Use count/scan objects shrinking interface

Move to the more recent interfaces for counting and scanning cached
objects to shrink.

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

ReleaseNotes.md

@@ -1,6 +1,41 @@
 Versity ScoutFS Release Notes
 =============================
 
+---
+v1.6
+\
+*Jul 7, 2022*
+
+* **Fix memory leaks in rare corner cases**
+\
+  Analysis tools found a few corner cases that leaked small structures,
+  generally around error handling or startup and shutdown.
+
+* **Add --skip-likely-huge scoutfs print command option**
+\
+  Add an option to scoutfs print to reduce the size of the output
+  so that it can be used to see system-wide metadata without being
+  overwhelmed by file-level details.
+
+---
+v1.5
+\
+*Jun 21, 2022*
+
+* **Fix persistent error during server startup**
+\
+  Fixed a case where the server would always hit a consistent error on
+  seartup, preventing the system from mounting.  This required a rare
+  but valid state across the clients.
+
+* **Fix a client hang that would lead to fencing**
+\
+  The client module's use of in-kernel networking was missing annotation
+  that could lead to communication hanging.  The server would fence the
+  client when it stopped communicating.  This could be identified by the
+  server fencing a client after it disconnected with no attempt by the
+  client to reconnect.
+
 ---
 v1.4
 \

kmod/src/Makefile

@@ -46,6 +46,10 @@ 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

@@ -34,3 +34,52 @@ 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

@@ -84,6 +84,21 @@ static u64 smallest_order_length(u64 len)
 	return 1ULL << (free_extent_order(len) * 3);
 }
 
+/*
+ * An extent modification dirties three distinct leaves of an allocator
+ * btree as it adds and removes the blkno and size sorted items for the
+ * old and new lengths of the extent.  Dirtying the paths to these
+ * leaves can grow the tree and grow/shrink neighbours at each level.
+ * We over-estimate the number of blocks allocated and freed (the paths
+ * share a root, growth doesn't free) to err on the simpler and safer
+ * side.  The overhead is minimal given the relatively large list blocks
+ * and relatively short allocator trees.
+ */
+static u32 extent_mod_blocks(u32 height)
+{
+	return ((1 + height) * 2) * 3;
+}
+
 /*
  * Free extents don't have flags and are stored in two indexes sorted by
  * block location and by length order, largest first.  The location key
@@ -877,6 +892,14 @@ 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
+ * modified trees to see if they need to continue moving extents.
+ *
  * The caller can specify that extents in the source tree should first
  * be found based on their zone bitmaps.  We'll first try to find
  * extents in the exclusive zones, then vacant zones, and then we'll
@@ -891,7 +914,7 @@ 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)
+		       __le64 *exclusive, __le64 *vacant, u64 zone_blocks, u64 meta_reserved)
 {
 	struct alloc_ext_args args = {
 		.alloc = alloc,
@@ -941,6 +964,14 @@ 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))) {
+			ret = -EINPROGRESS;
+			break;
+		}
+
 		/* searching set start/len, finish initializing alloced extent */
 		ext.map = found.map ? ext.start - found.start + found.map : 0;
 		ext.flags = found.flags;
@@ -1065,15 +1096,6 @@ out:
  * than completely exhausting the avail list or overflowing the freed
  * list.
  *
- * An extent modification dirties three distinct leaves of an allocator
- * btree as it adds and removes the blkno and size sorted items for the
- * old and new lengths of the extent.  Dirtying the paths to these
- * leaves can grow the tree and grow/shrink neighbours at each level.
- * We over-estimate the number of blocks allocated and freed (the paths
- * share a root, growth doesn't free) to err on the simpler and safer
- * side.  The overhead is minimal given the relatively large list blocks
- * and relatively short allocator trees.
- *
  * The caller tells us how many extents they're about to modify and how
  * many other additional blocks they may cow manually.  And finally, the
  * caller could be the first to dirty the avail and freed blocks in the
@@ -1082,7 +1104,7 @@ out:
 static bool list_has_blocks(struct super_block *sb, struct scoutfs_alloc *alloc,
 			    struct scoutfs_alloc_root *root, u32 extents, u32 addl_blocks)
 {
-	u32 tree_blocks = (((1 + root->root.height) * 2) * 3) * extents;
+	u32 tree_blocks = extent_mod_blocks(root->root.height) * extents;
 	u32 most = 1 + tree_blocks + addl_blocks;
 
 	if (le32_to_cpu(alloc->avail.first_nr) < most) {

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);
+		       __le64 *exclusive, __le64 *vacant, u64 zone_blocks, u64 meta_reserved);
 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);

kmod/src/block.c

@@ -21,6 +21,7 @@
 #include <linux/blkdev.h>
 #include <linux/rhashtable.h>
 #include <linux/random.h>
+#include <linux/sched/mm.h>
 
