v1.7 Release

Finish the release notes for the 1.7 release.

Signed-off-by: Zach Brown <zab@versity.com>

ReleaseNotes.md

@@ -1,6 +1,29 @@
 Versity ScoutFS Release Notes
 =============================
 
+---
+v1.7
+\
+*Aug 26, 2022*
+
+* **Fixed possible persistent errors moving freed data extents**
+\
+  Fixed a case where the server could hit persistent errors trying to
+  move a client's freed extents in one commit.  The client had to free
+  a large number of extents that occupied distant positions in the
+  global free extent btree.  Very large fragmented files could cause
+  this.  The server now moves the freed extents in multiple commits and
+  can always ensure forward progress.
+
+* **Fixed possible persistent errors from freed duplicate extents**
+\
+  Background orphan deletion wasn't properly synchronizing with
+  foreground tasks deleting very large files.  If a deletion took long
+  enough then background deletion could also attempt to delete inode items
+  while the deletion was making progress.  This could create duplicate
+  deletions of data extent items which causes the server to abort when
+  it later discovers the duplicate extents as it merges free lists.
+
 ---
 v1.6
 \

kmod/src/Makefile

@@ -46,10 +46,6 @@ scoutfs-y +=			\
 	volopt.o		\
 	xattr.o
 
-ifdef KC_BUILD_KERNELCOMPAT
-scoutfs-y += kernelcompat.o
-endif
-
 #
 # The raw types aren't available in userspace headers.  Make sure all
 # the types we use in the headers are the exported __ versions.

kmod/src/Makefile.kernelcompat

@@ -34,52 +34,3 @@ endif
 ifneq (,$(shell grep 'FMODE_KABI_ITERATE' include/linux/fs.h))
 ccflags-y += -DKC_FMODE_KABI_ITERATE
 endif
-
-#
-# v4.11-12447-g104b4e5139fe
-#
-# Renamed __percpu_counter_add to percpu_counter_add_batch to clarify
-# that the __ wasn't less safe, just took an extra parameter.
-#
-ifneq (,$(shell grep 'percpu_counter_add_batch' include/linux/percpu_counter.h))
-ccflags-y += -DKC_PERCPU_COUNTER_ADD_BATCH
-endif
-
-#
-# v4.11-4550-g7dea19f9ee63
-#
-# Introduced memalloc_nofs_{save,restore} preferred instead of _noio_.
-#
-ifneq (,$(shell grep 'memalloc_nofs_save' include/linux/sched/mm.h))
-ccflags-y += -DKC_MEMALLOC_NOFS_SAVE
-endif
-
-#
-# v4.7-12414-g1eff9d322a44
-#
-# Renamed bi_rw to bi_opf to force old code to catch up.  We use it as a
-# single switch between old and new bio structures.
-#
-ifneq (,$(shell grep 'bi_opf' include/linux/blk_types.h))
-ccflags-y += -DKC_BIO_BI_OPF
-endif
-
-#
-# v4.12-rc2-201-g4e4cbee93d56
-#
-# Moves to bi_status BLK_STS_ API instead of having a mix of error
-# end_io args or bi_error.
-#
-ifneq (,$(shell grep 'bi_status' include/linux/blk_types.h))
-ccflags-y += -DKC_BIO_BI_STATUS
-endif
-
-#
-# v3.11-8765-ga0b02131c5fc
-#
-# Remove the old ->shrink() API, ->{scan,count}_objects is preferred.
-#
-ifneq (,$(shell grep '(*shrink)' include/linux/shrinker.h))
-ccflags-y += -DKC_SHRINKER_SHRINK
-KC_BUILD_KERNELCOMPAT=1
-endif

kmod/src/alloc.c

@@ -892,12 +892,11 @@ static int find_zone_extent(struct super_block *sb, struct scoutfs_alloc_root *r
  * -ENOENT is returned if we run out of extents in the source tree
  * before moving the total.
  *
- * If meta_reserved is non-zero then -EINPROGRESS can be returned if the
- * current meta allocator's avail blocks or room for freed blocks would
- * have fallen under the reserved amount.  The could have been
- * successfully dirtied in this case but the number of blocks moved is
- * not returned.  The caller is expected to deal with the partial
- * progress by commiting the dirty trees and examining the resulting
+ * If meta_budget is non-zero then -EINPROGRESS can be returned if the
+ * the caller's budget is consumed in the allocator during this call
+ * (though not necessarily by us, we don't have per-thread tracking of
+ * allocator consumption :/).  The call can still have made progress and
+ * caller is expected commit the dirty trees and examining the resulting
  * modified trees to see if they need to continue moving extents.
  *
  * The caller can specify that extents in the source tree should first
@@ -914,7 +913,7 @@ int scoutfs_alloc_move(struct super_block *sb, struct scoutfs_alloc *alloc,
 		       struct scoutfs_block_writer *wri,
 		       struct scoutfs_alloc_root *dst,
 		       struct scoutfs_alloc_root *src, u64 total,
-		       __le64 *exclusive, __le64 *vacant, u64 zone_blocks, u64 meta_reserved)
+		       __le64 *exclusive, __le64 *vacant, u64 zone_blocks, u64 meta_budget)
 {
 	struct alloc_ext_args args = {
 		.alloc = alloc,
@@ -922,6 +921,8 @@ int scoutfs_alloc_move(struct super_block *sb, struct scoutfs_alloc *alloc,
 	};
 	struct scoutfs_extent found;
 	struct scoutfs_extent ext;
+	u32 avail_start = 0;
+	u32 freed_start = 0;
 	u64 moved = 0;
 	u64 count;
 	int ret = 0;
@@ -932,6 +933,9 @@ int scoutfs_alloc_move(struct super_block *sb, struct scoutfs_alloc *alloc,
 		vacant = NULL;
 	}
 
