After 40 minutes, time out a test-before-evict entry and just evict without
testing. Otherwise, if we were unable to test an entry for some reason, we
might break using feelers altogether.
Fixes a bug where feelers could be stuck trying to resolve a collision in the
tried table that is to an address in the same netgroup as an existing outbound peer.
Thanks to Muoi Tran for the original bug report and detailed debug logs to track
this down.
9999879f56 refactor: Use RPCHelpMan::IsValidNumArgs in getrawtransaction (MarcoFalke)
fa9ff8fe21 doc: Remove misleading hint in getrawtransaction (MarcoFalke)
Pull request description:
For 0.18.0
I asked this line to be added in #15159, which was wrong because getmempoolentry does not return the raw transaction hex.
Tree-SHA512: 7ac85500c8192314347b7283cd369196bb959c124863642b6c1ce73d5662b1cbe4f42ded9c374dac6657458ab70b01810caf1235dd1d2b404bf376ebf09efa69
a720a98301 gui: Fix async open wallet call order (João Barbosa)
Pull request description:
Fixes#15455. Must call `OpenWalletActivity::open` asynchronously only after all connections are made to the `OpenWalletActivity` instance, otherwise signals can be missed.
Tree-SHA512: 4e5fdbd09d2ca017ed07a1813c2707c09f96275f1498779804e322e0a4dbd7dcff0c2e9cd6ec18463cd427b88b192a8d02373de9edc2b03ba5e4b8484b264417
ef362f2773 rpc/gui: Remove 'Unknown block versions being mined' warning (Wladimir J. van der Laan)
Pull request description:
Due to miners inserting garbage into the version numbers causing false positives, the current version signalling has become completely useless. This removes the "unknown block versions" warning which has the tendency to scare users unnecessarily (and might get them to "update" to something bad).
It preserves the warning in the logs. Whether this is desirable can be a point of discussion.
Tree-SHA512: 51407ccd24a571462465d9c7180f0f28307c50b82a03284abe783e181d8ab7e0638dbb710698d883f28de8a609db70763e39be2470d956e67c833da0768e43e9
fa2cdc9ac2 test: Simplify create_cache (MarcoFalke)
fa25210d62 qa: Fix wallet_txn_doublespend issue (MarcoFalke)
1111aecbb5 qa: Always refresh stale cache to be out of ibd (MarcoFalke)
fab0d85802 qa: Remove mocktime unless required (MarcoFalke)
Pull request description:
When starting a test, we are always in IBD because the timestamps on cached blocks are in the past. Usually, we solve that by generating a block at the beginning of the test.
That is clumsy and might even lead to other problems such as #15360 and https://github.com/bitcoin/bitcoin/issues/14446#issuecomment-461926598
So fix that by getting rid of mocktime and always refreshing the last block of the cache when starting the test framework.
Should fix#14446
Tree-SHA512: 6af09800f9c86131349a103af617a54551f5f3f3260d38e14e3f30fdd3d91a0feb0100c56cbb12eae4aeac5571ae4b530b16345cbb831d2670237b53351a22c1
Due to miners inserting garbage into the version numbers, the current
version signalling has become completely useless. This removes the
"unknown block versions" warning which has the tendency to scare
users unnecessarily (and might get them to "update" to something
bad).
It preserves the warning in the logs. Whether this is desirable can
be a point of discussion.
3f5ad622e5 Enable PID file creation on Windows - Add available WIN PID function - Consider WIN32 in each relevant case - Add new preprocessor definitions to suppress warning - Update error message for generic OS (riordant)
Pull request description:
# Introduction
As discussed with @laanwj on IRC:
- PID file creation was never enabled for Windows, as the `pid_t` filetype is not available for it. However, the WIN32 API contains the header [`Processthreadsapi.h`](https://github.com/CodeShark/x86_64-w64-mingw32/blob/master/include/processthreadsapi.h) which in turn contains the function [`GetCurrentProcessId()`](https://docs.microsoft.com/en-gb/windows/desktop/api/processthreadsapi/nf-processthreadsapi-getcurrentprocessid). ~~This function is called at a higher level by [`_getpid()`](https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/getpid?view=vs-2017)~~ EDIT: `_getpid()` is not available to the MSVC compiler used in the AppVeyor build. As a result, I have changed the function call to`GetCurrentProcessId()`, which performs the same function and is available to both MinGW & MSVC.
