Attempting to mount initial drive while s3backer is running corrupts data

[GoogleCodeExporter]

What steps will reproduce the problem?
1. mount the first filesystem 
2. mount the child filesystem, and create a bunch of data 
3. unmount both filesystems
4. ps -Af | grep s3backer, and note the process is still running (sending data)
5. mount the first filesystem again.  Note that 'file' is now corrupted.

What is the expected output? What do you see instead?
I would expect it to either fail to mount the first filesystem, or wait for it 
to finish sending the necessary data.  As is, it appears to be killing the 
original process.

What version of the product are you using? On what operating system?
centos 6, s3backer 1.3.4

Please provide any additional information below.
fstab:
s3backer#backup     /s3/dev/s3backer/backup      fuse    
noauto,size=500g,blockSize=1024k,encrypt,compress=9,passwordFile=xxxxxxxxxxxx,ac
cessFile=xxxxxxxxxxxx,bloc\
kCacheSize=512000,md5CacheSize=512000,md5CacheTime=0,blockCacheFile=/s3/cache/ba
ckup/cachefile,blockCacheWriteDelay=60000,blockCacheThreads=5,timeout=60,listBlo
cks,rrs,ssl,insecure,vhost  0 0     

# backup disk looped onto s3backer mount                                        


/s3/dev/s3backer/backup/file          /s3/backup        ext4 
noauto,loop,noatime,nodiratime,sync  0       0

Original issue reported on code.google.com by bren...@thecontes.info on 30 Apr 2013 at 2:48