summaryrefslogtreecommitdiff
path: root/fs/bcachefs
diff options
context:
space:
mode:
Diffstat (limited to 'fs/bcachefs')
-rw-r--r--fs/bcachefs/Kconfig1
-rw-r--r--fs/bcachefs/Makefile3
-rw-r--r--fs/bcachefs/bcachefs.h2
-rw-r--r--fs/bcachefs/btree_cache.c16
-rw-r--r--fs/bcachefs/btree_key_cache.c14
-rw-r--r--fs/bcachefs/closure.h5
-rw-r--r--fs/bcachefs/data_update.c4
-rw-r--r--fs/bcachefs/errcode.h5
-rw-r--r--fs/bcachefs/fs.c14
-rw-r--r--fs/bcachefs/io_read.c13
-rw-r--r--fs/bcachefs/movinggc.c44
-rw-r--r--fs/bcachefs/nocow_locking.c3
-rw-r--r--fs/bcachefs/sysfs.c117
-rw-r--r--fs/bcachefs/util.h2
-rw-r--r--fs/bcachefs/vendor/closure.c218
-rw-r--r--fs/bcachefs/vendor/closure.h490
16 files changed, 922 insertions, 29 deletions
diff --git a/fs/bcachefs/Kconfig b/fs/bcachefs/Kconfig
index 8b846c09350b..5455412b2b75 100644
--- a/fs/bcachefs/Kconfig
+++ b/fs/bcachefs/Kconfig
@@ -3,7 +3,6 @@ config BCACHEFS_FS
tristate "bcachefs filesystem support (EXPERIMENTAL)"
depends on BLOCK
select EXPORTFS
- select CLOSURES
select CRC32
select CRC64
select FS_POSIX_ACL
diff --git a/fs/bcachefs/Makefile b/fs/bcachefs/Makefile
index a4258615dffa..1e87eee962ec 100644
--- a/fs/bcachefs/Makefile
+++ b/fs/bcachefs/Makefile
@@ -98,7 +98,8 @@ bcachefs-y := \
two_state_shared_lock.o \
util.o \
varint.o \
- xattr.o
+ xattr.o \
+ vendor/closure.o
obj-$(CONFIG_MEAN_AND_VARIANCE_UNIT_TEST) += mean_and_variance_test.o
diff --git a/fs/bcachefs/bcachefs.h b/fs/bcachefs/bcachefs.h
index 83d6ab9c1a91..3ccca855f05e 100644
--- a/fs/bcachefs/bcachefs.h
+++ b/fs/bcachefs/bcachefs.h
@@ -196,7 +196,6 @@
#include <linux/backing-dev-defs.h>
#include <linux/bug.h>
#include <linux/bio.h>
-#include <linux/closure.h>
#include <linux/kobject.h>
#include <linux/list.h>
#include <linux/math64.h>
@@ -217,6 +216,7 @@
#include "bcachefs_format.h"
#include "btree_journal_iter_types.h"
+#include "closure.h"
#include "disk_accounting_types.h"
#include "errcode.h"
#include "fast_list.h"
diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c
index 59638d09e1fd..3b1d694dcb3a 100644
--- a/fs/bcachefs/btree_cache.c
+++ b/fs/bcachefs/btree_cache.c
@@ -15,6 +15,7 @@
#include <linux/prefetch.h>
#include <linux/sched/mm.h>
+#include <linux/seq_buf.h>
#include <linux/swap.h>
const char * const bch2_btree_node_flags[] = {
@@ -565,6 +566,19 @@ static unsigned long bch2_btree_cache_count(struct shrinker *shrink,
return btree_cache_can_free(list);
}
+static void bch2_btree_cache_shrinker_to_text(struct seq_buf *s, struct shrinker *shrink)
+{
+ struct btree_cache_list *list = shrink->private_data;
+ struct btree_cache *bc = container_of(list, struct btree_cache, live[list->idx]);
+
+ char *cbuf;
+ size_t buflen = seq_buf_get_buf(s, &cbuf);
+ struct printbuf out = PRINTBUF_EXTERN(cbuf, buflen);
+
+ bch2_btree_cache_to_text(&out, bc);
+ seq_buf_commit(s, out.pos);
+}
+
void bch2_fs_btree_cache_exit(struct bch_fs *c)
{
struct btree_cache *bc = &c->btree_cache;
@@ -659,6 +673,7 @@ int bch2_fs_btree_cache_init(struct bch_fs *c)
bc->live[0].shrink = shrink;
shrink->count_objects = bch2_btree_cache_count;
shrink->scan_objects = bch2_btree_cache_scan;