 #include "format.h"
 #include "super.h"
@@ -57,7 +58,7 @@ struct block_info {
 	atomic64_t access_counter;
 	struct rhashtable ht;
 	wait_queue_head_t waitq;
-	struct shrinker shrinker;
+	KC_DEFINE_SHRINKER(shrinker);
 	struct work_struct free_work;
 	struct llist_head free_llist;
 };
@@ -128,7 +129,7 @@ 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 noio_flags;
+	unsigned int nofs_flags;
 
 	/*
 	 * If we had multiple blocks per page we'd need to be a little
@@ -156,9 +157,9 @@ static struct block_private *block_alloc(struct super_block *sb, u64 blkno)
 		 * spurious reclaim-on dependencies and warnings.
 		 */
 		lockdep_off();
-		noio_flags = memalloc_noio_save();
+		nofs_flags = memalloc_nofs_save();
 		bp->virt = __vmalloc(SCOUTFS_BLOCK_LG_SIZE, GFP_NOFS | __GFP_HIGHMEM, PAGE_KERNEL);
-		memalloc_noio_restore(noio_flags);
+		memalloc_nofs_restore(nofs_flags);
 		lockdep_on();
 
 		if (!bp->virt) {
@@ -436,11 +437,10 @@ 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, int rw,
+static void block_end_io(struct super_block *sb, unsigned int opf,
 			 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, int rw,
 	if (!atomic_dec_and_test(&bp->io_count))
 		return;
 
-	if (is_read && !test_bit(BLOCK_BIT_ERROR, &bp->bits))
+	if (!op_is_write(opf) && !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, int rw,
 		wake_up(&binf->waitq);
 }
 
-static void block_bio_end_io(struct bio *bio, int err)
+static void KC_DECLARE_BIO_END_IO(block_bio_end_io, struct bio *bio)
 {
 	struct block_private *bp = bio->bi_private;
 	struct super_block *sb = bp->sb;
 
 	TRACE_BLOCK(end_io, bp);
-	block_end_io(sb, bio->bi_rw, bp, err);
+	block_end_io(sb, kc_bio_get_opf(bio), bp, kc_bio_get_errno(bio));
 	bio_put(bio);
 }
 
@@ -477,7 +477,7 @@ static void block_bio_end_io(struct bio *bio, int err)
  * Kick off IO for a single block.
  */
 static int block_submit_bio(struct super_block *sb, struct block_private *bp,
-			    int rw)
+			    unsigned int opf)
 {
 	struct scoutfs_sb_info *sbi = SCOUTFS_SB(sb);
 	struct bio *bio = NULL;
@@ -510,8 +510,9 @@ static int block_submit_bio(struct super_block *sb, struct block_private *bp,
 				break;
 			}
 
-			bio->bi_sector = sector + (off >> 9);
-			bio->bi_bdev = sbi->meta_bdev;
+			kc_bio_set_opf(bio, opf);
+			kc_bio_set_sector(bio, sector + (off >> 9));
+			bio_set_dev(bio, sbi->meta_bdev);
 			bio->bi_end_io = block_bio_end_io;
 			bio->bi_private = bp;
 
@@ -528,18 +529,18 @@ 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(rw, bio);
+			submit_bio(bio);
 			bio = NULL;
 		}
 	}
 
 	if (bio)
-		submit_bio(rw, bio);
+		submit_bio(bio);
 
 	blk_finish_plug(&plug);
 
 	/* let racing end_io know we're done */
-	block_end_io(sb, rw, bp, ret);
+	block_end_io(sb, opf, bp, ret);
 
 	return ret;
 }
@@ -640,7 +641,7 @@ 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, READ);
+		ret = block_submit_bio(sb, bp, REQ_OP_READ);
 		if (ret < 0)
 			goto out;
 	}
@@ -939,7 +940,7 @@ 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, WRITE);
+		ret = block_submit_bio(sb, bp, REQ_OP_WRITE);
 		if (ret < 0)
 			break;
 	}
@@ -1039,6 +1040,17 @@ 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.
  *
@@ -1059,23 +1071,19 @@ u64 scoutfs_block_writer_dirty_bytes(struct super_block *sb,
  * atomically remove blocks when the only references are ours and the
  * hash table.
  */
-static int block_shrink(struct shrinker *shrink, struct shrink_control *sc)
+static unsigned long block_scan_objects(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;
 