+	if (meta_budget != 0)
+		scoutfs_alloc_meta_remaining(alloc, &avail_start, &freed_start);
+
 	while (moved < total) {
 		count = total - moved;
 
@@ -964,10 +968,10 @@ int scoutfs_alloc_move(struct super_block *sb, struct scoutfs_alloc *alloc,
 		if (ret < 0)
 			break;
 
-		if (meta_reserved != 0 &&
-		    scoutfs_alloc_meta_low(sb, alloc, meta_reserved +
-					   extent_mod_blocks(src->root.height) +
-					   extent_mod_blocks(dst->root.height))) {
+		if (meta_budget != 0 &&
+		    scoutfs_alloc_meta_low_since(alloc, avail_start, freed_start, meta_budget,
+						 extent_mod_blocks(src->root.height) +
+						 extent_mod_blocks(dst->root.height))) {
 			ret = -EINPROGRESS;
 			break;
 		}
@@ -1351,6 +1355,27 @@ void scoutfs_alloc_meta_remaining(struct scoutfs_alloc *alloc, u32 *avail_total,
 	} while (read_seqretry(&alloc->seqlock, seq));
 }
 
+/*
+ * Returns true if the caller's consumption of nr from either avail or
+ * freed would end up exceeding their budget relative to the starting
+ * remaining snapshot they took.
+ */
+bool scoutfs_alloc_meta_low_since(struct scoutfs_alloc *alloc, u32 avail_start, u32 freed_start,
+				  u32 budget, u32 nr)
+{
+	u32 avail_use;
+	u32 freed_use;
+	u32 avail;
+	u32 freed;
+
+	scoutfs_alloc_meta_remaining(alloc, &avail, &freed);
+
+	avail_use = avail_start - avail;
+	freed_use = freed_start - freed;
+
+	return ((avail_use + nr) > budget) || ((freed_use + nr) > budget);
+}
+
 bool scoutfs_alloc_test_flag(struct super_block *sb,
 			    struct scoutfs_alloc *alloc, u32 flag)
 {

kmod/src/alloc.h

@@ -131,7 +131,7 @@ int scoutfs_alloc_move(struct super_block *sb, struct scoutfs_alloc *alloc,
 		       struct scoutfs_block_writer *wri,
 		       struct scoutfs_alloc_root *dst,
 		       struct scoutfs_alloc_root *src, u64 total,
-		       __le64 *exclusive, __le64 *vacant, u64 zone_blocks, u64 meta_reserved);
+		       __le64 *exclusive, __le64 *vacant, u64 zone_blocks, u64 meta_budget);
 int scoutfs_alloc_insert(struct super_block *sb, struct scoutfs_alloc *alloc,
 			 struct scoutfs_block_writer *wri, struct scoutfs_alloc_root *root,
 			 u64 start, u64 len);
@@ -159,6 +159,8 @@ int scoutfs_alloc_splice_list(struct super_block *sb,
 bool scoutfs_alloc_meta_low(struct super_block *sb,
 			    struct scoutfs_alloc *alloc, u32 nr);
 void scoutfs_alloc_meta_remaining(struct scoutfs_alloc *alloc, u32 *avail_total, u32 *freed_space);
+bool scoutfs_alloc_meta_low_since(struct scoutfs_alloc *alloc, u32 avail_start, u32 freed_start,
+				  u32 budget, u32 nr);
 bool scoutfs_alloc_test_flag(struct super_block *sb,
 			    struct scoutfs_alloc *alloc, u32 flag);
 

kmod/src/block.c

@@ -21,7 +21,6 @@
 #include <linux/blkdev.h>
 #include <linux/rhashtable.h>
 #include <linux/random.h>
-#include <linux/sched/mm.h>
 
 #include "format.h"
 #include "super.h"
@@ -58,7 +57,7 @@ struct block_info {
 	atomic64_t access_counter;
 	struct rhashtable ht;
 	wait_queue_head_t waitq;
-	KC_DEFINE_SHRINKER(shrinker);
+	struct shrinker shrinker;
 	struct work_struct free_work;
 	struct llist_head free_llist;
 };
@@ -129,7 +128,7 @@ static __le32 block_calc_crc(struct scoutfs_block_header *hdr, u32 size)
 static struct block_private *block_alloc(struct super_block *sb, u64 blkno)
 {
 	struct block_private *bp;
-	unsigned int nofs_flags;
+	unsigned int noio_flags;
 
