If the device contains on old logfs image and the journal is moved to
segment that have never been used by the current logfs and not all
journal segments are erased before the next mount, the old content can
confuse mount code.  To prevent this, always erase the new journal
segments.
Signed-off-by: NJoern Engel <joern@logfs.org>

Fixes a GC livelock.
Signed-off-by: NJoern Engel <joern@logfs.org>

do_logfs_journal_wl_pass() must call freeseg(), thereby clear
PagePrivate on all pages of the current journal segment.
Signed-off-by: NJoern Engel <joern@logfs.org>

A comment in the old code read:
        /* The math in this function can surely use some love */

And indeed it did.  In the case that area->a_used_bytes is exactly
4096 bytes below segment size it fell apart.  pad_wbuf is now split
into two helpers that are significantly less complicated.
Signed-off-by: NJoern Engel <joern@logfs.org>

The comment was correct, so make the code match the comment.  As the
new comment indicates, we might be able to do a little less work.  But
for the current -rc series let's keep it simple and just fix the bug.
Signed-off-by: NJoern Engel <joern@logfs.org>

Found by Al Viro.
Signed-off-by: NJoern Engel <joern@logfs.org>

rootdir was already allocated, so we must iput it again.
Found by Al Viro.
Signed-off-by: NJoern Engel <joern@logfs.org>

If the first superblock is wrong and the second gets written, there
will still be a mismatch on next mount.  Write both to make sure they
match.
Signed-off-by: NJoern Engel <joern@logfs.org>

Apparently filldir can sleep, which forbids kmap_atomic.
Signed-off-by: NJoern Engel <joern@logfs.org>

Signed-off-by: NJoern Engel <joern@logfs.org>

Intel SSDs have a limit of 0xffff as queue_max_hw_sectors(q).  Such a
limit may make sense from a hardware pov, but it causes bio_alloc() to
return NULL.
Signed-off-by: NJoern Engel <joern@logfs.org>

logfs_recover_sb() needs it open.
Signed-off-by: NJoern Engel <joern@logfs.org>

Many changes were made during development that could result in old
versions of mklogfs and the kernel code being subtly incompatible.
Not being a friend of subtleties, I hereby change the magic number.
Any old version of mklogfs is now guaranteed to fail.

Incompatible change: h_compr is moved up so the padding is all in one chunk.

This prevents unnecessary journal writes.  More importantly it prevents
an oops due to a journal write on failed mount.

Erases for block devices were always just emulated by writing 0xff.
Some time back the write was removed and only the page cache was
changed to 0xff.  Superficialy a good idea with two problems:
1. Touching the page cache isn't necessary either.
2. However, writing out 0xff _is_ necessary for the journal.  As the
   journal is scanned linearly, an old non-overwritten commit entry
   can be used on next mount and cause havoc.

This should fix both aspects.

Andrew Morton sayeth:
fs/logfs/journal.c: In function 'logfs_init_journal':
fs/logfs/journal.c:266: warning: 'last_len' may be used uninitialized in this function

Can this be squished please?

Randy Dunlap caught this built error on i386:

fs/built-in.o: In function `hash_index':
dir.c:(.text+0x6c1f2): undefined reference to `__umoddi3'

Spotted by Dan Carpenter.

This is a new flash file system. See
Documentation/filesystems/logfs.txt
Signed-off-by: NJoern Engel <joern@logfs.org>