-	nr = sc->nr_to_scan;
-	if (nr == 0)
-		goto out;
+	scoutfs_inc_counter(sb, block_cache_scan_objects);
 
-	scoutfs_inc_counter(sb, block_cache_shrink);
-
-	nr = DIV_ROUND_UP(nr, SCOUTFS_BLOCK_LG_PAGES_PER);
+	nr = DIV_ROUND_UP(sc->nr_to_scan, SCOUTFS_BLOCK_LG_PAGES_PER);
 
 restart:
 	recently = accessed_recently(binf);
@@ -1118,6 +1126,7 @@ 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);
@@ -1126,9 +1135,8 @@ restart:
 
 	rhashtable_walk_stop(&iter);
 	rhashtable_walk_exit(&iter);
-out:
-	return min_t(u64, (u64)atomic_read(&binf->total_inserted) * SCOUTFS_BLOCK_LG_PAGES_PER,
-		     INT_MAX);
+
+	return freed;
 }
 
 struct sm_block_completion {
@@ -1136,11 +1144,11 @@ struct sm_block_completion {
 	int err;
 };
 
-static void sm_block_bio_end_io(struct bio *bio, int err)
+static void KC_DECLARE_BIO_END_IO(sm_block_bio_end_io, struct bio *bio)
 {
 	struct sm_block_completion *sbc = bio->bi_private;
 
-	sbc->err = err;
+	sbc->err = kc_bio_get_errno(bio);
 	complete(&sbc->comp);
 	bio_put(bio);
 }
@@ -1155,9 +1163,8 @@ static void sm_block_bio_end_io(struct bio *bio, int err)
  * 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, int rw, 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, unsigned int opf,
+		       u64 blkno, struct scoutfs_block_header *hdr, size_t len, __le32 *blk_crc)
 {
 	struct scoutfs_block_header *pg_hdr;
 	struct sm_block_completion sbc;
@@ -1171,7 +1178,7 @@ static int sm_block_io(struct super_block *sb, struct block_device *bdev, int rw
 		return -EIO;
 
 	if (WARN_ON_ONCE(len > SCOUTFS_BLOCK_SM_SIZE) ||
-	    WARN_ON_ONCE(!(rw & WRITE) && !blk_crc))
+	    WARN_ON_ONCE(!op_is_write(opf) && !blk_crc))
 		return -EINVAL;
 
 	page = alloc_page(GFP_NOFS);
@@ -1180,7 +1187,7 @@ static int sm_block_io(struct super_block *sb, struct block_device *bdev, int rw
 
 	pg_hdr = page_address(page);
 
-	if (rw & WRITE) {
+	if (op_is_write(opf)) {
 		memcpy(pg_hdr, hdr, len);
 		if (len < SCOUTFS_BLOCK_SM_SIZE)
 			memset((char *)pg_hdr + len, 0,
@@ -1194,8 +1201,9 @@ static int sm_block_io(struct super_block *sb, struct block_device *bdev, int rw
 		goto out;
 	}
 
-	bio->bi_sector = blkno << (SCOUTFS_BLOCK_SM_SHIFT - 9);
-	bio->bi_bdev = bdev;
+	bio->bi_opf = opf | REQ_SYNC;
+	kc_bio_set_sector(bio, blkno << (SCOUTFS_BLOCK_SM_SHIFT - 9));
+	bio_set_dev(bio, bdev);
 	bio->bi_end_io = sm_block_bio_end_io;
 	bio->bi_private = &sbc;
 	bio_add_page(bio, page, SCOUTFS_BLOCK_SM_SIZE, 0);
@@ -1203,12 +1211,12 @@ static int sm_block_io(struct super_block *sb, struct block_device *bdev, int rw
 	init_completion(&sbc.comp);
 	sbc.err = 0;
 
-	submit_bio((rw & WRITE) ? WRITE_SYNC : READ_SYNC, bio);
+	submit_bio(bio);
 
 	wait_for_completion(&sbc.comp);
 	ret = sbc.err;
 