+ shrink->to_text = bch2_btree_cache_shrinker_to_text;
shrink->seeks = 2;
shrink->private_data = &bc->live[0];
shrinker_register(shrink);
@@ -669,6 +684,7 @@ int bch2_fs_btree_cache_init(struct bch_fs *c)
bc->live[1].shrink = shrink;
shrink->count_objects = bch2_btree_cache_count;
shrink->scan_objects = bch2_btree_cache_scan;
+ shrink->to_text = bch2_btree_cache_shrinker_to_text;
shrink->seeks = 8;
shrink->private_data = &bc->live[1];
shrinker_register(shrink);
diff --git a/fs/bcachefs/btree_key_cache.c b/fs/bcachefs/btree_key_cache.c
index 4890cbc88e7c..e3336ab27ccc 100644
--- a/fs/bcachefs/btree_key_cache.c
+++ b/fs/bcachefs/btree_key_cache.c
@@ -13,6 +13,7 @@
#include "trace.h"
#include <linux/sched/mm.h>
+#include <linux/seq_buf.h>
static inline bool btree_uses_pcpu_readers(enum btree_id id)
{
@@ -808,6 +809,18 @@ void bch2_fs_btree_key_cache_init_early(struct btree_key_cache *c)
{
}
+static void bch2_btree_key_cache_shrinker_to_text(struct seq_buf *s, struct shrinker *shrink)
+{
+ struct bch_fs *c = shrink->private_data;
+ struct btree_key_cache *bc = &c->btree_key_cache;
+ char *cbuf;
+ size_t buflen = seq_buf_get_buf(s, &cbuf);
+ struct printbuf out = PRINTBUF_EXTERN(cbuf, buflen);
+
+ bch2_btree_key_cache_to_text(&out, bc);
+ seq_buf_commit(s, out.pos);
+}
+
int bch2_fs_btree_key_cache_init(struct btree_key_cache *bc)
{
struct bch_fs *c = container_of(bc, struct bch_fs, btree_key_cache);
@@ -832,6 +845,7 @@ int bch2_fs_btree_key_cache_init(struct btree_key_cache *bc)
bc->shrink = shrink;
shrink->count_objects = bch2_btree_key_cache_count;
shrink->scan_objects = bch2_btree_key_cache_scan;
+ shrink->to_text = bch2_btree_key_cache_shrinker_to_text;
shrink->batch = 1 << 14;
shrink->seeks = 0;
shrink->private_data = c;
diff --git a/fs/bcachefs/closure.h b/fs/bcachefs/closure.h
new file mode 100644
index 000000000000..d8d4c7093ce0
--- /dev/null
+++ b/fs/bcachefs/closure.h
@@ -0,0 +1,5 @@
+#include "vendor/closure.h"
+
+#define closure_wait bch2_closure_wait
+#define closure_return_sync bch2_closure_return_sync
+#define __closure_wake_up __bch2_closure_wake_up
diff --git a/fs/bcachefs/data_update.c b/fs/bcachefs/data_update.c
index 155c1ad42fc1..62d5d17d681e 100644
--- a/fs/bcachefs/data_update.c
+++ b/fs/bcachefs/data_update.c
@@ -91,8 +91,10 @@ bool __bkey_nocow_lock(struct bch_fs *c, struct moving_context *ctxt, struct bke
move_ctxt_wait_event(ctxt,
(locked = bch2_bucket_nocow_trylock(&c->nocow_locks, bucket, 0)) ||
list_empty(&ctxt->ios));
- if (!locked)
+ if (!locked) {
+ bch2_trans_unlock(ctxt->trans);
bch2_bucket_nocow_lock(&c->nocow_locks, bucket, 0);
+ }
}
return true;
}
diff --git a/fs/bcachefs/errcode.h b/fs/bcachefs/errcode.h
index c4344a1d6976..cbf1eedddad7 100644
--- a/fs/bcachefs/errcode.h
+++ b/fs/bcachefs/errcode.h
@@ -188,6 +188,11 @@
x(BCH_ERR_recovery_will_run, recovery_pass_will_run) \
x(0, data_update_done) \
x(0, bkey_was_deleted) \
+ x(0, bucket_not_moveable) \
+ x(BCH_ERR_bucket_not_moveable, bucket_not_moveable_dev_not_rw) \
+ x(BCH_ERR_bucket_not_moveable, bucket_not_moveable_bucket_open) \
