The stack trace did not make it evidant which request had caused this
issue, so add extra logging for the request that caused it. Sanitize
this request if it may contain any secrets.
Additionally, in this situation, begin shutting down the wallet rather
than simply dropping the response. This will help to catch the issue
since it's easier to notice all requests failing, rather than just one
that was dropped. If shutdown takes an unreasonably long time, panic.
Instead of using the zero value, explicitly set the last chain index
to -1, which represents the root key. If no additional keys are read
from the io.Reader, this could result in panics when looking up keys
at index 0, when no additional keys have been created yet.
Fixes#119.
Previously, registerations for wallet notifications (new txs, changed
account balances) were only passed up to websocket clients if the
wallet was loaded off disk (SetWallet was called with a non-nil
wallet), and not for the case when the RPC server would create the
wallet (if it wasn't created yet, and the user manually created it
with createencryptedwallet). This change fixes that by registering
for these notifications when this code path is taken.
The notified chain server connection state was being passed through
the wallet and then notified to the RPC server, which prevented this
notification from ever firing if a wallet didn't exist yet. Instead,
make the RPC server register for these notifications directly from the
chain server RPC client.
I'm not happy with this notification and how it's handled in the code,
but to not break existing clients this change is being made. Fixing
the notifiation mess and modifying existing clients to use a new
notification API will need to be done sometime later.
This prevents a hang when attempting to set the wallet (and register
for wallet notifications) when the process is interrupted and the rpc
server begins shutting down.
If a websocket client was already connected and the wallet and/or
chain server is loaded into the rpc server (enabling the handlers
specific to those components), the btcdconnected notifications were
not being sent, and this could break clients that expected the
notification. I'm not happy with this change, but since this is how
notifications are currently done (unsolicited), and to not break
compatibility yet, I'm adding these back in for now.
Eventually, this notification will require explicit registration
before it is received by a client. See issue #84.
Closes#115.
Addresses do no have balances. In situations where a payment is
required and just a single address was provided, it is better to track
the unspent outputs themselves, rather than watching some artificial
measure of payment.
This commit is the result of several big changes being made to the
wallet. In particular, the "handshake" (initial sync to the chain
server) was quite racy and required proper synchronization. To make
fixing this race easier, several other changes were made to the
internal wallet data structures and much of the RPC server ended up
being rewritten.
First, all account support has been removed. The previous Account
struct has been replaced with a Wallet structure, which includes a
keystore for saving keys, and a txstore for storing relevant
transactions. This decision has been made since it is the opinion of
myself and other developers that bitcoind accounts are fundamentally
broken (as accounts implemented by bitcoind support both arbitrary
address groupings as well as moving balances between accounts -- these
are fundamentally incompatible features), and since a BIP0032 keystore
is soon planned to be implemented (at which point, "accounts" can
return as HD extended keys). With the keystore handling the grouping
of related keys, there is no reason have many different Account
structs, and the AccountManager has been removed as well. All RPC
handlers that take an account option will only work with "" (the
default account) or "*" if the RPC allows specifying all accounts.
Second, much of the RPC server has been cleaned up. The global
variables for the RPC server and chain server client have been moved
to part of the rpcServer struct, and the handlers for each RPC method
that are looked up change depending on which components have been set.
Passthrough requests are also no longer handled specially, but when
the chain server is set, a handler to perform the passthrough will be
returned if the method is not otherwise a wallet RPC. The
notification system for websocket clients has also been rewritten so
wallet components can send notifications through channels, rather than
requiring direct access to the RPC server itself, or worse still,
sending directly to a websocket client's send channel. In the future,
this will enable proper registration of notifications, rather than
unsolicited broadcasts to every connected websocket client (see
issue #84).
Finally, and the main reason why much of this cleanup was necessary,
the races during intial sync with the chain server have been fixed.
Previously, when the 'Handshake' was run, a rescan would occur which
would perform modifications to Account data structures as
notifications were received. Synchronization was provided with a
single binary semaphore which serialized all access to wallet and
account data. However, the Handshake itself was not able to run with
this lock (or else notifications would block), and many data races
would occur as both notifications were being handled. If GOMAXPROCS
was ever increased beyond 1, btcwallet would always immediately crash
due to invalid addresses caused by the data races on startup. To fix
this, the single lock for all wallet access has been replaced with
mutexes for both the keystore and txstore. Handling of btcd
notifications and client requests may now occur simultaneously.
GOMAXPROCS has also been set to the number of logical CPUs at the
beginning of main, since with the data races fixed, there's no reason
to prevent the extra parallelism gained by increasing it.
Closes#78.
Closes#101.
Closes#110.
The responses chan for a websocket client was being closed by one of
the websocket goroutines, but it was not the only sender to this
channel. There was also the notification handler, run by the server
to handle notifications to all websocket clients. It was possible to
hit cases where sends to this channel would still occur (the select
statement doesn't guarantee that the picked channel operation won't
panic, even if there's another that won't). To fix this, wait on the
client being removed from the notification group, or if the server is
already shutting down, wait on the notification handler completely
closing, to ensure that no more sends to the channel will occur,
before closing the channel.
Fixes#110.
This package is used solely for the storage of private and public
keys, and the addresses they represent. Since "wallet" is an
overloaded term and a working wallet requires transaction history as
well, rename this package and its data structures to more clearly
reflect what it is for.
When a BIP0032 wallet is implemented and multiple address chains can
be supported by a single keystore, the Account structure will
represent a single wallet (and be renamed to reflect that change),
rather than keeping the collection of Account structs as currently
managed by the AccountManager. In preperation for this, and to remove
a global variable, move the fee increment for created transactions to
this structure. When setting the fee, look it up from the default
account.
Pass the RPC client to the notification handlers. Update the last
seen block for blockconnected notifications in the client structure
directly, protecting access with a mutex.
When the addmultisigaddress RPC was called, the wallet with the
imported address was not being written to disk, and if no more writes
were scheduled, the address could be lost. This change immediately
writes the updated keystore to disk before the RPC returns.
Closes#98.
This change fixes the asynchronous deferred locking that used to be
performed after some timeout after a call to walletpassphrase by
managing the locked state of each account in a new account manager
goroutine. The timeouts for new unlock requests replace any running
timeouts for older requests, rather than allowing previous timeouts to
expire before the most recent one.
Fixes#105.
Instead of checking whether the chain server client is currently
connected (as opposed to in a reconnect loop), always shutdown the
client connection so other rpc calls begin erroring.