 	/*
 	 * If we had multiple blocks per page we'd need to be a little
@@ -157,9 +156,9 @@ static struct block_private *block_alloc(struct super_block *sb, u64 blkno)
 		 * spurious reclaim-on dependencies and warnings.
 		 */
 		lockdep_off();
-		nofs_flags = memalloc_nofs_save();
+		noio_flags = memalloc_noio_save();
 		bp->virt = __vmalloc(SCOUTFS_BLOCK_LG_SIZE, GFP_NOFS | __GFP_HIGHMEM, PAGE_KERNEL);
-		memalloc_nofs_restore(nofs_flags);
+		memalloc_noio_restore(noio_flags);
 		lockdep_on();
 
 		if (!bp->virt) {
@@ -437,10 +436,11 @@ static void block_remove_all(struct super_block *sb)
  * possible.  Final freeing, verifying checksums, and unlinking errored
  * blocks are all done by future users of the blocks.
  */
-static void block_end_io(struct super_block *sb, unsigned int opf,
+static void block_end_io(struct super_block *sb, int rw,
 			 struct block_private *bp, int err)
 {
 	DECLARE_BLOCK_INFO(sb, binf);
+	bool is_read = !(rw & WRITE);
 
 	if (err) {
 		scoutfs_inc_counter(sb, block_cache_end_io_error);
@@ -450,7 +450,7 @@ static void block_end_io(struct super_block *sb, unsigned int opf,
 	if (!atomic_dec_and_test(&bp->io_count))
 		return;
 
-	if (!op_is_write(opf) && !test_bit(BLOCK_BIT_ERROR, &bp->bits))
+	if (is_read && !test_bit(BLOCK_BIT_ERROR, &bp->bits))
 		set_bit(BLOCK_BIT_UPTODATE, &bp->bits);
 
 	clear_bit(BLOCK_BIT_IO_BUSY, &bp->bits);
@@ -463,13 +463,13 @@ static void block_end_io(struct super_block *sb, unsigned int opf,
 		wake_up(&binf->waitq);
 }
 
-static void KC_DECLARE_BIO_END_IO(block_bio_end_io, struct bio *bio)
+static void block_bio_end_io(struct bio *bio, int err)
 {
 	struct block_private *bp = bio->bi_private;
 	struct super_block *sb = bp->sb;
 
 	TRACE_BLOCK(end_io, bp);
-	block_end_io(sb, kc_bio_get_opf(bio), bp, kc_bio_get_errno(bio));
+	block_end_io(sb, bio->bi_rw, bp, err);
 	bio_put(bio);
 }
 
@@ -477,7 +477,7 @@ static void KC_DECLARE_BIO_END_IO(block_bio_end_io, struct bio *bio)
  * Kick off IO for a single block.
  */
 static int block_submit_bio(struct super_block *sb, struct block_private *bp,
-			    unsigned int opf)
+			    int rw)
 {
 	struct scoutfs_sb_info *sbi = SCOUTFS_SB(sb);
 	struct bio *bio = NULL;
@@ -510,9 +510,8 @@ static int block_submit_bio(struct super_block *sb, struct block_private *bp,
 				break;
 			}
 
-			kc_bio_set_opf(bio, opf);
-			kc_bio_set_sector(bio, sector + (off >> 9));
-			bio_set_dev(bio, sbi->meta_bdev);
+			bio->bi_sector = sector + (off >> 9);
+			bio->bi_bdev = sbi->meta_bdev;
 			bio->bi_end_io = block_bio_end_io;
 			bio->bi_private = bp;
 
@@ -529,18 +528,18 @@ static int block_submit_bio(struct super_block *sb, struct block_private *bp,
 			BUG();
 
 		if (!bio_add_page(bio, page, PAGE_SIZE, 0)) {
-			submit_bio(bio);
+			submit_bio(rw, bio);
 			bio = NULL;
 		}
 	}
 
 	if (bio)
-		submit_bio(bio);
+		submit_bio(rw, bio);
 
 	blk_finish_plug(&plug);
 
 	/* let racing end_io know we're done */
-	block_end_io(sb, opf, bp, ret);
+	block_end_io(sb, rw, bp, ret);
 
 	return ret;
 }
@@ -641,7 +640,7 @@ static struct block_private *block_read(struct super_block *sb, u64 blkno)
 
 	if (!test_bit(BLOCK_BIT_UPTODATE, &bp->bits) &&
 	     test_and_clear_bit(BLOCK_BIT_NEW, &bp->bits)) {
-		ret = block_submit_bio(sb, bp, REQ_OP_READ);
+		ret = block_submit_bio(sb, bp, READ);
 		if (ret < 0)
 			goto out;
 	}
@@ -940,7 +939,7 @@ int scoutfs_block_writer_write(struct super_block *sb,
 		/* retry previous write errors */
 		clear_bit(BLOCK_BIT_ERROR, &bp->bits);
 
-		ret = block_submit_bio(sb, bp, REQ_OP_WRITE);
+		ret = block_submit_bio(sb, bp, WRITE);
 		if (ret < 0)
 			break;
 	}
@@ -1040,17 +1039,6 @@ u64 scoutfs_block_writer_dirty_bytes(struct super_block *sb,
 	return wri->nr_dirty_blocks * SCOUTFS_BLOCK_LG_SIZE;
 }
 