+ x(BCH_ERR_bucket_not_moveable, bucket_not_moveable_bp_mismatch) \
+ x(BCH_ERR_bucket_not_moveable, bucket_not_moveable_lru_race) \
x(BCH_ERR_data_update_done, data_update_done_would_block) \
x(BCH_ERR_data_update_done, data_update_done_unwritten) \
x(BCH_ERR_data_update_done, data_update_done_no_writes_needed) \
diff --git a/fs/bcachefs/fs.c b/fs/bcachefs/fs.c
index 958849c30071..f1849eb8327d 100644
--- a/fs/bcachefs/fs.c
+++ b/fs/bcachefs/fs.c
@@ -1521,6 +1521,7 @@ static const struct vm_operations_struct bch_vm_ops = {
.page_mkwrite = bch2_page_mkwrite,
};
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(6,17,0)
static int bch2_mmap_prepare(struct vm_area_desc *desc)
{
file_accessed(desc->file);
@@ -1528,6 +1529,15 @@ static int bch2_mmap_prepare(struct vm_area_desc *desc)
desc->vm_ops = &bch_vm_ops;
return 0;
}
+#else
+static int bch2_mmap(struct file *file, struct vm_area_struct *vma)
+{
+ file_accessed(file);
+
+ vma->vm_ops = &bch_vm_ops;
+ return 0;
+}
+#endif
/* Directories: */
@@ -1719,7 +1729,11 @@ static const struct file_operations bch_file_operations = {
.llseek = bch2_llseek,
.read_iter = bch2_read_iter,
.write_iter = bch2_write_iter,
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(6,17,0)
.mmap_prepare = bch2_mmap_prepare,
+#else
+ .mmap = bch2_mmap,
+#endif
.get_unmapped_area = thp_get_unmapped_area,
.fsync = bch2_fsync,
.splice_read = filemap_splice_read,
diff --git a/fs/bcachefs/io_read.c b/fs/bcachefs/io_read.c
index ca480b8f8dae..ac545f962ce9 100644
--- a/fs/bcachefs/io_read.c
+++ b/fs/bcachefs/io_read.c
@@ -42,12 +42,6 @@ module_param_named(read_corrupt_device, bch2_read_corrupt_device, int, 0644);
MODULE_PARM_DESC(read_corrupt_device, "");
#endif
-static bool bch2_poison_extents_on_checksum_error;
-module_param_named(poison_extents_on_checksum_error,
- bch2_poison_extents_on_checksum_error, bool, 0644);
-MODULE_PARM_DESC(poison_extents_on_checksum_error,
- "Extents with checksum errors are marked as poisoned - unsafe without read fua support");
-
#ifndef CONFIG_BCACHEFS_NO_LATENCY_ACCT
static inline u32 bch2_dev_congested_read(struct bch_dev *ca, u64 now)
@@ -551,9 +545,6 @@ static void get_rbio_extent(struct btree_trans *trans,
static noinline int maybe_poison_extent(struct btree_trans *trans, struct bch_read_bio *rbio,
enum btree_id btree, struct bkey_s_c read_k)
{
- if (!bch2_poison_extents_on_checksum_error)
- return 0;
-
struct bch_fs *c = trans->c;
struct data_update *u = rbio_data_update(rbio);
@@ -1291,6 +1282,10 @@ retry_pick:
async_object_list_add(c, rbio, rbio, &rbio->list_idx);
+ /* XXX: also nvme read recovery level */
+ if (unlikely(failed && bch2_dev_io_failures(failed, pick.ptr.dev)))
+ rbio->bio.bi_opf |= REQ_FUA;
+
if (rbio->bounce)
trace_and_count(c, io_read_bounce, &rbio->bio);
diff --git a/fs/bcachefs/movinggc.c b/fs/bcachefs/movinggc.c
index f36d60b8fb07..0f7e35684bc8 100644
--- a/fs/bcachefs/movinggc.c
+++ b/fs/bcachefs/movinggc.c
@@ -62,25 +62,38 @@ static int bch2_bucket_is_movable(struct btree_trans *trans,
{
struct bch_fs *c = trans->c;
- if (bch2_bucket_is_open(c, b->k.bucket.inode, b->k.bucket.offset))
+ /*
+ * Valid bucket?
