functions don't provide additional
value, all uses are replaced by new(big.Int).SetBytes().
BigToBytes is now called PaddedBigBytes, its minimum output size
parameter is now specified as the number of bytes instead of bits. The
single use of this function is in the EVM's MSTORE instruction.
Big and String2Big are replaced by ParseBig, which is slightly stricter.
It previously accepted leading zeros for hexadecimal inputs but treated
decimal inputs as octal if a leading zero digit was present.
ParseUint64 is used in places where String2Big was used to decode a
uint64.
The new functions MustParseBig and MustParseUint64 are now used in many
places where parsing errors were previously ignored.
* common: delete unused big integer variables
* accounts/abi: replace uses of BytesToBig with use of encoding/binary
* common: remove BytesToBig
* common: remove Bytes2Big
* common: remove BigTrue
* cmd/utils: add BigFlag and use it for error-checked integer flags
While here, remove environment variable processing for DirectoryFlag
because we don't use it.
* core: add missing error checks in genesis block parser
* common: remove String2Big
* cmd/evm: use utils.BigFlag
* common/math: check for 256 bit overflow in ParseBig
This is supposed to prevent silent overflow/truncation of values in the
genesis block JSON. Without this check, a genesis block that set a
balance larger than 256 bits would lead to weird behaviour in the VM.
* cmd/utils: fixup import
Reworked the EVM gas instructions to use 64bit integers rather than
arbitrary size big ints. All gas operations, be it additions,
multiplications or divisions, are checked and guarded against 64 bit
integer overflows.
In additon, most of the protocol paramaters in the params package have
been converted to uint64 and are now constants rather than variables.
* common/math: added overflow check ops
* core: vmenv, env renamed to evm
* eth, internal/ethapi, les: unmetered eth_call and cancel methods
* core/vm: implemented big.Int pool for evm instructions
* core/vm: unexported intPool methods & verification methods
* core/vm: added memoryGasCost overflow check and test
Reworked the EVM gas instructions to use 64bit integers rather than
arbitrary size big ints. All gas operations, be it additions,
multiplications or divisions, are checked and guarded against 64 bit
integer overflows.
In additon, most of the protocol paramaters in the params package have
been converted to uint64 and are now constants rather than variables.
* common/math: added overflow check ops
* core: vmenv, env renamed to evm
* eth, internal/ethapi, les: unmetered eth_call and cancel methods
* core/vm: implemented big.Int pool for evm instructions
* core/vm: unexported intPool methods & verification methods
* core/vm: added memoryGasCost overflow check and test
The Subscription type is gone, all uses are replaced by
*TypeMuxSubscription. This change is prep-work for the
introduction of the new Subscription type in a later commit.
gorename -from '"github.com/ethereum/go-ethereum/event"::Event' -to TypeMuxEvent
gorename -from '"github.com/ethereum/go-ethereum/event"::muxsub' -to TypeMuxSubscription
gofmt -w -r 'Subscription -> *TypeMuxSubscription' ./event/*.go
find . -name '*.go' -and -not -regex '\./vendor/.*' \| xargs gofmt -w -r 'event.Subscription -> *event.TypeMuxSubscription'
This significantly reduces the dependency closure of ethclient, which no
longer depends on core/vm as of this change.
All uses of vm.Logs are replaced by []*types.Log. NewLog is gone too,
the constructor simply returned a literal.
The run loop, which previously contained custom opcode executes have been
removed and has been simplified to a few checks.
Each operation consists of 4 elements: execution function, gas cost function,
stack validation function and memory size function. The execution function
implements the operation's runtime behaviour, the gas cost function implements
the operation gas costs function and greatly depends on the memory and stack,
the stack validation function validates the stack and makes sure that enough
items can be popped off and pushed on and the memory size function calculates
the memory required for the operation and returns it.
This commit also allows the EVM to go unmetered. This is helpful for offline
operations such as contract calls.
The transaction pool keeps track of the current nonce in its local pendingState. When a
new block comes in the pendingState is reset. During the reset it fetches multiple times
the current state through the use of the currentState callback. When a second block comes
in during the reset its possible that the state changes during the reset. If that block
holds transactions that are currently in the pool the local pendingState that is used to
determine nonces can get out of sync.
Environment is now a struct (not an interface). This
reduces a lot of tech-debt throughout the codebase where a virtual
machine environment had to be implemented in order to test or run it.
The new environment is suitable to be used en the json tests, core
consensus and light client.
This commit implements EIP158 part 1, 2, 3 & 4
1. If an account is empty it's no longer written to the trie. An empty
account is defined as (balance=0, nonce=0, storage=0, code=0).
2. Delete an empty account if it's touched
3. An empty account is redefined as either non-existent or empty.
4. Zero value calls and zero value suicides no longer consume the 25k
reation costs.
params: moved core/config to params
Signed-off-by: Jeffrey Wilcke <jeffrey@ethereum.org>
These accessors were introduced by light client changes, but
the only method that is actually used is GetNumberU64. This
commit replaces all uses of .GetNumberU64 with .Number.Uint64.
* trie: store nodes as pointers
This avoids memory copies when unwrapping node interface values.
name old time/op new time/op delta
Get 388ns ± 8% 215ns ± 2% -44.56% (p=0.000 n=15+15)
GetDB 363ns ± 3% 202ns ± 2% -44.21% (p=0.000 n=15+15)
UpdateBE 1.57µs ± 2% 1.29µs ± 3% -17.80% (p=0.000 n=13+15)
UpdateLE 1.92µs ± 2% 1.61µs ± 2% -16.25% (p=0.000 n=14+14)
HashBE 2.16µs ± 6% 2.18µs ± 6% ~ (p=0.436 n=15+15)
HashLE 7.43µs ± 3% 7.21µs ± 3% -2.96% (p=0.000 n=15+13)
* trie: close temporary databases in GetDB benchmark
* trie: don't keep []byte from DB load around
Nodes decoded from a DB load kept hashes and values as sub-slices of
the DB value. This can be a problem because loading from leveldb often
returns []byte with a cap that's larger than necessary, increasing
memory usage.
* trie: unload old cached nodes
* trie, core/state: use cache unloading for account trie
* trie: use explicit private flags (fixes Go 1.5 reflection issue).
* trie: fixup cachegen overflow at request of nick
* core/state: rename journal size constant
This commit replaces the deep-copy based state revert mechanism with a
linear complexity journal. This commit also hides several internal
StateDB methods to limit the number of ways in which calling code can
use the journal incorrectly.
As usual consultation and bug fixes to the initial implementation were
provided by @karalabe, @obscuren and @Arachnid. Thank you!
Péter Szilágyi
Martin Holst Swende
Felföldi Zsolt
Matthew Halpern
Felix Lange
b00ris
Iskander (Alex) Sharipov
Sheldon
gary rong
Sheldon
kiel barry
Kyuntae Ethan Kim
Miya Chen
Egon Elbre
bas-vk
Felix Lange
Jeffrey Wilcke
Nick Johnson
Gustav Simonsson
Ricardo Catalinas Jiménez
Bas van Kervel