-	if (ret == 0 && !(rw & WRITE)) {
+	if (ret == 0 && !op_is_write(opf)) {
 		memcpy(hdr, pg_hdr, len);
 		*blk_crc = block_calc_crc(pg_hdr, SCOUTFS_BLOCK_SM_SIZE);
 	}
@@ -1222,14 +1230,14 @@ 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, READ, blkno, hdr, len, blk_crc);
+	return sm_block_io(sb, bdev, REQ_OP_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, WRITE, blkno, hdr, len, NULL);
+	return sm_block_io(sb, bdev, REQ_OP_WRITE, blkno, hdr, len, NULL);
 }
 
 int scoutfs_block_setup(struct super_block *sb)
@@ -1254,7 +1262,8 @@ 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);
-	binf->shrinker.shrink = block_shrink;
+	KC_INIT_SHRINKER_FUNCS(struct block_info, shrinker,
+			       &binf->shrinker, block_count_objects, block_scan_objects);
 	binf->shrinker.seeks = DEFAULT_SEEKS;
 	register_shrinker(&binf->shrinker);
 	INIT_WORK(&binf->free_work, block_free_work);

kmod/src/counters.h

@@ -30,6 +30,8 @@
 	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)		\
@@ -235,12 +237,12 @@ struct scoutfs_counters {
 #define SCOUTFS_PCPU_COUNTER_BATCH (1 << 30)
 
 #define scoutfs_inc_counter(sb, which)					\
-	__percpu_counter_add(&SCOUTFS_SB(sb)->counters->which, 1,	\
-			     SCOUTFS_PCPU_COUNTER_BATCH)
+	percpu_counter_add_batch(&SCOUTFS_SB(sb)->counters->which, 1,	\
+				 SCOUTFS_PCPU_COUNTER_BATCH)
 
 #define scoutfs_add_counter(sb, which, cnt)				\
-	__percpu_counter_add(&SCOUTFS_SB(sb)->counters->which, cnt,	\
-			     SCOUTFS_PCPU_COUNTER_BATCH)
+	percpu_counter_add_batch(&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/kernelcompat.c

@@ -0,0 +1,23 @@
+
+#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

@@ -46,4 +46,81 @@ 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/net.c

@@ -355,6 +355,7 @@ static int submit_send(struct super_block *sb,
 		}
 		if (rid != 0) {
 			spin_unlock(&conn->lock);
+			kfree(msend);
 			return -ENOTCONN;
 		}
 	}
@@ -991,6 +992,8 @@ static void scoutfs_net_listen_worker(struct work_struct *work)
 		if (ret < 0)
 			break;
 
+		acc_sock->sk->sk_allocation = GFP_NOFS;
+
 		/* inherit accepted request funcs from listening conn */
 		acc_conn = scoutfs_net_alloc_conn(sb, conn->notify_up,
 						  conn->notify_down,
@@ -1053,6 +1056,8 @@ static void scoutfs_net_connect_worker(struct work_struct *work)
 	if (ret)
 		goto out;
 
+	sock->sk->sk_allocation = GFP_NOFS;
+
 	/* caller specified connect timeout */
 	tv.tv_sec = conn->connect_timeout_ms / MSEC_PER_SEC;
 	tv.tv_usec = (conn->connect_timeout_ms % MSEC_PER_SEC) * USEC_PER_MSEC;
@@ -1341,10 +1346,12 @@ scoutfs_net_alloc_conn(struct super_block *sb,
 	if (!conn)
 		return NULL;
 
-	conn->info = kzalloc(info_size, GFP_NOFS);
-	if (!conn->info) {
-		kfree(conn);
-		return NULL;
+	if (info_size) {
+		conn->info = kzalloc(info_size, GFP_NOFS);
+		if (!conn->info) {
+			kfree(conn);
+			return NULL;
+		}
 	}
 
 	conn->workq = alloc_workqueue("scoutfs_net_%s",
@@ -1450,6 +1457,8 @@ int scoutfs_net_bind(struct super_block *sb,
 	if (ret)
 		goto out;
 
+	sock->sk->sk_allocation = GFP_NOFS;
+
 	optval = 1;
 	ret = kernel_setsockopt(sock, SOL_SOCKET, SO_REUSEADDR,
 				(char *)&optval, sizeof(optval));

kmod/src/omap.c

@@ -157,6 +157,15 @@ static int free_rid(struct omap_rid_list *list, struct omap_rid_entry *entry)
 	return nr;
 }
 
+static void free_rid_list(struct omap_rid_list *list)
+{
+	struct omap_rid_entry *entry;
+	struct omap_rid_entry *tmp;
+
+	list_for_each_entry_safe(entry, tmp, &list->head, head)
+		free_rid(list, entry);
+}
+
 static int copy_rids(struct omap_rid_list *to, struct omap_rid_list *from, spinlock_t *from_lock)
 {
 	struct omap_rid_entry *entry;
@@ -804,6 +813,10 @@ void scoutfs_omap_server_shutdown(struct super_block *sb)
 	llist_for_each_entry_safe(req, tmp, requests, llnode)
 		kfree(req);
 