-static unsigned long block_count_objects(struct shrinker *shrink, struct shrink_control *sc)
-{
-	struct block_info *binf = container_of(shrink, struct block_info, shrinker);
-	struct super_block *sb = binf->sb;
-
-	scoutfs_inc_counter(sb, block_cache_scan_objects);
-
-	return min_t(u64, (u64)atomic_read(&binf->total_inserted) * SCOUTFS_BLOCK_LG_PAGES_PER,
-		     ULONG_MAX / 2); /* magic numbers as we approach ~0UL :/ */
-}
-
 /*
  * Remove a number of cached blocks that haven't been used recently.
  *
@@ -1071,19 +1059,23 @@ static unsigned long block_count_objects(struct shrinker *shrink, struct shrink_
  * atomically remove blocks when the only references are ours and the
  * hash table.
  */
-static unsigned long block_scan_objects(struct shrinker *shrink, struct shrink_control *sc)
+static int block_shrink(struct shrinker *shrink, struct shrink_control *sc)
 {
-	struct block_info *binf = container_of(shrink, struct block_info, shrinker);
+	struct block_info *binf = container_of(shrink, struct block_info,
+					       shrinker);
 	struct super_block *sb = binf->sb;
 	struct rhashtable_iter iter;
 	struct block_private *bp;
-	unsigned long freed = 0;
 	unsigned long nr;
 	u64 recently;
 
-	scoutfs_inc_counter(sb, block_cache_scan_objects);
+	nr = sc->nr_to_scan;
+	if (nr == 0)
+		goto out;
 
-	nr = DIV_ROUND_UP(sc->nr_to_scan, SCOUTFS_BLOCK_LG_PAGES_PER);
+	scoutfs_inc_counter(sb, block_cache_shrink);
+
+	nr = DIV_ROUND_UP(nr, SCOUTFS_BLOCK_LG_PAGES_PER);
 
 restart:
 	recently = accessed_recently(binf);
@@ -1126,7 +1118,6 @@ restart:
 			if (block_remove_solo(sb, bp)) {
 				scoutfs_inc_counter(sb, block_cache_shrink_remove);
 				TRACE_BLOCK(shrink, bp);
-				freed++;
 				nr--;
 			}
 			block_put(sb, bp);
@@ -1135,8 +1126,9 @@ restart:
 
 	rhashtable_walk_stop(&iter);
 	rhashtable_walk_exit(&iter);
-
-	return freed;
+out:
+	return min_t(u64, (u64)atomic_read(&binf->total_inserted) * SCOUTFS_BLOCK_LG_PAGES_PER,
+		     INT_MAX);
 }
 
 struct sm_block_completion {
@@ -1144,11 +1136,11 @@ struct sm_block_completion {
 	int err;
 };
 
-static void KC_DECLARE_BIO_END_IO(sm_block_bio_end_io, struct bio *bio)
+static void sm_block_bio_end_io(struct bio *bio, int err)
 {
 	struct sm_block_completion *sbc = bio->bi_private;
 
-	sbc->err = kc_bio_get_errno(bio);
+	sbc->err = err;
 	complete(&sbc->comp);
 	bio_put(bio);
 }
@@ -1163,8 +1155,9 @@ static void KC_DECLARE_BIO_END_IO(sm_block_bio_end_io, struct bio *bio)
  * only layer that sees the full block buffer so we pass the calculated
  * crc to the caller for them to check in their context.
  */
-static int sm_block_io(struct super_block *sb, struct block_device *bdev, unsigned int opf,
-		       u64 blkno, struct scoutfs_block_header *hdr, size_t len, __le32 *blk_crc)
+static int sm_block_io(struct super_block *sb, struct block_device *bdev, int rw, u64 blkno,
+		       struct scoutfs_block_header *hdr, size_t len,
+		       __le32 *blk_crc)
 {
 	struct scoutfs_block_header *pg_hdr;
 	struct sm_block_completion sbc;
@@ -1178,7 +1171,7 @@ static int sm_block_io(struct super_block *sb, struct block_device *bdev, unsign
 		return -EIO;
 
 	if (WARN_ON_ONCE(len > SCOUTFS_BLOCK_SM_SIZE) ||
-	    WARN_ON_ONCE(!op_is_write(opf) && !blk_crc))
+	    WARN_ON_ONCE(!(rw & WRITE) && !blk_crc))
 		return -EINVAL;
 
 	page = alloc_page(GFP_NOFS);
@@ -1187,7 +1180,7 @@ static int sm_block_io(struct super_block *sb, struct block_device *bdev, unsign
 
 	pg_hdr = page_address(page);
 
-	if (op_is_write(opf)) {
+	if (rw & WRITE) {
 		memcpy(pg_hdr, hdr, len);
 		if (len < SCOUTFS_BLOCK_SM_SIZE)
 			memset((char *)pg_hdr + len, 0,
@@ -1201,9 +1194,8 @@ static int sm_block_io(struct super_block *sb, struct block_device *bdev, unsign
 		goto out;
 	}
 
