Flood of retarget messages, if internal miner is used #1902

My public testnet3 node uses the internal miner. Immediately after block #32255, a flood of debug messages appeared. Many messages per second...

received block 000000000207e63e68f2
SetBestChain: new best=000000000207e63e68f2  height=32255  work=1162848638650514
  date=10/02/12 21:55:17
ProcessBlock: ACCEPTED
CreateNewBlock(): total size 1000
  nActualTimespan = 326059  before bounds
GetNextWorkRequired RETARGET
nTargetTimespan = 1209600    nActualTimespan = 326059
Before: 1c04b5af  0000000004b5af00000000000000000000000000000000000000000000000000
After:  1c014500  00000000014500d8027027027027027027027027027027027027027027027027
  nActualTimespan = 326059  before bounds
GetNextWorkRequired RETARGET
nTargetTimespan = 1209600    nActualTimespan = 326059
Before: 1c04b5af  0000000004b5af00000000000000000000000000000000000000000000000000
After:  1c014500  00000000014500d8027027027027027027027027027027027027027027027027
Running BitcoinMiner with 1 transactions in block (188 bytes)
81.161.177.190:18333 /Satoshi:0.7.0.3/ 32229 getblocks -1 to 00000000000000000000 limit 500
  nActualTimespan = 326059  before bounds
GetNextWorkRequired RETARGET
nTargetTimespan = 1209600    nActualTimespan = 326059
Before: 1c04b5af  0000000004b5af00000000000000000000000000000000000000000000000000
After:  1c014500  00000000014500d8027027027027027027027027027027027027027027027027
  nActualTimespan = 326059  before bounds
GetNextWorkRequired RETARGET
nTargetTimespan = 1209600    nActualTimespan = 326059
Before: 1c04b5af  0000000004b5af00000000000000000000000000000000000000000000000000
After:  1c014500  00000000014500d8027027027027027027027027027027027027027027027027
96.21.113.151:18333 /Satoshi:0.7.0.99/ 32229 getblocks -1 to 00000000000000000000 limit 500
88.198.7.53:18333 /Satoshi:0.7.0.3/ 32229 getblocks -1 to 00000000000000000000 limit 500
89.40.16.241:18333 /Satoshi:0.7.0.3/ 32229 getblocks -1 to 00000000000000000000 limit 500
178.63.48.141:18333 /Satoshi:0.7.0.3/ 32229 getblocks -1 to 00000000000000000000 limit 500
  nActualTimespan = 326059  before bounds
GetNextWorkRequired RETARGET
nTargetTimespan = 1209600    nActualTimespan = 326059
Before: 1c04b5af  0000000004b5af00000000000000000000000000000000000000000000000000
After:  1c014500  00000000014500d8027027027027027027027027027027027027027027027027

Retarget message continues ad infinitum, until internal miner is disabled (setgenerate false).

100% reproducible, when internal miner is turned on. 0% reproducible, when internal miner is turned off.

Turned the internal miner back on at height=32259. No flood of messages.

I'm guessing this occurs (a) only on testnet, and (b) only at a difficulty retarget boundary.