+	spin_lock(&ominf->lock);
+	free_rid_list(&ominf->rids);
+	spin_unlock(&ominf->lock);
+
 	synchronize_rcu();
 }
 
@@ -864,6 +877,10 @@ void scoutfs_omap_destroy(struct super_block *sb)
 		rhashtable_walk_stop(&iter);
 		rhashtable_walk_exit(&iter);
 
+		spin_lock(&ominf->lock);
+		free_rid_list(&ominf->rids);
+		spin_unlock(&ominf->lock);
+
 		rhashtable_destroy(&ominf->group_ht);
 		rhashtable_destroy(&ominf->req_ht);
 		kfree(ominf);

kmod/src/server.c

@@ -689,23 +689,18 @@ static int alloc_move_refill_zoned(struct super_block *sb, struct scoutfs_alloc_
 	return scoutfs_alloc_move(sb, &server->alloc, &server->wri, dst, src,
 				  min(target - le64_to_cpu(dst->total_len),
 				      le64_to_cpu(src->total_len)),
-				  exclusive, vacant, zone_blocks);
-}
-
-static inline int alloc_move_refill(struct super_block *sb, struct scoutfs_alloc_root *dst,
-				    struct scoutfs_alloc_root *src, u64 lo, u64 target)
-{
-	return alloc_move_refill_zoned(sb, dst, src, lo, target, NULL, NULL, 0);
+				  exclusive, vacant, zone_blocks, 0);
 }
 
 static int alloc_move_empty(struct super_block *sb,
 			    struct scoutfs_alloc_root *dst,
-			    struct scoutfs_alloc_root *src)
+			    struct scoutfs_alloc_root *src, u64 meta_reserved)
 {
 	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);
+				  dst, src, le64_to_cpu(src->total_len), NULL, NULL, 0,
+				  meta_reserved);
 }
 
 /*
@@ -1272,6 +1267,7 @@ static int server_get_log_trees(struct super_block *sb,
 	char *err_str = NULL;
 	u64 nr;
 	int ret;
+	int err;
 
 	if (arg_len != 0) {
 		ret = -EINVAL;
@@ -1315,16 +1311,27 @@ static int server_get_log_trees(struct super_block *sb,
 		goto unlock;
 	}
 
+	if (ret != -ENOENT) {
+		/* need to sync lt with respect to changes in other structures */
+		scoutfs_key_init_log_trees(&key, le64_to_cpu(lt.rid), le64_to_cpu(lt.nr));
+		ret = scoutfs_btree_dirty(sb, &server->alloc, &server->wri,
+					  &super->logs_root, &key);
+		if (ret < 0) {
+			err_str = "dirtying lt btree key";
+			goto unlock;
+		}
+	}
+
 	/* drops and re-acquires the mutex and commit if it has to wait */
 	ret = finalize_and_start_log_merge(sb, &lt, rid, &hold);
 	if (ret < 0)
-		goto unlock;
+		goto update;
 
 	if (get_volopt_val(server, SCOUTFS_VOLOPT_DATA_ALLOC_ZONE_BLOCKS_NR, &data_zone_blocks)) {
 		ret = get_data_alloc_zone_bits(sb, rid, exclusive, vacant, data_zone_blocks);
 		if (ret < 0) {
 			err_str = "getting alloc zone bits";
-			goto unlock;
+			goto update;
 		}
 	} else {
 		data_zone_blocks = 0;
@@ -1341,13 +1348,13 @@ static int server_get_log_trees(struct super_block *sb,
 					&lt.meta_freed);
 	if (ret < 0) {
 		err_str = "splicing committed meta_freed";
-		goto unlock;
+		goto update;
 	}
 
-	ret = alloc_move_empty(sb, &super->data_alloc, &lt.data_freed);
+	ret = alloc_move_empty(sb, &super->data_alloc, &lt.data_freed, 0);
 	if (ret < 0) {
 		err_str = "emptying committed data_freed";
-		goto unlock;
+		goto update;
 	}
 
 	ret = scoutfs_alloc_fill_list(sb, &server->alloc, &server->wri,
@@ -1356,7 +1363,7 @@ static int server_get_log_trees(struct super_block *sb,
 				      SCOUTFS_SERVER_META_FILL_TARGET);
 	if (ret < 0) {
 		err_str = "filling meta_avail";
-		goto unlock;
+		goto update;
 	}
 
 	if (le64_to_cpu(server->meta_avail->total_len) <= scoutfs_server_reserved_meta_blocks(sb))
@@ -1369,7 +1376,7 @@ static int server_get_log_trees(struct super_block *sb,
 				      exclusive, vacant, data_zone_blocks);
 	if (ret < 0) {
 		err_str = "refilling data_avail";
-		goto unlock;
+		goto update;
 	}
 