+ *
+ * XXX: we should kill the LRU entry here if it's not
+ */
+ CLASS(bch2_dev_bucket_tryget, ca)(c, b->k.bucket);
+ if (!ca)
return 0;
- CLASS(btree_iter, iter)(trans, BTREE_ID_alloc, b->k.bucket, BTREE_ITER_cached);
- struct bkey_s_c k = bch2_btree_iter_peek_slot(&iter);
- int ret = bkey_err(k);
- if (ret)
- return ret;
-
- CLASS(bch2_dev_bucket_tryget, ca)(c, k.k->p);
- if (!ca)
+ if (ca->mi.state != BCH_MEMBER_STATE_rw ||
+ !bch2_dev_is_online(ca)) {
+ bch_err_throw(c, bucket_not_moveable_dev_not_rw);
return 0;
+ }
- if (bch2_bucket_bitmap_test(&ca->bucket_backpointer_mismatch, b->k.bucket.offset))
+ /* Bucket still being written? */
+ if (bch2_bucket_is_open(c, b->k.bucket.inode, b->k.bucket.offset)) {
+ bch_err_throw(c, bucket_not_moveable_bucket_open);
return 0;
+ }
- if (ca->mi.state != BCH_MEMBER_STATE_rw ||
- !bch2_dev_is_online(ca))
+ /* We won't be able to evacuate it if there's missing backpointers */
+ if (bch2_bucket_bitmap_test(&ca->bucket_backpointer_mismatch, b->k.bucket.offset)) {
+ bch_err_throw(c, bucket_not_moveable_bp_mismatch);
return 0;
+ }
+
+ CLASS(btree_iter, iter)(trans, BTREE_ID_alloc, b->k.bucket, BTREE_ITER_cached);
+ struct bkey_s_c k = bch2_btree_iter_peek_slot(&iter);
+ int ret = bkey_err(k);
+ if (ret)
+ return ret;
struct bch_alloc_v4 _a;
const struct bch_alloc_v4 *a = bch2_alloc_to_v4(k, &_a);
@@ -88,7 +101,12 @@ static int bch2_bucket_is_movable(struct btree_trans *trans,
b->sectors = bch2_bucket_sectors_dirty(*a);
u64 lru_idx = alloc_lru_idx_fragmentation(*a, ca);
- return lru_idx && lru_idx <= time;
+ if (!lru_idx || lru_idx > time) {
+ bch_err_throw(c, bucket_not_moveable_lru_race);
+ return 0;
+ }
+
+ return true;
}
static void move_bucket_free(struct buckets_in_flight *list,
diff --git a/fs/bcachefs/nocow_locking.c b/fs/bcachefs/nocow_locking.c
index 58cfd540c6d6..71b17f18e90c 100644
--- a/fs/bcachefs/nocow_locking.c
+++ b/fs/bcachefs/nocow_locking.c
@@ -2,11 +2,10 @@
#include "bcachefs.h"
#include "bkey_methods.h"
+#include "closure.h"
#include "nocow_locking.h"
#include "util.h"
-#include <linux/closure.h>
-
bool bch2_bucket_nocow_is_locked(struct bucket_nocow_lock_table *t, struct bpos bucket)
{
u64 dev_bucket = bucket_to_u64(bucket);
diff --git a/fs/bcachefs/sysfs.c b/fs/bcachefs/sysfs.c
index 40adefe7170f..ef6312c50f88 100644
--- a/fs/bcachefs/sysfs.c
+++ b/fs/bcachefs/sysfs.c
@@ -45,6 +45,7 @@
#include <linux/blkdev.h>
#include <linux/sort.h>
+#include <linux/string_choices.h>
#include <linux/sched/clock.h>
#include "util.h"
@@ -157,6 +158,7 @@ write_attribute(trigger_recalc_capacity);
write_attribute(trigger_delete_dead_snapshots);
write_attribute(trigger_emergency_read_only);
read_attribute(gc_gens_pos);
+__sysfs_attribute(read_fua_test, 0400);
read_attribute(uuid);
read_attribute(minor);
@@ -304,6 +306,116 @@ static void bch2_fs_usage_base_to_text(struct printbuf *out, struct bch_fs *c)
prt_printf(out, "reserved:\t\t%llu\n", b.reserved);
}
+static int bch2_read_fua_test(struct printbuf *out, struct bch_dev *ca)
+{
+ struct bch_fs *c = ca->fs;
