1
2 - support asynchronous operation -- add a per-fs 'reserved_space' count,
3   let each outstanding write reserve the _maximum_ amount of physical
4   space it could take. Let GC flush the outstanding writes because the
5   reservations will necessarily be pessimistic. With this we could even
6   do shared writable mmap, if we can have a fs hook for do_wp_page() to
7   make the reservation.
8 - disable compression in commit_write()?
9 - fine-tune the allocation / GC thresholds
10 - chattr support - turning on/off and tuning compression per-inode
11 - checkpointing (do we need this? scan is quite fast)
12 - make the scan code populate real inodes so read_inode just after 
13	mount doesn't have to read the flash twice for large files.
14	Make this a per-inode option, changable with chattr, so you can
15	decide which inodes should be in-core immediately after mount.
16 - test, test, test
17
18 - NAND flash support:
19	- almost done :)
20	- use bad block check instead of the hardwired byte check
21
22 - Optimisations:
23   - Split writes so they go to two separate blocks rather than just c->nextblock.
24	By writing _new_ nodes to one block, and garbage-collected REF_PRISTINE
25	nodes to a different one, we can separate clean nodes from those which
26	are likely to become dirty, and end up with blocks which are each far
27	closer to 100% or 0% clean, hence speeding up later GC progress dramatically.
28   - Stop keeping name in-core with struct jffs2_full_dirent. If we keep the hash in 
29     the full dirent, we only need to go to the flash in lookup() when we think we've
30     got a match, and in readdir(). 
31   - Doubly-linked next_in_ino list to allow us to free obsoleted raw_node_refs immediately?
32   - Remove size from jffs2_raw_node_frag. 
33
34dedekind:
351. __jffs2_flush_wbuf() has a strange 'pad' parameter. Eliminate.
362. get_sb()->build_fs()->scan() path... Why get_sb() removes scan()'s crap in
37   case of failure? scan() does not clean everything. Fix.
38