 	if (le64_to_cpu(lt.data_avail.total_len) < SCOUTFS_SERVER_DATA_FILL_LO)
@@ -1389,7 +1396,7 @@ static int server_get_log_trees(struct super_block *sb,
 		if (ret < 0) {
 			zero_data_alloc_zone_bits(&lt);
 			err_str = "setting data_avail zone bits";
-			goto unlock;
+			goto update;
 		}
 
 		lt.data_alloc_zone_blocks = cpu_to_le64(data_zone_blocks);
@@ -1398,13 +1405,18 @@ static int server_get_log_trees(struct super_block *sb,
 	/* give the transaction a new seq (must have been ==) */
 	lt.get_trans_seq = cpu_to_le64(scoutfs_server_next_seq(sb));
 
+update:
 	/* update client's log tree's item */
-	scoutfs_key_init_log_trees(&key, le64_to_cpu(lt.rid),
-				   le64_to_cpu(lt.nr));
-	ret = scoutfs_btree_force(sb, &server->alloc, &server->wri,
+	scoutfs_key_init_log_trees(&key, le64_to_cpu(lt.rid), le64_to_cpu(lt.nr));
+	err = scoutfs_btree_force(sb, &server->alloc, &server->wri,
 				  &super->logs_root, &key, &lt, sizeof(lt));
-	if (ret < 0)
-		err_str = "updating log trees";
+	BUG_ON(err < 0); /* can duplicate extents.. move dst in super, still in in lt src */
+	if (err < 0) {
+		if (ret == 0) {
+			ret = err;
+			err_str = "updating log trees";
+		}
+	}
 
 unlock:
 	if (unlock_alloc)
@@ -1544,9 +1556,11 @@ static int server_get_roots(struct super_block *sb,
  * read and we finalize the tree so that it will be merged.  We reclaim
  * all the allocator items.
  *
- * The caller holds the commit rwsem which means we do all this work in
- * one server commit.  We'll need to keep the total amount of blocks in
- * trees in check.
+ * The caller holds the commit rwsem which means we have to do our work
+ * in one commit.  The alocator btrees can be very large and very
+ * fragmented.  We return -EINPROGRESS if we couldn't fully reclaim the
+ * allocators in one commit.   The caller should apply the current
+ * commit and call again in a new commit.
  *
  * By the time we're evicting a client they've either synced their data
  * or have been forcefully removed.  The free blocks in the allocator
@@ -1606,9 +1620,9 @@ static int reclaim_open_log_tree(struct super_block *sb, u64 rid)
 	}
 
 	/*
-	 * All of these can return errors after having modified the
-	 * allocator trees.  We have to try and update the roots in the
-	 * log item.
+	 * All of these can return errors, perhaps indicating successful
+	 * partial progress, after having modified the allocator trees.
+	 * We always have to update the roots in the log item.
 	 */
 	mutex_lock(&server->alloc_mutex);
 	ret = (err_str = "splice meta_freed to other_freed",
@@ -1618,18 +1632,21 @@ static int reclaim_open_log_tree(struct super_block *sb, u64 rid)
 	       scoutfs_alloc_splice_list(sb, &server->alloc, &server->wri, server->other_freed,
 					 &lt.meta_avail)) ?:
 	      (err_str = "empty data_avail",
-	       alloc_move_empty(sb, &super->data_alloc, &lt.data_avail)) ?:
+	       alloc_move_empty(sb, &super->data_alloc, &lt.data_avail, 100)) ?:
 	      (err_str = "empty data_freed",
-	       alloc_move_empty(sb, &super->data_alloc, &lt.data_freed));
+	       alloc_move_empty(sb, &super->data_alloc, &lt.data_freed, 100));
 	mutex_unlock(&server->alloc_mutex);
 
-	/* the transaction is no longer open */
-	lt.commit_trans_seq = lt.get_trans_seq;
+	/* only finalize, allowing merging, once the allocators are fully freed */
+	if (ret == 0) {
+		/* the transaction is no longer open */
+		lt.commit_trans_seq = lt.get_trans_seq;
 