This allows one to capture the PID in Windows, without any additional includes - the above function is already available.
- Within this PR, I have added a separate line that calls `GetCurrentProcessId()` in the case of a WIN compilation, and the usual `getpid()` otherwise. All code blocks processing PID file logic that avoid WIN32 have been changed to consider it. I have also updated the preprocessor definitions in `libbitcoin_server.vcxproj.in` to suppress a warning related to `std::strerror` for the MSVC build, that was causing the AppVeyor build to fail (see @fanquake comment below).
# Rationale
- Consistency between OS's running Bitcoin
- Applications which build off of `bitcoind`, such as novel front-end clients, often need access to the PID in order to control the daemon. Instead of designing some alternate way of doing this for one system, it should be consistent between all of them.
In collaboration with @joernroeder
Tree-SHA512: 22fcbf866e99115d12ed29716e68d200d4c118ae2f7b188b7705dc0cf5f0cd0ce5fb18f772744c6238eecd9e6d0922c615e2f0e12a7fe7c810062a79d97aa6a2
- Add available WIN PID function
- Consider WIN32 in each relevant case
- Add new preprocessor definitions to suppress warning
- Update error message for generic OS
Co-authored-by: Jörn Röder <kontakt@joernroeder.de>
fa4ce7038d rpc: Actually throw help when passed invalid number of params (MarcoFalke)
fa05626ca7 rpc: Add RPCHelpMan::IsValidNumArgs() (MarcoFalke)
Pull request description:
Can be tested by
* running the included test against an old binary (compiled without this patch)
* calling `setban 1 "add" 3 4 5 6 7 8 9 0` in the gui
Tree-SHA512: aa6a25bbe6f40722913ea292252a62a4012c964eed9f4035335a2e2d13be98eb60f368e8a3251a104a26a62c08b2cb926b06e5ab1418ef1cf4abdd71d87c2919
Note that the former 'else' branch in RewindBlockIndex is now
dealt with more naturally inside the EraseBlockData call (by
checking whether the parent needs to be re-added as candidate
after deleting a child).
39e20fc54f Add missing #include. (Daniel Kraft)
Pull request description:
bd0dbe8763 introduced a dependency of `rpc/util.h` on `RPCErrorCode`, defined in `rpc/protocol.h`. The latter file is only included from `rpc/util.cpp`, though. This commit fixes the missing include, by moving the `#include` of `rpc/protocol.h` to `rpc/util.h`.
Tree-SHA512: 75c03cfadb28a309d6deb36feeb0ee6ce0b38e8a1176919bc611ea720feff8c42ec9ed0ac8ab74ba9c531a3b7ec9ccbed0c8692ebdf5f9fc17867b9750a1d9f6
This also changes the uiInterface.LoadWallet signal argument type from
shared_ptr<CWallet> to unique_ptr<interfaces::Wallet> because CWallet is an
internal wallet class that shouldn't be used in non-wallet code (and also can't
be passed across process boundaries).
This commit does not change behavior.
This avoids a bogus circular dependency error in the next commit:
interfaces/chain -> interfaces/wallet -> wallet/wallet -> interfaces/chain
Which is incorrect, because interfaces/chain.cpp depends only on the
interfaces/wallet.h file, not the interfaces/wallet.cpp file, and it is
wrong to treat these as a unit. Inside the interfaces directory, .h files
contain abstract class definitions and .cpp files contain implementations of
those classes, so you don't need to link against .cpp files if you're only
using the abstract class definition in the .h file.
An alternative fix might be to rename all the cpp files in the interfaces
directory like: chain.cpp->chain_impl.cpp, node.cpp->node_impl.cpp. But just
getting the linter to treat these files as independent dependencies seemed
like it would allow keeping code organization straightforward and avoiding
the need to rename things.
