- Check sha256 digests instead of full blobs (much less memory used)
- Replace `find->insert` sequence with a single `insert` - 2x fewer hashset accesses
- Preallocate the required size for the hashset (no full-table rehashes)
1. Doesn't allocate a `boost::shared_ptr`
2. Doesn't use a vtable
3. Doesn't use `std::function` wrapper
4. Doesn't necessarily copy functional object
5. Doesn't need a factory function for simple declaration (yay CTAD)
6. Isolated to its own header
7. Name doesn't suck as much
Co-authored-by: nahuhh
- jberman review
- cryptonote_protocol: don't arbitrarily download 1000 blocks ahead
- further restrict `proceed` to require `queue_proceed` in all cases.
ensure queue_proceed is true if we need the next block, even if we
already exceed the span and size limits
cryptonote_protocol: improved logging + const usage in span downloader
Without this commit:
1) read height from DB
2) add block to chain in separate thread
3) read chain for block id's and request them from peer
4) ERR in handle_response_chain_entry, peer's first block is the
one that was added to the chain, which has block idx=height from
step 1.
This commit reads the chain for height and highest block id's
in one go while holding the m_blockchain_lock to avoid the race.
Changes complexity from M*N to (2*N+M)*log2(M). The FCMP++ stressnet recently hit mempool sizes of ~55k txs.
If the requesting node's mempool is populated, this results in an average of (55000*55000)/2
(about 1.5 billion) comparisons for the responding node. Under this commit, this would be reduced to
(55000+55000)*log2(55000) comparisons (about 2.6 million), a 99.83% reduction.
update_checkpoints() makes a few DNS requests and can take up to 20-30 seconds to complete (3-6 seconds on average). It is currently called from core::handle_incoming_block() which holds m_incoming_tx_lock, so it blocks all incoming transactions and blocks processing while update_checkpoints() is running. This PR moves it to until after a new block has been processed and relayed, to avoid full monerod locking.
In this repo, `boost::interprocess` was being used soley to make `uint32_t` operations atomic. So I replaced each instance of
`boost::interprocess::ipcdetail::atomic(...)32` with `std::atomic` methods. I replaced member declarations as applicable. For example,
when I needed to change a `volatile uint32_t` into a `std::atomic<uint32_t>`. Sometimes, a member was being used a boolean flag, so
I replaced it with `std::atomic<bool>`.
You may notice that I didn't touch `levin_client_async.h`. That is because this file is entirely unused and will be deleted in PR monero-project#8211.
Additional changes from review:
* Make some local variables const
* Change postfix operators to prefix operators where value was not need
when kicking a peer for inactivity, clear the set of requested blocks,
or next time we requests blocks from it, we'll probably reject the
incoming blocks due to missing the previous requested blocks
There are quite a few variables in the code that are no longer
(or perhaps never were) in use. These were discovered by enabling
compiler warnings for unused variables and cleaning them up.
In most cases where the unused variables were the result
of a function call the call was left but the variable
assignment removed, unless it was obvious that it was
a simple getter with no side effects.
SChernykh