Olaoluwa Osuntokun
d5cd95b9a9
wire: optimize parsing for CFCheckpkt message, reduce allocs by 96%
...
In this commit, we optimize the decoding for the CFCheckpkt message. The
old decode routine would do a fresh alloc for each hash to be read out.
Instead, we'll now allocate enough memory for the entire set of headers
to be decoded, then read them into that contiguous slice, and point to
members of this slice in the wire message itself.
We've also added benchmarks to show the improvement:
```
⛰ cat bench-cmp.txt
goos: darwin
goarch: arm64
pkg: github.com/btcsuite/btcd/wire
cpu: Apple M4 Max
│ bench-old.txt │ bench-new.txt │
│ sec/op │ sec/op vs base │
MsgCFCheckptDecode/headers_1000-16 14.354µ ± ∞ ¹ 6.919µ ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptDecode/headers_10000-16 146.77µ ± ∞ ¹ 70.23µ ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptDecode/headers_100000-16 1473.6µ ± ∞ ¹ 564.2µ ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptEncode/headers_1000-16 7.021µ ± ∞ ¹ 7.196µ ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptEncode/headers_10000-16 88.97µ ± ∞ ¹ 90.40µ ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptEncode/headers_100000-16 861.0µ ± ∞ ¹ 875.0µ ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptDecodeEmpty-16 68.28n ± ∞ ¹ 66.61n ± ∞ ¹ ~ (p=1.000 n=1) ²
geomean 37.98µ 26.98µ -28.98%
¹ need >= 6 samples for confidence interval at level 0.95
² need >= 4 samples to detect a difference at alpha level 0.05
│ bench-old.txt │ bench-new.txt │
│ B/op │ B/op vs base │
MsgCFCheckptDecode/headers_1000-16 39.36Ki ± ∞ ¹ 40.11Ki ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptDecode/headers_10000-16 392.6Ki ± ∞ ¹ 400.1Ki ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptDecode/headers_100000-16 3.817Mi ± ∞ ¹ 3.820Mi ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptEncode/headers_1000-16 63.98Ki ± ∞ ¹ 63.98Ki ± ∞ ¹ ~ (p=1.000 n=1) ³
MsgCFCheckptEncode/headers_10000-16 1.000Mi ± ∞ ¹ 1.000Mi ± ∞ ¹ ~ (p=1.000 n=1) ³
MsgCFCheckptEncode/headers_100000-16 8.000Mi ± ∞ ¹ 8.000Mi ± ∞ ¹ ~ (p=1.000 n=1) ³
MsgCFCheckptDecodeEmpty-16 112.0 ± ∞ ¹ 112.0 ± ∞ ¹ ~ (p=1.000 n=1) ³
geomean 166.5Ki 167.4Ki +0.55%
¹ need >= 6 samples for confidence interval at level 0.95
² need >= 4 samples to detect a difference at alpha level 0.05
³ all samples are equal
│ bench-old.txt │ bench-new.txt │
│ allocs/op │ allocs/op vs base │
MsgCFCheckptDecode/headers_1000-16 1003.000 ± ∞ ¹ 4.000 ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptDecode/headers_10000-16 10003.000 ± ∞ ¹ 4.000 ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptDecode/headers_100000-16 100003.000 ± ∞ ¹ 4.000 ± ∞ ¹ ~ (p=1.000 n=1) ²
MsgCFCheckptEncode/headers_1000-16 11.00 ± ∞ ¹ 11.00 ± ∞ ¹ ~ (p=1.000 n=1) ³
MsgCFCheckptEncode/headers_10000-16 15.00 ± ∞ ¹ 15.00 ± ∞ ¹ ~ (p=1.000 n=1) ³
MsgCFCheckptEncode/headers_100000-16 18.00 ± ∞ ¹ 18.00 ± ∞ ¹ ~ (p=1.000 n=1) ³
MsgCFCheckptDecodeEmpty-16 2.000 ± ∞ ¹ 2.000 ± ∞ ¹ ~ (p=1.000 n=1) ³
geomean 179.3 6.268 -96.50%
¹ need >= 6 samples for confidence interval at level 0.95
² need >= 4 samples to detect a difference at alpha level 0.05
³ all samples are equal
```
Old bench:
```
goos: darwin
goarch: arm64
pkg: github.com/btcsuite/btcd/wire
cpu: Apple M4 Max
BenchmarkMsgCFCheckptDecode/headers_1000-16 74678 14354 ns/op 40304 B/op 1003 allocs/op
BenchmarkMsgCFCheckptDecode/headers_10000-16 8234 146770 ns/op 402033 B/op 10003 allocs/op
BenchmarkMsgCFCheckptDecode/headers_100000-16 822 1473622 ns/op 4002931 B/op 100003 allocs/op
BenchmarkMsgCFCheckptEncode/headers_1000-16 173762 7021 ns/op 65520 B/op 11 allocs/op
BenchmarkMsgCFCheckptEncode/headers_10000-16 13459 88968 ns/op 1048564 B/op 15 allocs/op
BenchmarkMsgCFCheckptEncode/headers_100000-16 1399 860985 ns/op 8388592 B/op 18 allocs/op
BenchmarkMsgCFCheckptDecodeEmpty-16 17459148 68.28 ns/op 112 B/op 2 allocs/op
PASS
ok github.com/btcsuite/btcd/wire 10.135s
```
New bench:
```
goos: darwin
goarch: arm64
pkg: github.com/btcsuite/btcd/wire
cpu: Apple M4 Max
BenchmarkMsgCFCheckptDecode/headers_1000-16 166368 6919 ns/op 41072 B/op 4 allocs/op
BenchmarkMsgCFCheckptDecode/headers_10000-16 17079 70227 ns/op 409712 B/op 4 allocs/op
BenchmarkMsgCFCheckptDecode/headers_100000-16 2062 564175 ns/op 4006003 B/op 4 allocs/op
BenchmarkMsgCFCheckptEncode/headers_1000-16 173940 7196 ns/op 65520 B/op 11 allocs/op
BenchmarkMsgCFCheckptEncode/headers_10000-16 13054 90401 ns/op 1048564 B/op 15 allocs/op
BenchmarkMsgCFCheckptEncode/headers_100000-16 1408 875012 ns/op 8388592 B/op 18 allocs/op
BenchmarkMsgCFCheckptDecodeEmpty-16 17256627 66.61 ns/op 112 B/op 2 allocs/op
PASS
ok github.com/btcsuite/btcd/wire 10.522s
```
As seen from the benchmarks, allocs have decreased by 96%, and the decode
speed by nearly 30%.
