In the previous code WaitGroup db.wg.Add(), Wait() and db.closed were
not treated in sync. In particular, it was theoretically possible to
first check closed, then set closed and Wait, and then call wg.Add()
while waiting (leading to WaitGroup panic).
In theory it was also possible that db.env.BeginTxn() is called on a
closed or nil db.env, because db.wg.Add() was called only after BeginTxn
(db.wg.Wait() could already return).
WaitGroup is replaced with a Cond variable.
Now it is not possible to increase the active transactions count on a
closed database. It is also not possible to call BeginTxn on a closed
database.
logic: 1/42 of ram, but not more than 2Gb for chandata and not more than
256mb for other databases.
---------
Co-authored-by: battlmonstr <battlmonstr@users.noreply.github.com>
This introduces _experimental_ RPC daemon run by embedded Silkworm
library. Same notes as in PR #8353 apply here plus the following ones:
- activated if `http` command-line option is enabled and `silkworm.path`
option is present, nothing more is required (i.e. currently, both block
execution and RPC daemon run by Silkworm when specifying
`silkworm.path`, just to keep things as simple as possible)
- only Execution API endpoints are implemented by Silkworm RPCDaemon,
whilst Engine API endpoints are still served by Erigon RPCDaemon
- some features are still missing, in particular:
- state change notification handling
- custom JSON RPC settings (i.e. Erigon RPC settings are not passed to
Silkworm yet)
This introduces _experimental_ block execution run by embedded Silkworm
API library:
- new command-line option `silkworm.path` to enable the feature by
specifying the path to the Silkworm library
- the Silkworm API shared library is dynamically loaded on-demand
- currently requires to build Silkworm library on the target machine
- available only on Linux at the moment: macOS has issue with [stack
size](https://github.com/golang/go/issues/28024) and Windows would
require [TDM-GCC-64](https://jmeubank.github.io/tdm-gcc/), both need
dedicated effort for an assessment
This is to fix an issue with resource usage if the db.size.limit is
increased from its default setting of 2TB. This is applied to the chain
DB, but should not be used on the consensus DB which has smaller data
requirements. Expanding both DBs results in excessive RAM being reserved
by the underlying OS.
