[ Monday, February 14, 2000 ] breed@almaden.ibm.com wrote:
> On a related issue, if there was a char generic_data[] member in the inode
> union that had a defined size (probably the size of the largest structure),
> people (like me) writing filesystems that aren't ready to be put into the
> sources yet could put data right in the union and know that it will fit
> without having to resort to using generic_ip and allocating an info block
> to store inode specific info. (Narly sentence huh?)
That would seem to place a restriction on the size of your inode data,
something generic_ip frees you from (at not a high cost imho). I'd
go ahead and advocate everyone having to go through the indirection
and rip out the entire union for a void *inode_info first (helping a
good bit towards freeing vfs from specific fs's). Then do the same
for the super_block struct and you seem to be well on your way to the
factored vfs, as you'll be rid of fs-specific includes in linux/fs.h
(attached). Whether the indirection and sacrificed cache line is worth
it is the call of another.
fs.h | 99 +------------------------------------------------------------------ 1 files changed, 2 insertions, 97 deletions
James
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Feb 15 2000 - 21:00:29 EST