The precompile at 0x09 wraps the BLAKE2b F compression function:
https://tools.ietf.org/html/rfc7693#section-3.2

The precompile requires 6 inputs tightly encoded, taking exactly 213
bytes, as explained below.

- `rounds` - the number of rounds - 32-bit unsigned big-endian word
- `h` - the state vector - 8 unsigned 64-bit little-endian words
- `m` - the message block vector - 16 unsigned 64-bit little-endian words
- `t_0, t_1` - offset counters - 2 unsigned 64-bit little-endian words
- `f` - the final block indicator flag - 8-bit word

[4 bytes for rounds][64 bytes for h][128 bytes for m][8 bytes for t_0]
[8 bytes for t_1][1 byte for f]

The boolean `f` parameter is considered as `true` if set to `1`.
The boolean `f` parameter is considered as `false` if set to `0`.
All other values yield an invalid encoding of `f` error.

The precompile should compute the F function as specified in the RFC
(https://tools.ietf.org/html/rfc7693#section-3.2) and return the updated
state vector `h` with unchanged encoding (little-endian).

See EIP-152 for details.

* Removed comment section referring to Cloudflare's bn curve parameters

* Added comment to clarify the nature of the parameters

* Changed value of xi to i+9

* Added checks for nil pointers in Marshal functions

* Set nil pointer to identity in GT before marshaling

Package crypto works with or without cgo, which is great. However, to make it
work without cgo required setting the build tag `nocgo`. It's common to disable
cgo by instead just setting the environment variable `CGO_ENABLED=0`. Setting
this environment variable does _not_ implicitly set the build tag `nocgo`. So
projects that try to build the crypto package with `CGO_ENABLED=0` will fail. I
have done this myself several times. Until today, I had just assumed that this
meant that this package requires cgo.

But a small build tag change will make this case work. Instead of using `nocgo`
and `!nocgo`, we can use `!cgo` and `cgo`, respectively. The `cgo` build tag is
automatically set if cgo is enabled, and unset if it is disabled.

* core/vm/runtim: benchmark create/create2

* core/vm: do less hashing in CREATE2

* core/vm: avoid storing jumpdest analysis for initcode

* core/vm: avoid unneccesary lookups, remove unused fields

* core/vm: go formatting tests

* core/vm: save jumpdest analysis locally

* core/vm: use common.Hash instead of nil, fix review comments

* core/vm: removed type destinations

* core/vm: correct check for empty hash

* eth: more elegant api_tracer

* core/vm: address review concerns

`(void)data;` may cause link error on Windows.

* core, crypto, params: implement CREATE2 evm instrction

* core/vm: add opcode to string mapping

* core: remove past fork checking

* core, crypto: use option2 to generate new address

Our original wrapper code had two parts. One taken from a third
party repository (who took it from upstream Go) licensed under
BSD-3. The second written by Jeff, Felix and Gustav, licensed
under LGPL. This made this package problematic to use from the
outside.

With the agreement of the original copyright holders, this commit
changes the license of the LGPL portions of the code to BSD-3:

---
I agree changing from LGPL to a BSD style license.

Jeff
---
Hey guys,

My preference would be to relicense to GNUBL, but I'm also OK with BSD.

Cheers,
Gustav
---
Felix Lange (fjl):
I would approve anything that makes our licensing less complicated
---

ToECDSAPub was unsafe because it returned a non-nil key with nil X, Y in
case of invalid input. This change replaces ToECDSAPub with
UnmarshalPubkey across the codebase.

The parameter `rand` is unused in `PrivateKey.Decrypt`. Decryption in
the ECIES encryption scheme is deterministic, so randomness isn't
needed.

* crypto/bn256: full switchover to cloudflare's code

* crypto/bn256: only use cloudflare for optimized architectures

* crypto/bn256: upstream fallback for non-optimized code

* .travis, build: drop support for Go 1.8 (need type aliases)

* crypto/bn256/cloudflare: enable curve mul lattice optimization

* core/vm, crypto/bn256: switch over to cloudflare library

* crypto/bn256: unmarshal constraint + start pure go impl

* crypto/bn256: combo cloudflare and google lib

* travis: drop 386 test job

Fixes #15744

* crypto: ensure that VerifySignature rejects malleable signatures

It already rejected them when using libsecp256k1, make sure the nocgo
version does the same thing.

* crypto: simplify check

* crypto: fix build

This adds the inverse to DecompressPubkey and improves a few minor
details in crypto/secp256k1.

With this change,

    key, err := crypto.HexToECDSA("000000...")
    
returns nil key and an error instead of a non-nil key with nil X
and Y inside. Issue found by @guidovranken.

We need those operations for p2p/enr.

Also upgrade github.com/btcsuite/btcd/btcec to the latest version
and improve BenchmarkSha3. The benchmark printed extra output 
that confused tools like benchstat and ignored N.

Generator in the current lib uses -2 as the y point when doing
ScalarBaseMult, this makes it so that points/signatures generated
from libs like py_ecc don't match/validate as pretty much all
other libs (including libsnark) have (1, 2) as the standard
generator.

This does not affect consensus as the generator is never used in
the VM, points are always explicitly defined and there is not
ScalarBaseMult op - it only makes it so that doing "import
github.com/ethereum/go-ethereum/crypto/bn256" doesn't generate
bad points in userland tools.