I am seeing this with Windows XP and Bitcoin Core 0.9.3. I had received an I/O error causing the client to crash. It is reindexing the block chain. I can't leave computer running indefinitely so I shutdown and begin again each day, hoping that it continues where it left off. This reindexing has been going on now for about a week working about 8 hours a day. During that time the CPU is 80-100%. Its difficult to do anything else on the laptop under these conditions. Its still about 20 weeks of data from completion. I will see how much progress it makes before I have to shutdown the laptop. I may take a backup of the Bitcoin folder and try erasing all the blocks to download from the very beginning. It would be worth it if I can use my laptop once again for other work, even if it takes longer to finish. My worry is that whatever is causing the extreme CPU usage might not go away if I let the reindexing continue. This 80-100% CPU usage is unacceptable even for the reindexing. If it should persist once the block chain becomes current, I would have to restart anyway.
With the backup of the Bitcoin folder, I will be able to resume the reindexing if I have to.
As a followup, I have been trying to get Bitcoin Core 0.9.3 to fully sync up for several weeks. I let the process run for about an hour before attempting to terminate.
1. It may be my imagination, but it seems to terminate more cleanly with File->Exit than if the Windows "X" option is invoked.
2. If it is running away with the CPU, it is usually not responsive to mouse clicks and has to be forcefully terminated.
3. Often something goes wrong as process tries to terminate from File->Exit request. Sometimes it then begins running away with the CPU. Sometimes, it issues the warning message not kill process or shutdown until the message disappears. Other times that message never disappears. I've made it a policy to wait not only for the message to disappear, but also to wait for the process to exit of its own accord.
If the process exit cleanly after a File->Exit request. I take a check-point backup of the data folder. If something has gone wrong, I take no chances and restore from the last check-point backup.
After several weeks, and almost 80 check-point backups taken, I'm about 6 months from being fully synced. Its slow going, but at least its making progress without corrupting the data folder as it is prone to do.