Eric Sandeen wrote:
Once I started looking at the pattern of extent buffer reductions before
and after calling compact_page/full
I noticed even when we did a partial move the number of total buffers
didn't go down.
I suppose you could end up with the stars and moon lining up just right
and you would
Lachlan McIlroy wrote:
Here's a patch to remove xfs_iext_irec_compact_full() like Russell
did in his original patch - are you guys happy with this?
I'm putting it through it's paces now and so far it looks good.
I'll have to think more about it, honestly. Probably fine, but I've not
looked at all the surrounding code, I was so far just looking for the
do enough partial moves to free up a page.
Since this is all incore buffers space we are talking about all these
space optimizations are
moot once the inode goes inactive and is flushed from cache.
I can't really think of a situation where not doing partial extent moves
is really going to
create an issue but I might be missing something.
(FWIW, compact_full *does* get called reasonably frequently, but the
memmove case is what's hard to hit...)
--- a/fs/xfs/xfs_inode.c 2008-09-19 13:08:08.000000000 +1000
+++ b/fs/xfs/xfs_inode.c 2008-09-19 13:16:34.000000000 +1000
@@ -4157,7 +4166,7 @@ xfs_iext_indirect_to_direct(
ASSERT(nextents <= XFS_LINEAR_EXTS);
size = nextents * sizeof(xfs_bmbt_rec_t);
ASSERT(ifp->if_real_bytes == XFS_IEXT_BUFSZ);
ep = ifp->if_u1.if_ext_irec->er_extbuf;