-	bio->bi_opf = opf | REQ_SYNC;
-	kc_bio_set_sector(bio, blkno << (SCOUTFS_BLOCK_SM_SHIFT - 9));
-	bio_set_dev(bio, bdev);
+	bio->bi_sector = blkno << (SCOUTFS_BLOCK_SM_SHIFT - 9);
+	bio->bi_bdev = bdev;
 	bio->bi_end_io = sm_block_bio_end_io;
 	bio->bi_private = &sbc;
 	bio_add_page(bio, page, SCOUTFS_BLOCK_SM_SIZE, 0);
@@ -1211,12 +1203,12 @@ static int sm_block_io(struct super_block *sb, struct block_device *bdev, unsign
 	init_completion(&sbc.comp);
 	sbc.err = 0;
 
-	submit_bio(bio);
+	submit_bio((rw & WRITE) ? WRITE_SYNC : READ_SYNC, bio);
 
 	wait_for_completion(&sbc.comp);
 	ret = sbc.err;
 
-	if (ret == 0 && !op_is_write(opf)) {
+	if (ret == 0 && !(rw & WRITE)) {
 		memcpy(hdr, pg_hdr, len);
 		*blk_crc = block_calc_crc(pg_hdr, SCOUTFS_BLOCK_SM_SIZE);
 	}
@@ -1230,14 +1222,14 @@ int scoutfs_block_read_sm(struct super_block *sb,
 			  struct scoutfs_block_header *hdr, size_t len,
 			  __le32 *blk_crc)
 {
-	return sm_block_io(sb, bdev, REQ_OP_READ, blkno, hdr, len, blk_crc);
+	return sm_block_io(sb, bdev, READ, blkno, hdr, len, blk_crc);
 }
 
 int scoutfs_block_write_sm(struct super_block *sb,
 			   struct block_device *bdev, u64 blkno,
 			   struct scoutfs_block_header *hdr, size_t len)
 {
-	return sm_block_io(sb, bdev, REQ_OP_WRITE, blkno, hdr, len, NULL);
+	return sm_block_io(sb, bdev, WRITE, blkno, hdr, len, NULL);
 }
 
 int scoutfs_block_setup(struct super_block *sb)
@@ -1262,8 +1254,7 @@ int scoutfs_block_setup(struct super_block *sb)
 	atomic_set(&binf->total_inserted, 0);
 	atomic64_set(&binf->access_counter, 0);
 	init_waitqueue_head(&binf->waitq);
-	KC_INIT_SHRINKER_FUNCS(struct block_info, shrinker,
-			       &binf->shrinker, block_count_objects, block_scan_objects);
+	binf->shrinker.shrink = block_shrink;
 	binf->shrinker.seeks = DEFAULT_SEEKS;
 	register_shrinker(&binf->shrinker);
 	INIT_WORK(&binf->free_work, block_free_work);

kmod/src/counters.h

@@ -30,8 +30,6 @@
 	EXPAND_COUNTER(block_cache_free)			\
 	EXPAND_COUNTER(block_cache_free_work)			\
 	EXPAND_COUNTER(block_cache_remove_stale)		\
-	EXPAND_COUNTER(block_cache_count_objects)		\
-	EXPAND_COUNTER(block_cache_scan_objects)		\
 	EXPAND_COUNTER(block_cache_shrink)			\
 	EXPAND_COUNTER(block_cache_shrink_next)			\
 	EXPAND_COUNTER(block_cache_shrink_recent)		\
@@ -237,12 +235,12 @@ struct scoutfs_counters {
 #define SCOUTFS_PCPU_COUNTER_BATCH (1 << 30)
 
 #define scoutfs_inc_counter(sb, which)					\
-	percpu_counter_add_batch(&SCOUTFS_SB(sb)->counters->which, 1,	\
-				 SCOUTFS_PCPU_COUNTER_BATCH)
+	__percpu_counter_add(&SCOUTFS_SB(sb)->counters->which, 1,	\
+			     SCOUTFS_PCPU_COUNTER_BATCH)
 
 #define scoutfs_add_counter(sb, which, cnt)				\
-	percpu_counter_add_batch(&SCOUTFS_SB(sb)->counters->which, cnt,	\
-				 SCOUTFS_PCPU_COUNTER_BATCH)
+	__percpu_counter_add(&SCOUTFS_SB(sb)->counters->which, cnt,	\
+			     SCOUTFS_PCPU_COUNTER_BATCH)
 
 void __init scoutfs_init_counters(void);
 int scoutfs_setup_counters(struct super_block *sb);

kmod/src/inode.c

@@ -1685,6 +1685,7 @@ static int try_delete_inode_items(struct super_block *sb, u64 ino)
 	struct scoutfs_lock *lock = NULL;
 	struct scoutfs_inode sinode;
 	struct scoutfs_key key;
+	bool clear_trying = false;
 	u64 group_nr;
 	int bit_nr;
 	int ret;
@@ -1704,6 +1705,7 @@ static int try_delete_inode_items(struct super_block *sb, u64 ino)
 		ret = 0;
 		goto out;
 	}
+	clear_trying = true;
 