-	/* the mount is no longer writing to the zones */
-	zero_data_alloc_zone_bits(&lt);
-	le64_add_cpu(&lt.flags, SCOUTFS_LOG_TREES_FINALIZED);
-	lt.finalize_seq = cpu_to_le64(scoutfs_server_next_seq(sb));
+		/* the mount is no longer writing to the zones */
+		zero_data_alloc_zone_bits(&lt);
+		le64_add_cpu(&lt.flags, SCOUTFS_LOG_TREES_FINALIZED);
+		lt.finalize_seq = cpu_to_le64(scoutfs_server_next_seq(sb));
+	}
 
 	err = scoutfs_btree_update(sb, &server->alloc, &server->wri,
 				  &super->logs_root, &key, &lt, sizeof(lt));
@@ -1638,7 +1655,7 @@ static int reclaim_open_log_tree(struct super_block *sb, u64 rid)
 out:
 	mutex_unlock(&server->logs_mutex);
 
-	if (ret < 0)
+	if (ret < 0 && ret != -EINPROGRESS)
 		scoutfs_err(sb, "server error %d reclaiming log trees for rid %016llx: %s",
 			    ret, rid, err_str);
 
@@ -3536,26 +3553,37 @@ struct farewell_request {
  * Reclaim all the resources for a mount which has gone away.  It's sent
  * us a farewell promising to leave or we actively fenced it.
  *
- * It's safe to call this multiple times for a given rid.  Each
- * individual action knows to recognize that it's already been performed
- * and return success.
+ * This can be called multiple times across different servers for
+ * different reclaim attempts.  The existence of the mounted_client item
+ * triggers reclaim and must be deleted last.  Each step knows that it
+ * can be called multiple times and safely recognizes that its work
+ * might have already been done.
+ *
+ * Some steps (reclaiming large fragmented allocators) may need multiple
+ * calls to complete.  They return -EINPROGRESS which tells us to apply
+ * the server commit and retry.
  */
 static int reclaim_rid(struct super_block *sb, u64 rid)
 {
 	COMMIT_HOLD(hold);
 	int ret;
+	int err;
 
-	server_hold_commit(sb, &hold);
+	do {
+		server_hold_commit(sb, &hold);
 
-	/* delete mounted client last, recovery looks for it */
-	ret = scoutfs_lock_server_farewell(sb, rid) ?:
-	      reclaim_open_log_tree(sb, rid) ?:
-	      cancel_srch_compact(sb, rid) ?:
-	      cancel_log_merge(sb, rid) ?:
-	      scoutfs_omap_remove_rid(sb, rid) ?:
-	      delete_mounted_client(sb, rid);
+		err = scoutfs_lock_server_farewell(sb, rid) ?:
+		      reclaim_open_log_tree(sb, rid) ?:
+		      cancel_srch_compact(sb, rid) ?:
+		      cancel_log_merge(sb, rid) ?:
+		      scoutfs_omap_remove_rid(sb, rid) ?:
+		      delete_mounted_client(sb, rid);
 
-	return server_apply_commit(sb, &hold, ret);
+		ret = server_apply_commit(sb, &hold, err == -EINPROGRESS ? 0 : err);
+
+	} while (err == -EINPROGRESS && ret == 0);
+
+	return ret;
 }
 
 /*

kmod/src/super.c

@@ -496,7 +496,7 @@ static int scoutfs_fill_super(struct super_block *sb, void *data, int silent)
 
 	ret = assign_random_id(sbi);
 	if (ret < 0)
-		return ret;
+		goto out;
 
 	spin_lock_init(&sbi->next_ino_lock);
 	spin_lock_init(&sbi->data_wait_root.lock);
@@ -505,7 +505,7 @@ static int scoutfs_fill_super(struct super_block *sb, void *data, int silent)
 	/* parse options early for use during setup */
 	ret = scoutfs_options_early_setup(sb, data);
 	if (ret < 0)
-		return ret;
+		goto out;
 	scoutfs_options_read(sb, &opts);
 
 	ret = sb_set_blocksize(sb, SCOUTFS_BLOCK_SM_SIZE);

utils/man/scoutfs.8

@@ -597,7 +597,7 @@ format.
 .PD
 
 .TP
-.BI "print META-DEVICE"
+.BI "print {-S|--skip-likely-huge} META-DEVICE"
 .sp
 Prints out all of the metadata in the file system.  This makes no effort
 to ensure that the structures are consistent as they're traversed and
@@ -607,6 +607,20 @@ output.
 .PD 0
 .TP
 .sp
+.B "-S, --skip-likely-huge"
+Skip printing structures that are likely to be very large.  The
+structures that are skipped tend to be global and whose size tends to be
+related to the size of the volume.   Examples of skipped structures include
+the global fs items, srch files, and metadata and data
+allocators.  Similar structures that are not skipped are related to the
+number of mounts and are maintained at a relatively reasonable size.
+These include per-mount log trees, srch files, allocators, and the
+metadata allocators used by server commits.
+.sp
+Skipping the larger structures limits the print output to a relatively
+constant size rather than being a large multiple of the used metadata
+space of the volume making the output much more useful for inspection.
+.TP
 .B "META-DEVICE"
 The path to the metadata device for the filesystem whose metadata will be
 printed.  Since this command reads via the host's buffer cache, it may not

