Merge tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4

Pull ext4 updates from Ted Ts'o:
 "For this cycle we add support for the shutdown ioctl, which is
  primarily used for testing, but which can be useful on production
  systems when a scratch volume is being destroyed and the data on it
  doesn't need to be saved.

  This found (and we fixed) a number of bugs with ext4's recovery to
  corrupted file system --- the bugs increased the amount of data that
  could be potentially lost, and in the case of the inline data feature,
  could cause the kernel to BUG.

  Also included are a number of other bug fixes, including in ext4's
  fscrypt, DAX, inline data support"

* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: (26 commits)
  ext4: rename EXT4_IOC_GOINGDOWN to EXT4_IOC_SHUTDOWN
  ext4: fix fencepost in s_first_meta_bg validation
  ext4: don't BUG when truncating encrypted inodes on the orphan list
  ext4: do not use stripe_width if it is not set
  ext4: fix stripe-unaligned allocations
  dax: assert that i_rwsem is held exclusive for writes
  ext4: fix DAX write locking
  ext4: add EXT4_IOC_GOINGDOWN ioctl
  ext4: add shutdown bit and check for it
  ext4: rename s_resize_flags to s_ext4_flags
  ext4: return EROFS if device is r/o and journal replay is needed
  ext4: preserve the needs_recovery flag when the journal is aborted
  jbd2: don't leak modified metadata buffers on an aborted journal
  ext4: fix inline data error paths
  ext4: move halfmd4 into hash.c directly
  ext4: fix use-after-iput when fscrypt contexts are inconsistent
  jbd2: fix use after free in kjournald2()
  ext4: fix data corruption in data=journal mode
  ext4: trim allocation requests to group size
  ext4: replace BUG_ON with WARN_ON in mb_find_extent()
  ...

1 files changed, 22 insertions, 3 deletions

diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 7ae43c59bc79..10c62de642c6 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -1556,7 +1556,17 @@ static int mb_find_extent(struct ext4_buddy *e4b, int block,
 		ex->fe_len += 1 << order;
 	}
 
-	BUG_ON(ex->fe_start + ex->fe_len > (1 << (e4b->bd_blkbits + 3)));
+	if (ex->fe_start + ex->fe_len > (1 << (e4b->bd_blkbits + 3))) {
+		/* Should never happen! (but apparently sometimes does?!?) */
+		WARN_ON(1);
+		ext4_error(e4b->bd_sb, "corruption or bug in mb_find_extent "
+			   "block=%d, order=%d needed=%d ex=%u/%d/%d@%u",
+			   block, order, needed, ex->fe_group, ex->fe_start,
+			   ex->fe_len, ex->fe_logical);
+		ex->fe_len = 0;
+		ex->fe_start = 0;
+		ex->fe_group = 0;
+	}
 	return ex->fe_len;
 }
 
@@ -2136,8 +2146,10 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 	 * We search using buddy data only if the order of the request
 	 * is greater than equal to the sbi_s_mb_order2_reqs
 	 * You can tune it via /sys/fs/ext4/<partition>/mb_order2_req
+	 * We also support searching for power-of-two requests only for
+	 * requests upto maximum buddy size we have constructed.
 	 */
-	if (i >= sbi->s_mb_order2_reqs) {
+	if (i >= sbi->s_mb_order2_reqs && i <= sb->s_blocksize_bits + 2) {
 		/*
 		 * This should tell if fe_len is exactly power of 2
 		 */
@@ -2207,7 +2219,7 @@ repeat:
 			}
 
 			ac->ac_groups_scanned++;
-			if (cr == 0 && ac->ac_2order < sb->s_blocksize_bits+2)
+			if (cr == 0)
 				ext4_mb_simple_scan_group(ac, &e4b);
 			else if (cr == 1 && sbi->s_stripe &&
 					!(ac->ac_g_ex.fe_len % sbi->s_stripe))
@@ -3123,6 +3135,13 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac,
 	if (ar->pright && start + size - 1 >= ar->lright)
 		size -= start + size - ar->lright;
 
+	/*
+	 * Trim allocation request for filesystems with artificially small
+	 * groups.
+	 */
+	if (size > EXT4_BLOCKS_PER_GROUP(ac->ac_sb))
+		size = EXT4_BLOCKS_PER_GROUP(ac->ac_sb);
+
 	end = start + size;
 
 	/* check we don't cross already preallocated blocks */