 	/* can't delete if it's cached in local or remote mounts */
 	if (scoutfs_omap_test(sb, ino) || test_bit_le(bit_nr, ldata->map.bits)) {
@@ -1730,7 +1732,7 @@ static int try_delete_inode_items(struct super_block *sb, u64 ino)
 
 	ret = delete_inode_items(sb, ino, &sinode, lock, orph_lock);
 out:
-	if (ldata)
+	if (clear_trying)
 		clear_bit(bit_nr, ldata->trying);
 
 	scoutfs_unlock(sb, lock, SCOUTFS_LOCK_WRITE);

kmod/src/kernelcompat.c

@@ -1,23 +0,0 @@
-
-#include "kernelcompat.h"
-
-#ifdef KC_SHRINKER_SHRINK
-#include <linux/shrinker.h>
-/*
- * If a target doesn't have that .{count,scan}_objects() interface then
- * we have a .shrink() helper that performs the shrink work in terms of
- * count/scan.
- */
-int kc_shrink_wrapper(struct shrinker *shrink, struct shrink_control *sc)
-{
-	struct kc_shrinker_funcs *funcs = KC_SHRINKER_FUNCS(shrink);
-	unsigned long nr;
-
-	if (sc->nr_to_scan != 0)
-		funcs->scan_objects(shrink, sc);
-
-	nr = funcs->count_objects(shrink, sc);
-
-	return min_t(unsigned long, nr, INT_MAX);
-}
-#endif

kmod/src/kernelcompat.h

@@ -46,81 +46,4 @@ static inline int dir_emit_dots(struct file *file, void *dirent,
 }
 #endif
 
-#ifndef KC_DIR_EMIT_DOTS
-#define percpu_counter_add_batch __percpu_counter_add
-#endif
-
-#ifndef KC_MEMALLOC_NOFS_SAVE
-#define memalloc_nofs_save memalloc_noio_save
-#define memalloc_nofs_restore memalloc_noio_restore
-#endif
-
-#ifdef KC_BIO_BI_OPF
-#define kc_bio_get_opf(bio)		\
-({					\
-	(bio)->bi_opf;			\
-})
-#define kc_bio_set_opf(bio, opf)	\
-do {					\
-	(bio)->bi_opf = opf;		\
-} while (0)
-#define kc_bio_set_sector(bio, sect)	\
-do {					\
-	(bio)->bi_iter.bi_sector = sect;\
-} while (0)
-#else
-#define kc_bio_get_opf(bio)		\
-({					\
-	(bio)->bi_rw;			\
-})
-#define kc_bio_set_opf(bio, opf)	\
-do {					\
-	(bio)->bio_rw = opf;		\
-} while (0)
-#define kc_bio_set_sector(bio, sect)	\
-do {					\
-	(bio)->bi_sector = sect;	\
-} while (0)
-#endif
-
-#ifdef KC_BIO_BI_STATUS
-#define KC_DECLARE_BIO_END_IO(name, bio)	name(bio)
-#define kc_bio_get_errno(bio)			({ blk_status_to_errno((bio)->bi_status); })
-#else
-#define KC_DECLARE_BIO_END_IO(name, bio)	name(bio, int _error_arg)
-#define kc_bio_get_errno(bio)			({ (int)((void)(bio), _error_arg); })
-#endif
-
-#ifndef KC_SHRINKER_SHRINK
-#define KC_DEFINE_SHRINKER(name) struct shrinker name
-#define KC_INIT_SHRINKER_FUNCS(type, name, shrink, count, scan) do {	\
-	__typeof__(shrink) _shrink = (shrink);				\
-	_shrink->count_objects = count;					\
-	_shrink->scan_objects = scan;					\
-} while (0)
-#else
-#include <linux/shrinker.h>
-struct kc_shrinker_funcs {
-	unsigned long (*count_objects)(struct shrinker *, struct shrink_control *sc);
-	unsigned long (*scan_objects)(struct shrinker *, struct shrink_control *sc);
-};
-/* using adjacent member of an unnamed struct */
-#define KC_DEFINE_SHRINKER(name)				\
-	{							\
-		struct kc_shrinker_funcs shrinker_funcs;	\
-		struct shinker name;				\
-	}
-#define KC_SHRINKER_FUNCS(shrinker) \
-	((void *)((long)(shrink) - sizeof(struct kc_shrinker_funcs)))
-#define KC_INIT_SHRINKER_FUNCS(type, name, shrink, count, scan) do {				\
-	BUILD_BUG_ON(offsetof(cont, shrink_funcs) + sizeof(struct kc_shrinker_funcs)) !=	\
-		     offsetof(cont, name) + sizeof(struct kc_shrinker_funcs);			\
-	struct kc_shrinker_funcs *_funcs = KC_SHRINKER_FUNCS(shrink)				\
-	__typeof__(shrink) _shrink = (shrink);							\
-	_funcs->count_objects = count;								\
-	_funcs->scan_objects = scan;								\
-	_shrink->shrink = kc_shrink_wrapper;							\
-} while (0)
-#endif
-
 #endif

kmod/src/server.c

@@ -694,13 +694,13 @@ static int alloc_move_refill_zoned(struct super_block *sb, struct scoutfs_alloc_
 
 static int alloc_move_empty(struct super_block *sb,
 			    struct scoutfs_alloc_root *dst,
-			    struct scoutfs_alloc_root *src, u64 meta_reserved)
+			    struct scoutfs_alloc_root *src, u64 meta_budget)
 {
 	DECLARE_SERVER_INFO(sb, server);
 
 	return scoutfs_alloc_move(sb, &server->alloc, &server->wri,
 				  dst, src, le64_to_cpu(src->total_len), NULL, NULL, 0,
-				  meta_reserved);
+				  meta_budget);
 }
 