utils/src/print.c

@@ -8,6 +8,7 @@
 #include <errno.h>
 #include <string.h>
 #include <stdarg.h>
+#include <stdbool.h>
 #include <ctype.h>
 #include <uuid/uuid.h>
 #include <sys/socket.h>
@@ -989,9 +990,10 @@ static void print_super_block(struct scoutfs_super_block *super, u64 blkno)
 
 struct print_args {
 	char *meta_device;
+	bool skip_likely_huge;
 };
 
-static int print_volume(int fd)
+static int print_volume(int fd, struct print_args *args)
 {
 	struct scoutfs_super_block *super = NULL;
 	struct print_recursion_args pa;
@@ -1041,23 +1043,26 @@ static int print_volume(int fd)
 			ret = err;
 	}
 
-	for (i = 0; i < array_size(super->meta_alloc); i++) {
-		snprintf(str, sizeof(str), "meta_alloc[%u]", i);
-		err = print_btree(fd, super, str, &super->meta_alloc[i].root,
+	if (!args->skip_likely_huge) {
+		for (i = 0; i < array_size(super->meta_alloc); i++) {
+			snprintf(str, sizeof(str), "meta_alloc[%u]", i);
+			err = print_btree(fd, super, str, &super->meta_alloc[i].root,
+					  print_alloc_item, NULL);
+			if (err && !ret)
+				ret = err;
+		}
+
+		err = print_btree(fd, super, "data_alloc", &super->data_alloc.root,
 				  print_alloc_item, NULL);
 		if (err && !ret)
 			ret = err;
 	}
 
-	err = print_btree(fd, super, "data_alloc", &super->data_alloc.root,
-			  print_alloc_item, NULL);
-	if (err && !ret)
-		ret = err;
-
 	err = print_btree(fd, super, "srch_root", &super->srch_root,
 			  print_srch_root_item, NULL);
 	if (err && !ret)
 		ret = err;
+
 	err = print_btree(fd, super, "logs_root", &super->logs_root,
 			  print_log_trees_item, NULL);
 	if (err && !ret)
@@ -1065,19 +1070,23 @@ static int print_volume(int fd)
 
 	pa.super = super;
 	pa.fd = fd;
-	err = print_btree_leaf_items(fd, super, &super->srch_root.ref,
-				     print_srch_root_files, &pa);
-	if (err && !ret)
-		ret = err;
+	if (!args->skip_likely_huge) {
+		err = print_btree_leaf_items(fd, super, &super->srch_root.ref,
+					     print_srch_root_files, &pa);
+		if (err && !ret)
+			ret = err;
+	}
 	err = print_btree_leaf_items(fd, super, &super->logs_root.ref,
 				     print_log_trees_roots, &pa);
 	if (err && !ret)
 		ret = err;
 
-	err = print_btree(fd, super, "fs_root", &super->fs_root,
-			  print_fs_item, NULL);
-	if (err && !ret)
-		ret = err;
+	if (!args->skip_likely_huge) {
+		err = print_btree(fd, super, "fs_root", &super->fs_root,
+				  print_fs_item, NULL);
+		if (err && !ret)
+			ret = err;
+	}
 
 out:
 	free(super);
@@ -1098,7 +1107,7 @@ static int do_print(struct print_args *args)
 		return ret;
 	}
 
-	ret = print_volume(fd);
+	ret = print_volume(fd, args);
 	close(fd);
 	return ret;
 };
@@ -1108,6 +1117,9 @@ static int parse_opt(int key, char *arg, struct argp_state *state)
 	struct print_args *args = state->input;
 
 	switch (key) {
+	case 'S':
+		args->skip_likely_huge = true;
+		break;
 	case ARGP_KEY_ARG:
 		if (!args->meta_device)
 			args->meta_device = strdup_or_error(state, arg);
@@ -1125,8 +1137,13 @@ static int parse_opt(int key, char *arg, struct argp_state *state)
 	return 0;
 }
 
+static struct argp_option options[] = {
+	{ "skip-likely-huge", 'S', NULL, 0, "Skip large structures to minimize output size"},
+	{ NULL }
+};
+
 static struct argp argp = {
-	NULL,
+	options,
 	parse_opt,
 	"META-DEV",
 	"Print metadata structures"