Some developers clearly don't get this and have been posting
"improvements" that create clear vulnerabilities. It should
have been better explained in the code, since the design
is somewhat subtle and getting it right is important.
This patch changes the implementation from one that stores 16 2-bit integers
in one uint32_t's, to one that stores the first bit of 64 2-bit integers in
one uint64_t and the second bit in another. This allows for 450x faster
refreshing and 2.2x faster average speed.
Change the few occurrences of the deprecated `auto_ptr` to c++11 `unique_ptr`.
Silences the deprecation warnings.
Also add a missing `std::` for consistency.
Bitwise logic combined with `<` with undefined signedness will
potentially results in undefined behavior. Fix this by defining the type
as a c++11 typed enum.
Fixes#6017.
DumpBanList currently does this:
- with lock: take a copy of the banmap
- perform I/O (write out the banmap)
- with lock: mark the banmap non-dirty
If a new ban is added during the I/O operation, it may never be persisted to
disk.
Reorder operations so that the data to be persisted cannot be older than the
time at which the banmap was marked non-dirty.
I made a silly mistake in a database wrapper where keys
were sorted by char instead of uint8_t. As x86 char is signed
the sorting for the block index database was messed up, resulting
in a segfault due to missing records.
Add a test to catch:
- Wrong sorting
- Seeking errors
- Iteration result not complete
ActivateBestChain uses chainActive after releasing the lock; reorder operations
to move all access to synchronized object into existing LOCK(cs_main) block.
A small GUI annoyance for me has always been that it's impossible to
have multiple transaction detail windows open, for example to compare
transactions.
This patch makes the window non-modal so that it is possible to open
transaction details at will.
Pass parent wrapper directly instead of obfuscation key. This
makes it possible for other databases which re-use this code
to use other properties from the database.
Add a namespace dbwrapper_private for private functions to be used
only in dbwrapper.h/cpp and dbwrapper_tests.
Using throw() specifications in function signatures is not only
not required in C++, it is considered deprecated for
[various reasons](https://stackoverflow.com/questions/1055387/throw-keyword-in-functions-signature).
It is not implemented by any of the common C++ compilers. The usage is
also inconsistent with the rest of the source code.
Without this patch:
- When I compile the GUI from the bitcoin directory itself, it works as
expected.
- When I build the GUI in an out-of-tree build, I cannot get it to
select tabs. When I click, say the "Receive" tab nothing happens,
the button selects but it doesn't switch the page. The rest - even
the debug window - seems to work.
See full discussion here:
https://github.com/bitcoin/bitcoin/pull/7911#issuecomment-212413442
This turned out to be caused by a mismatch in the arguments to moc,
preventing it from finding `bitcoin-config.h`. Fix this by passing
`$(DEFAULT_INCLUDES)` to it, which gets set to the appropriate
path by autoconf itself.
SetString seems to be passing the length of the wrong variable to
memory_cleanse, resulting in the last byte of the temporary buffer not being
securely erased.
This will avoid sending more pointless INVs around updates, and
prevents using filter updates to timetag transactions.
Also adds locking for fRelayTxes.
By eliminating queued entries from the mempool response and responding only at
trickle time, this makes the mempool no longer leak transaction arrival order
information (as the mempool itself is also sorted)-- at least no more than
relay itself leaks it.
Previously we would assert that if every block in vBlockHashesToAnnounce is in
chainActive, then the blocks to be announced must connect. However, there are
edge cases where this assumption could be violated (eg using invalidateblock /
reconsiderblock), so just check for this case and revert to inv-announcement
instead.
Rather than allowing CNetAddr/CService/CSubNet to launch DNS queries, require
that addresses are already resolved.
This greatly simplifies async resolve logic, and makes it harder to
accidentally leak DNS queries.
Note: Some seeds aren't actually returning an IP for their name entries, so
they're being added to addrman with a source of [::].
This commit shouldn't change that behavior, for better or worse.
Previously Bitcoin would send 1/4 of transactions out to all peers
instantly. This causes high overhead because it makes >80% of
INVs size 1. Doing so harms privacy, because it limits the
amount of source obscurity a transaction can receive.
These randomized broadcasts also disobeyed transaction dependencies
and required use of the orphan pool. Because the orphan pool is
so small this leads to poor propagation for dependent transactions.
When the bypass wasn't in effect, transactions were sent in the
order they were received. This avoided creating orphans but
undermines privacy fairly significantly.
This commit:
Eliminates the bypass. The bypass is replaced by halving the
average delay for outbound peers.
Sorts candidate transactions for INV by their topological
depth then by their feerate (then hash); removing the
information leakage and providing priority service to
higher fee transactions.
Limits the amount of transactions sent in a single INV to
7tx/sec (and twice that for outbound); this limits the
harm of low fee transaction floods, gives faster relay
service to higher fee transactions. The 7 sounds lower
than it really is because received advertisements need
not be sent, and because the aggregate rate is multipled
by the number of peers.
leveldb's buildsystem causes us a few problems:
- breaks out-of-tree builds
- forces flags used for some tools
- limits cross builds
Rather than continuing to add wrappers around it, simply integrate it into our
build.
Gregory Maxwell