 /*
@@ -1226,6 +1226,82 @@ static int finalize_and_start_log_merge(struct super_block *sb, struct scoutfs_l
 	return ret;
 }
 
+/*
+ * The calling get_log_trees ran out of available blocks in its commit's
+ * metadata allocator while moving extents from the log tree's
+ * data_freed into the core data_avail.  This finishes moving the
+ * extents in as many additional commits as it takes.   The logs mutex
+ * is nested inside holding commits so we recheck the persistent item
+ * each time we commit to make sure it's still what we think.   The
+ * caller is still going to send the item to the client so we update the
+ * caller's each time we make progress.  This is a best-effort attempt
+ * to clean up and it's valid to leave extents in data_freed we don't
+ * return errors to the caller.  The client will continue the work later
+ * in get_log_trees or as the rid is reclaimed.
+ */
+static void try_drain_data_freed(struct super_block *sb, struct scoutfs_log_trees *lt)
+{
+	DECLARE_SERVER_INFO(sb, server);
+	struct scoutfs_super_block *super = &SCOUTFS_SB(sb)->super;
+	const u64 rid = le64_to_cpu(lt->rid);
+	const u64 nr = le64_to_cpu(lt->nr);
+	struct scoutfs_log_trees drain;
+	struct scoutfs_key key;
+	COMMIT_HOLD(hold);
+	int ret = 0;
+	int err;
+
+	scoutfs_key_init_log_trees(&key, rid, nr);
+
+	while (lt->data_freed.total_len != 0) {
+		server_hold_commit(sb, &hold);
+		mutex_lock(&server->logs_mutex);
+
+		ret = find_log_trees_item(sb, &super->logs_root, false, rid, U64_MAX, &drain);
+		if (ret < 0)
+			break;
+
+		/* careful to only keep draining the caller's specific open trans */
+		if (drain.nr != lt->nr || drain.get_trans_seq != lt->get_trans_seq ||
+		    drain.commit_trans_seq != lt->commit_trans_seq || drain.flags != lt->flags) {
+			ret = -ENOENT;
+			break;
+		}
+
+		ret = scoutfs_btree_dirty(sb, &server->alloc, &server->wri,
+					  &super->logs_root, &key);
+		if (ret < 0)
+			break;
+
+		/* moving can modify and return errors, always update caller and item */
+		mutex_lock(&server->alloc_mutex);
+		ret = alloc_move_empty(sb, &super->data_alloc, &drain.data_freed,
+				       COMMIT_HOLD_ALLOC_BUDGET / 2);
+		mutex_unlock(&server->alloc_mutex);
+		if (ret == -EINPROGRESS)
+			ret = 0;
+
+		*lt = drain;
+		err = scoutfs_btree_force(sb, &server->alloc, &server->wri,
+					  &super->logs_root, &key, &drain, sizeof(drain));
+		BUG_ON(err < 0); /* dirtying must guarantee success */
+
+		mutex_unlock(&server->logs_mutex);
+
+		ret = server_apply_commit(sb, &hold, ret);
+		if (ret < 0) {
+			ret = 0; /* don't try to abort, ignoring ret */
+			break;
+		}
+	}
+
+	/* try to cleanly abort and write any partial dirty btree blocks, but ignore result */
+	if (ret < 0) {
+		mutex_unlock(&server->logs_mutex);
+		server_apply_commit(sb, &hold, 0);
+	}
+}
+
 /*
  * Give the client roots to all the trees that they'll use to build
  * their transaction.
@@ -1351,7 +1427,9 @@ static int server_get_log_trees(struct super_block *sb,
 		goto update;
 	}
 
-	ret = alloc_move_empty(sb, &super->data_alloc, &lt.data_freed, 0);
+	ret = alloc_move_empty(sb, &super->data_alloc, &lt.data_freed, 100);
+	if (ret == -EINPROGRESS)
+		ret = 0;
 	if (ret < 0) {
 		err_str = "emptying committed data_freed";
 		goto update;
@@ -1429,6 +1507,10 @@ out:
 		scoutfs_err(sb, "error %d getting log trees for rid %016llx: %s",
 			    ret, rid, err_str);
 
+	/* try to drain excessive data_freed with additional commits, if needed, ignoring err */
+	if (ret == 0)
+		try_drain_data_freed(sb, &lt);
+
 	return scoutfs_net_response(sb, conn, cmd, id, ret, &lt, sizeof(lt));
 }
 

tests/Makefile

@@ -10,7 +10,8 @@ BIN := src/createmany			\
 	src/bulk_create_paths		\
 	src/stage_tmpfile		\
 	src/find_xattrs			\
-	src/create_xattr_loop
+	src/create_xattr_loop		\
+	src/fragmented_data_extents
 