+ struct bio *bio = NULL;
+ void *buf = NULL;
+ unsigned bs = c->opts.block_size, iters;
+ u64 end, test_duration = NSEC_PER_SEC * 2;
+ struct bch2_time_stats stats_nofua, stats_fua, stats_random;
+ int ret = 0;
+
+ bch2_time_stats_init_no_pcpu(&stats_nofua);
+ bch2_time_stats_init_no_pcpu(&stats_fua);
+ bch2_time_stats_init_no_pcpu(&stats_random);
+
+ if (!bch2_dev_get_ioref(c, ca->dev_idx, READ, BCH_DEV_READ_REF_read_fua_test)) {
+ prt_str(out, "offline\n");
+ return 0;
+ }
+
+ struct block_device *bdev = ca->disk_sb.bdev;
+
+ bio = bio_kmalloc(1, GFP_KERNEL);
+ if (!bio) {
+ ret = -ENOMEM;
+ goto err;
+ }
+
+ buf = kmalloc(bs, GFP_KERNEL);
+ if (!buf)
+ goto err;
+
+ end = ktime_get_ns() + test_duration;
+ for (iters = 0; iters < 1000 && time_before64(ktime_get_ns(), end); iters++) {
+ bio_init(bio, bdev, bio->bi_inline_vecs, 1, READ);
+ bch2_bio_map(bio, buf, bs);
+
+ u64 submit_time = ktime_get_ns();
+ ret = submit_bio_wait(bio);
+ bch2_time_stats_update(&stats_nofua, submit_time);
+
+ if (ret)
+ goto err;
+ }
+
+ end = ktime_get_ns() + test_duration;
+ for (iters = 0; iters < 1000 && time_before64(ktime_get_ns(), end); iters++) {
+ bio_init(bio, bdev, bio->bi_inline_vecs, 1, REQ_FUA|READ);
+ bch2_bio_map(bio, buf, bs);
+
+ u64 submit_time = ktime_get_ns();
+ ret = submit_bio_wait(bio);
+ bch2_time_stats_update(&stats_fua, submit_time);
+
+ if (ret)
+ goto err;
+ }
+
+ u64 dev_size = ca->mi.nbuckets * bucket_bytes(ca);
+
+ end = ktime_get_ns() + test_duration;
+ for (iters = 0; iters < 1000 && time_before64(ktime_get_ns(), end); iters++) {
+ bio_init(bio, bdev, bio->bi_inline_vecs, 1, READ);
+ bio->bi_iter.bi_sector = (bch2_get_random_u64_below(dev_size) & ~((u64) bs - 1)) >> 9;
+ bch2_bio_map(bio, buf, bs);
+
+ u64 submit_time = ktime_get_ns();
+ ret = submit_bio_wait(bio);
+ bch2_time_stats_update(&stats_random, submit_time);
+
+ if (ret)
+ goto err;
+ }
+
+ u64 ns_nofua = mean_and_variance_get_mean(stats_nofua.duration_stats);
+ u64 ns_fua = mean_and_variance_get_mean(stats_fua.duration_stats);
+ u64 ns_rand = mean_and_variance_get_mean(stats_random.duration_stats);
+
+ u64 stddev_nofua = mean_and_variance_get_stddev(stats_nofua.duration_stats);
+ u64 stddev_fua = mean_and_variance_get_stddev(stats_fua.duration_stats);
+ u64 stddev_rand = mean_and_variance_get_stddev(stats_random.duration_stats);
+
+ printbuf_tabstop_push(out, 8);
+ printbuf_tabstop_push(out, 12);
+ printbuf_tabstop_push(out, 12);
+ prt_printf(out, "This test must be run on an idle drive for accurate results\n");
+ prt_printf(out, "%s\n", dev_name(&ca->disk_sb.bdev->bd_device));
+ prt_printf(out, "fua support advertized: %s\n", str_yes_no(bdev_fua(bdev)));
+ prt_newline(out);
+ prt_printf(out, "ns:\tlatency\rstddev\r\n");
+ prt_printf(out, "nofua\t%llu\r%llu\r\n", ns_nofua, stddev_nofua);
+ prt_printf(out, "fua\t%llu\r%llu\r\n", ns_fua, stddev_fua);
+ prt_printf(out, "random\t%llu\r%llu\r\n", ns_rand, stddev_rand);
+
+ bool read_cache = ns_nofua * 2 < ns_rand;
+ bool fua_cached = read_cache && ns_fua < (ns_nofua + ns_rand) / 2;
+
