Bypass reloading blocks from disk #4764

Pass the currently-being-processed block down to ConnectTip, so it can avoid loading it again from disk.

This results in a 15% speedup for me for reindexing until block 172896 (134s vs 155s).

Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/p4764_92bb6f2f1729192703dd49bb982c1dd4c0062fa0/ for binaries and test log. This test script verifies pulls every time they are updated. It, however, dies sometimes and fails to test properly. If you are waiting on a test, please check timestamps to verify that the test.log is moving at http://jenkins.bluematt.me/pull-tester/current/ Contact BlueMatt on freenode if something looks broken.

Nice, code changes look good to me, going to test this.

Edit: I benchmarked a full testnet reindex with and without this patch (with time src/bitcoind -datadir=/store/tmp/testbtc -testnet -reindex -stopafterblockimport -disablewallet):

With:
real    21m2.153s user    74m18.716s sys     0m43.476s
real    22m30.468s user    73m45.495s sys     0m41.732s
real    20m12.181s user    74m13.455s sys     0m42.349s

Without:
real    21m2.175s user    74m38.421s sys     0m47.715s
real    20m15.575s user    74m15.827s sys     0m46.672s

Amount of time in 'sys' is a noticeably less with the patch, but the overall time doesn't decrease within the variance of the results.

Repeating your test on the mainnet (time src/bitcoind -datadir=/store/tmp/testbtc -reindex -stopafterblockimport -disablewallet -stopafterblock=172896, dropping kernel caches in between):

With:
real    5m10.073s user    3m11.530s sys     0m19.505s
real    5m6.246s user    3m8.900s sys     0m18.981s
real    5m8.438s user    3m5.312s sys     0m19.045s

Without:
real    5m32.379s user    3m30.707s sys     0m22.530s
real    5m37.136s user    3m32.757s sys     0m23.048s
real    5m33.290s user    3m30.284s sys     0m22.501s

Although I don't get to 15% there is a clear improvement here. ACK.

ut ACK

ut ACK