 DEPS := $(wildcard src/*.d)
 

tests/golden/large-fragmented-free

@@ -0,0 +1,3 @@
+== creating fragmented extents
+== unlink file with moved extents to free extents per block
+== cleanup

tests/sequence

@@ -9,6 +9,7 @@ fallocate.sh
 setattr_more.sh
 offline-extent-waiting.sh
 move-blocks.sh
+large-fragmented-free.sh
 enospc.sh
 srch-basic-functionality.sh
 simple-xattr-unit.sh

tests/src/fragmented_data_extents.c

@@ -0,0 +1,113 @@
+/*
+ * Copyright (C) 2021 Versity Software, Inc.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License v2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ */
+
+/*
+ * This creates fragmented data extents.
+ *
+ * A file is created that has alternating free and allocated extents.
+ * This also results in the global allocator having the matching
+ * fragmented free extent pattern.  While that file is being created,
+ * occasionally an allocated extent is moved to another file.   This
+ * results in a file that has fragmented extents at a given stride that
+ * can be deleted to create free data extents with a given stride.
+ *
+ * We don't have hole punching so to do this quickly we use a goofy
+ * combination of fallocate, truncate, and our move_blocks ioctl.
+ */
+
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE
+#endif
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+#include <fcntl.h>
+#include <errno.h>
+#include <linux/types.h>
+#include <assert.h>
+
+#include "ioctl.h"
+
+#define BLOCK_SIZE 4096
+
+int main(int argc, char **argv)
+{
+	struct scoutfs_ioctl_move_blocks mb = {0,};
+	unsigned long long freed_extents;
+	unsigned long long move_stride;
+	unsigned long long i;
+	int alloc_fd;
+	int trunc_fd;
+	off_t off;
+	int ret;
+
+	if (argc != 5) {
+		printf("%s <freed_extents> <move_stride> <alloc_file> <trunc_file>\n", argv[0]);
+		return 1;
+	}
+
+	freed_extents = strtoull(argv[1], NULL, 0);
+	move_stride = strtoull(argv[2], NULL, 0);
+
+	alloc_fd = open(argv[3], O_RDWR | O_CREAT | O_TRUNC, S_IRUSR | S_IWUSR);
+	if (alloc_fd == -1) {
+		fprintf(stderr, "error opening %s: %d (%s)\n", argv[3], errno, strerror(errno));
+		exit(1);
+	}
+
+	trunc_fd = open(argv[4], O_RDWR | O_CREAT | O_TRUNC, S_IRUSR | S_IWUSR);
+	if (trunc_fd == -1) {
+		fprintf(stderr, "error opening %s: %d (%s)\n", argv[4], errno, strerror(errno));
+		exit(1);
+	}
+
+	for (i = 0, off = 0; i < freed_extents; i++, off += BLOCK_SIZE * 2) {
+
+		ret = fallocate(alloc_fd, 0, off, BLOCK_SIZE * 2);
+		if (ret < 0) {
+			fprintf(stderr, "fallocate at off %llu error: %d (%s)\n",
+				(unsigned long long)off, errno, strerror(errno));
+			exit(1);
+		}
+
+		ret = ftruncate(alloc_fd, off + BLOCK_SIZE);
+		if (ret < 0) {
+			fprintf(stderr, "truncate to off %llu error: %d (%s)\n",
+				(unsigned long long)off + BLOCK_SIZE, errno, strerror(errno));
+			exit(1);
+		}
+
+		if ((i % move_stride) == 0) {
+			mb.from_fd = alloc_fd;
+			mb.from_off = off;
+			mb.len = BLOCK_SIZE;
+			mb.to_off = i * BLOCK_SIZE;
+
+			ret = ioctl(trunc_fd, SCOUTFS_IOC_MOVE_BLOCKS, &mb);
+			if (ret < 0) {
+				fprintf(stderr, "move from off %llu error: %d (%s)\n",
+					(unsigned long long)off,
+					errno, strerror(errno));
+			}
+		}
+	}
+
+	if (alloc_fd > -1)
+		close(alloc_fd);
+	if (trunc_fd > -1)
+		close(trunc_fd);
+
+	return 0;
+}

tests/tests/large-fragmented-free.sh

@@ -0,0 +1,22 @@
+#
+# Make sure the server can handle a transaction with a data_freed whose
+# blocks all hit different btree blocks in the main free list.  It
+# probably has to be merged in multiple commits.
+#
+
+t_require_commands fragmented_data_extents
+
+EXTENTS_PER_BTREE_BLOCK=600
+EXTENTS_PER_LIST_BLOCK=8192
+FREED_EXTENTS=$((EXTENTS_PER_BTREE_BLOCK * EXTENTS_PER_LIST_BLOCK))
+
+echo "== creating fragmented extents"
+fragmented_data_extents $FREED_EXTENTS $EXTENTS_PER_BTREE_BLOCK "$T_D0/alloc" "$T_D0/move"
+
+echo "== unlink file with moved extents to free extents per block"
+rm -f "$T_D0/move"
+
+echo "== cleanup"
+rm -f "$T_D0/alloc"
+
+t_pass