file corruption running eCryptfs on top of vmhgfs

[Michael R. <mr...@us...>]

Hi,

I am trying to run an eCryptfs volume on top of a vmhgfs and I run into problems: when writing to this stack with high bandwidth (copying many large files to it), there is a pretty good chance of data corruption. The persistent store contains data different from what has actually been written. The files have the correct size, but the content is wrong.

I have to make some more experiments to find out the exact circumstances. But it appears to work fine on my system, when copying directly to vmhgfs or when copying to an eCryptfs not running on vmhgfs. After some initial tests, it looks like the combination of eCryptfs and vmhgfs does not work, although the individual components do work.

I will try to have a look at the code to figure out what code paths in vmhgfs are triggered by eCryptfs. My first guess is that vmhgfs makes some hidden assumption that holds when using it through the system call interface directly, but that does not hold when the client is eCryptfs.

Do you have any suggestions where I should look? Do you have any ideas what could be going wrong here?

Michael