2025-07-14 18:36:27 -07:00
Bruno Garcia
bc52e090ce
p2p: remove alert msg
2025-07-10 13:07:50 -03:00
yyforyongyu
c96d061314
wire: refactor BtcDecode and patch unit test
2025-06-23 21:31:26 +08:00
yyforyongyu
3913a72e2e
main+wire: fix line too long
...
This commit provides a minor fix on the format.
2025-06-23 20:03:33 +08:00
Eugene Siegel
f9c0058e46
wire: introduce encoding/decoding funcs for v2 messages
2025-05-26 13:46:25 -07:00
Oleg Bondar
2aa26d3a0a
wire: add TestNet4
2025-03-18 11:18:37 -05:00
argentpapa
7af88526af
multi: remove deprecated io/ioutil
2025-03-09 15:47:34 +08:00
Olaoluwa Osuntokun
f3bd1f58e7
Merge pull request #2276 from starius/wire-signet
...
wire: add bitcoin network magic for default SigNet
2025-01-13 18:03:32 -08:00
Calvin Kim
48c02954d2
wire, peer: fix broken ibd
...
IBD for new nodes were broken due to the version handshake failing
between nodes that recognized wtxid based relays. Reverting the changes
that were made so that the node is able to connect to those nodes.
2024-11-29 15:07:01 +09:00
Boris Nagaev
38af045ab1
wire: add bitcoin network magic for default SigNet
...
Added constant wire.SigNet, added it to stringer for BitcoinNet type.
Added a test in chaincfg to compare calculated magic value with the constant.
2024-11-25 10:29:15 -03:00
Marco Peereboom
684d64ad74
Add support for wtxidrelay message. ( #2272 )
...
* Add support for wtxidrelay message.
This adds support for the wtxidrelay (BIP339) message in wire.
While here, add tests for sendaddrv2 and rename AddrV2Version to
SendAddrV2Version in order to make the code consistent with all other
messages. This diff does keep the old AddrV2Version constant for
backwards compatibility.
* Bump copyrights
2024-11-06 13:03:27 +00:00
Jonathan Harvey-Buschel
ee7cf5e564
wire: check TXID length before creating outpoint
2024-08-14 16:29:08 -04:00
Olaoluwa Osuntokun
c4ed92fb52
Merge pull request #2178 from ProofOfKeags/standardness-cleanup
...
Update standardness rules congruent to Bitcoin Core
2024-05-21 15:44:24 -07:00
Keagan McClelland
da2f3b1e5f
mempool: make txn's below 65 non-witness bytes non-standard
...
This is to mitigate CVE-2017-12842. Along the way, also error when
deserializing transactions that have the witness marker flag set
but have no witnesses. This matches Bitcoin Core's behaviour initially
introduced here https://github.com/bitcoin/bitcoin/pull/14039 . Allowing
such transactions is benign, but this makes sure that our parsing code
matches Core's exactly.
2024-05-21 12:30:17 -07:00
Oliver Gugger
a03259b26b
Merge pull request #1991 from ffranr/add-json-support
...
Convert `witnessToHex` into a method `ToHexStrings` on `TxWitness`
2024-03-26 13:03:46 -06:00
mattn
3cb9f602e8
fix typos ( #2100 )
2024-03-25 09:44:25 -04:00
ffranr
75fe7e4790
wire: make witnessToHex a TxWitness method called ToHexStrings
2024-03-25 12:56:12 +00:00
ffranr
2a225e9d36
wire: add method TxID to MsgTx
...
This commit adds a method to MsgTx called TxID. This method returns the
transaction ID (txid) of the subject transaction.
