Fixes for obscure mempool-checkpoint interaction #2056

pull sipa wants to merge 2 commits into bitcoin:master from sipa:fix_2052 changing 1 files +5 −3

sipa commented at 11:45 PM on November 30, 2012: member
Implements the suggestions listed in #2052:
- Only move disconnected transactions to the mempool after the last checkpoint
- Always very scripts of transactions that do
Enable script verification for reorganized mempool tx 038784b979
BitcoinPullTester commented at 11:59 PM on November 30, 2012: none

Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/cb1e827f6d558b7cb311bc73e1bbdb770c468776 for binaries and test log.
in src/main.cpp:None in cb1e827f6d outdated
```
1705 | @@ -1706,7 +1706,7 @@ bool SetBestChain(CBlockIndex* pindexNew)
1706 |  
1707 |          // Queue memory transactions to resurrect
1708 |          BOOST_FOREACH(const CTransaction& tx, block.vtx)
1709 | -            if (!tx.IsCoinBase())
1710 | +            if (!tx.IsCoinBase() && pindex->nHeight > Checkpoints::GetTotalBlocksEstimate())
```
gavinandresen commented at 12:02 AM on December 1, 2012:

ACK if you add a comment something like: // height will be < checkpoint only when reading an old blockchain via -loadblock or -reindex // (skip re-org'ed transactions in those cases)

sipa commented at 12:03 AM on December 1, 2012:

@gavinandresen During normal IBD it will be < checkpoint too. As a follow-up, we should probably disable accepting things to the mempool during IBD altogether...

EDIT: right, but you shouldn't be receiving side chains when doing normal network IBD. Nothing prevents a peer from sending one nonetheless though.
Only send reorged txn to mempool after checkpoint 8259c57321
sipa commented at 1:12 AM on December 1, 2012: member

@gavinandresen Updated.
BitcoinPullTester commented at 1:30 AM on December 1, 2012: none

Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/8259c573212213e3cd8da7146af525e067c4ddaa for binaries and test log.
gavinandresen commented at 2:49 AM on December 1, 2012: contributor

ACK
Diapolo commented at 8:42 PM on December 4, 2012: none

I've been using this for a while and did not observe strange things. But my observations (as always) should be considered just entry-level in terms of client-usage-and-test-scenarios ^^, sorry :D.
jgarzik commented at 10:10 PM on December 5, 2012: contributor

ACK
sipa referenced this in commit cd1391ae4d on Dec 5, 2012
sipa merged this on Dec 5, 2012
sipa closed this on Dec 5, 2012
sipa deleted the branch on May 3, 2013
laudney referenced this in commit 71fbe178ad on Mar 19, 2014
DrahtBot locked this on Sep 8, 2021

Contributors

Linked (view graph)

#2052 Obscure checkpoint/mempool interaction