Hi Pete,
>
> Random stuff in pvfs2-cache.c, test/configure{,.in} shouldn't be in
> the patch.
Ok, checked these in separately :)
>
> The introduction of new FIXMEs is a bit worrisome.
>
> You don't need to hash in the fs_id anymore, like you were doing
> last year? I don't grok the internals well, so ignore if it's
> not a sensible question.
>
I wish I could answer all these questions with certainty.
The reason I was hashing the fs_id earlier was so that I could have a high
probability of a unique inode->i_ino field.
I am not sure what/if any tools depend on the i_ino field to detect if 2
directory entries are the same or different object. Programs like tar
probably depend on i_ino to check if 2 objects are hard links or not,
although it is a moot point for us since we do not support hard links.
So I decided to run some more tests (perhaps with a smaller handle range )
and see if I can coax the code through corner cases and hold off checking
it in for a while..
Thanks again for the reviews!
Murali
_______________________________________________
Pvfs2-developers mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers
