github
/
bitcoin
mirror of https://github.com/bitcoin/bitcoin
1
Fork 0
Code Releases Activity

index: race fix, lock cs_main while 'm_synced' is subject to change 

This ensures that the index does not miss any 'new block' signals
occurring in-between reading the 'next block' and setting 'm_synced'.
Because, if this were to happen, the ignored blocks would never be
indexed, thus stalling the index forever.
This commit is contained in:
Ryan Ofsky 2024-04-17 17:24:05 -03:00 committed by furszy
parent 3f6a6da3b0
commit 65951e0418
No known key found for this signature in database
GPG Key ID: 5DD23CCC686AA623
1 changed files with 14 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
src/index/base.cpp
View File

@ -160,12 +160,24 @@ void BaseIndex::Sync()
}
const CBlockIndex* pindex_next = WITH_LOCK(cs_main, return NextSyncBlock(pindex, m_chainstate->m_chain));
// If pindex_next is null, it means pindex is the chain tip, so
// commit data indexed so far.
if (!pindex_next) {
SetBestBlockIndex(pindex);
// No need to handle errors in Commit. See rationale above.
Commit();
m_synced = true;
break;
// If pindex is still the chain tip after committing, exit the
// sync loop. It is important for cs_main to be locked while
// setting m_synced = true, otherwise a new block could be
// attached while m_synced is still false, and it would not be
// indexed.
LOCK(::cs_main);
pindex_next = NextSyncBlock(pindex, m_chainstate->m_chain);
if (!pindex_next) {
m_synced = true;
break;
}
}
if (pindex_next->pprev != pindex && !Rewind(pindex, pindex_next->pprev)) {
FatalErrorf("%s: Failed to rewind index %s to a previous chain tip", __func__, GetName());