>>>>> " " == Juan J Quintela <quintela@fi.udc.es> writes:
> Trond, I have not an SMP machine (yet), and I can not tell you
> numbers now. I put the counter there to show that we *may*
> want to limit the latency there. I am thinking in the write of
> a big file, that can take a lot to free all the pages, but I
I'm pretty SMP-less myself at the moment (I'm visiting in Strasbourg
again), so I'm afraid I cannot run the test for you.
> By the way, while we are here, the only difference between
> truncate_inode_pages and invalidate_inode_pages is the one that
> you told here before? I am documenting some of the MM stuff,
> and your comments in that aspect are really wellcome. (You
> will have noted now that I am quite newbie here).
Well. As far as NFS and other non-disk based systems are concerned
that is the functional difference between the two. That and the fact
that truncate_inode_pages() takes an offset as an argument.
For disk-based systems, they are very different beasts, since
truncate_inode_pages() will also attempt to invalidate and/or wait on
any pending buffers on the pages it clears out.
Strictly speaking therefore, one should not confuse the two, however
truncate_inode_pages() is (ab)used as a sleeping substitute for
invalidate_inode_pages() by some of the icache pruning code in
fs/inode.c.
Cheers,
Trond
-
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 : Mon May 15 2000 - 21:00:19 EST