[Unionfs] RHEL 2.6.9 hangs with unionfs-1.1.5

[Erez Zadok]

In message <20061130183141.GA32201 at filer.fsl.cs.sunysb.edu>, Josef Sipek writes:

> As I mentioned above, there is no way that we can know if a file is sparse
> or not. We could of course try to re-sparsify the file during copyup, but
> checking every block for being full of \0 is rather expensive.

struct stat contains both the number of bytes and number of blocks in the
file.  With some simple math, you can figure out if a file is sparse or not
(but not where the holes are).

When we copyup a file, what'd happen if we simply NOT copy over pages which
are all zeros?  Wouldn't that be the same as seeking past the end of the
file and writing the next non-zero page?  Wouldn't that create a copied-up
file that's sparse (assuming the target f/s supports it).

> Thanks,

Erez.