2024-03-25 12:21:41 +00:00
yyforyongyu
9cda0f7e95
wire: add Copy method to MsgBlock
2024-01-21 18:43:15 +08:00
Olaoluwa Osuntokun
16684f6cbc
Merge pull request #2073 from Roasbeef/wire-opts
...
wire: only borrow/return binaryFreeList buffers at the message level
2023-12-28 18:52:53 -08:00
Olaoluwa Osuntokun
b0e9636689
wire: consistently use defer for returning scratch buffers
2023-12-28 18:37:43 -08:00
Olaoluwa Osuntokun
c3c3545f9b
multi: update main package to chainhash/v1.1.0, use optimized dsha256
...
In this commit, we update the top-level btcd package to use the latest
version of btcutil and also the chainhash package. With this version
bump, we can now use the new optimized dsha256 routine where applicable.
With this commit, I've covered most of the areas we'll hash an entire
transaction/block/header, but we may want to optimize some other areas
further, in particular, the witness sighash calc.
2023-12-19 15:01:55 -08:00
Conner Fromknecht
8c4da83890
wire/msgblock+msgtx: user block-level script slab
2023-12-15 16:38:02 -08:00
Conner Fromknecht
d7396dc13d
wire/msgtx: use tx-level script slab
2023-12-15 16:37:57 -08:00
Conner Fromknecht
2e6eefcab1
wire/msgtx: remove unused read/writeOutPoint
2023-12-15 16:37:54 -08:00
Conner Fromknecht
4cc4f76c55
wire/msgtx: remove unused readScript
2023-12-15 16:37:51 -08:00
Conner Fromknecht
e0fa866890
wire/msgtx: remove unused readTxIn
2023-12-15 16:37:49 -08:00
Conner Fromknecht
f0184e58ba
wire/msgtx: remove unused writeTxIn
2023-12-15 16:37:46 -08:00
Conner Fromknecht
da89ed68ff
wire/msgtx: remove unused writeTxWitness
2023-12-15 16:37:44 -08:00
Conner Fromknecht
80ae5d342c
wire/msgpong: remove usage of read/writeElement
2023-12-15 16:37:41 -08:00
Conner Fromknecht
3698f2deab
wire/msgping: remove usage for read/writeElement
2023-12-15 16:37:39 -08:00
Conner Fromknecht
7207967135
wire/msgmerkleblock: optimize serialization by reusing small buffers
2023-12-15 16:37:36 -08:00
Conner Fromknecht
8bf07cc0bf
wire/netaddress: add optimiezed read/writeNetAddressBuf
2023-12-15 16:37:34 -08:00
Conner Fromknecht
dc4fbb04b3
wire/msgreject: optimize serialization by reusing small buffers
2023-12-15 16:37:31 -08:00
Conner Fromknecht
57daac345e
wire/common: add optimized writeVarStrBuf an readVarStrBuf
2023-12-15 16:37:29 -08:00
Conner Fromknecht
1cd5e02838
wire/invvect: remove unused readInvVect and writeInvVect
2023-12-15 16:37:26 -08:00
Conner Fromknecht
efcf964be2
wire/msgnotfound: optimize serialization by reusing small buffers
2023-12-15 16:37:24 -08:00
Conner Fromknecht
834febbb86
wire/msgcfilter: optimize serialization by reusing small buffers
2023-12-15 16:37:21 -08:00
Conner Fromknecht
ddeba60026
wiree/msggetcfilters: optimize serialization by reusing small buffers
2023-12-15 16:37:19 -08:00
Conner Fromknecht
d6594daa49
wire/msggetdata: optimize serialization by reusing small buffers
2023-12-15 16:37:16 -08:00
Conner Fromknecht
2383a04977
wire/msginv: optimize by reusing small buffers
2023-12-15 16:37:13 -08:00
Conner Fromknecht
f37f4750dc
wire/msgcfcheckpt: optimize serialization by reusing small buffers
2023-12-15 16:37:09 -08:00
Conner Fromknecht
1990555eee
wire/msggetcfcheckpt: optimize by removing read/writeElement
2023-12-15 16:37:02 -08:00
Conner Fromknecht
1c525dbd0e
wire/msgcfheaders: optimize encode/decode by using one small buffer
2023-12-15 16:36:56 -08:00
Conner Fromknecht
d042fe0586
wire/msggetcfheaders: use single small buffer for encode/decode
2023-12-15 16:36:54 -08:00
Conner Fromknecht
83675cb393
wire/msgheaders: optimize serialization by reusing small buffers
2023-12-15 16:36:51 -08:00
Conner Fromknecht
c0d35e6d92
wire/msggetheaders: optimize serialization by reusing small buffers
2023-12-15 16:36:49 -08:00
Conner Fromknecht
d8e08170bd
wire/msgblock: optimize DeserializeTxLoc by reusing small buffers
2023-12-15 16:36:46 -08:00
Conner Fromknecht
ee1f8077d8
wire/msgblock: use only one small buffer per block encode/decode
2023-12-15 16:36:44 -08:00
Conner Fromknecht
4ebc651689
wire/msggetblocks: optimize by reusing small buffer
2023-12-15 16:36:41 -08:00