+ if (!read_cache)
+ prt_str(out, "reads don't appear to be cached - safe\n");
+ else if (!fua_cached)
+ prt_str(out, "fua reads don't appear to be cached - safe\n");
+ else
+ prt_str(out, "fua reads appear to be cached - unsafe\n");
+err:
+ kfree(buf);
+ kfree(bio);
+ enumerated_ref_put(&ca->io_ref[READ], BCH_DEV_READ_REF_read_fua_test);
+ bch_err_fn(c, ret);
+ return ret;
+}
+
SHOW(bch2_fs)
{
struct bch_fs *c = container_of(kobj, struct bch_fs, kobj);
@@ -847,6 +959,9 @@ SHOW(bch2_dev)
if (attr == &sysfs_open_buckets)
bch2_open_buckets_to_text(out, c, ca);
+ if (attr == &sysfs_read_fua_test)
+ return bch2_read_fua_test(out, ca);
+
int opt_id = bch2_opt_lookup(attr->name);
if (opt_id >= 0)
return sysfs_opt_show(c, ca, opt_id, out);
@@ -911,6 +1026,8 @@ struct attribute *bch2_dev_files[] = {
&sysfs_congested,
#endif
+ &sysfs_read_fua_test,
+
/* debug: */
&sysfs_alloc_debug,
&sysfs_open_buckets,
diff --git a/fs/bcachefs/util.h b/fs/bcachefs/util.h
index 52ac8230be9f..555e0d8f3cf0 100644
--- a/fs/bcachefs/util.h
+++ b/fs/bcachefs/util.h
@@ -4,7 +4,6 @@
#include <linux/bio.h>
#include <linux/blkdev.h>
-#include <linux/closure.h>
#include <linux/errno.h>
#include <linux/freezer.h>
#include <linux/kernel.h>
@@ -21,6 +20,7 @@
#include <linux/vmalloc.h>
#include <linux/workqueue.h>
+#include "closure.h"
#include "mean_and_variance.h"
#include "darray.h"
diff --git a/fs/bcachefs/vendor/closure.c b/fs/bcachefs/vendor/closure.c
new file mode 100644
index 000000000000..bdafd3a57386
--- /dev/null
+++ b/fs/bcachefs/vendor/closure.c
@@ -0,0 +1,218 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Asynchronous refcounty things
+ *
+ * Copyright 2010, 2011 Kent Overstreet <kent.overstreet@gmail.com>
+ * Copyright 2012 Google, Inc.
+ */
+
+#include "closure.h"
+#include <linux/debugfs.h>
+#include <linux/export.h>
+#include <linux/rcupdate.h>
+#include <linux/seq_file.h>
+#include <linux/sched/debug.h>
+
+static void closure_val_checks(struct closure *cl, unsigned new, int d)
+{
+ unsigned count = new & CLOSURE_REMAINING_MASK;
+
+ if (WARN(new & CLOSURE_GUARD_MASK,
+ "closure %ps has guard bits set: %x (%u), delta %i",
+ cl->fn,
+ new, (unsigned) __fls(new & CLOSURE_GUARD_MASK), d))
+ new &= ~CLOSURE_GUARD_MASK;
+
+ WARN(!count && (new & ~(CLOSURE_DESTRUCTOR|CLOSURE_SLEEPING)),
+ "closure %ps ref hit 0 with incorrect flags set: %x (%u)",
+ cl->fn,
+ new, (unsigned) __fls(new));
+}
+
+enum new_closure_state {
+ CLOSURE_normal_put,
+ CLOSURE_requeue,
+ CLOSURE_done,
+};
+
+/* For clearing flags with the same atomic op as a put */
+void bch2_closure_sub(struct closure *cl, int v)
+{
+ enum new_closure_state s;
+ struct task_struct *sleeper;
+
+ /* rcu_read_lock, atomic_read_acquire() are both for cl->sleeper: */
+ guard(rcu)();
+
+ int old = atomic_read_acquire(&cl->remaining), new;
+ do {
+ new = old - v;
+
+ if (new & CLOSURE_REMAINING_MASK) {
+ s = CLOSURE_normal_put;
+ } else {
+ if ((cl->fn || (new & CLOSURE_SLEEPING)) &&
+ !(new & CLOSURE_DESTRUCTOR)) {
+ s = CLOSURE_requeue;
+ new += CLOSURE_REMAINING_INITIALIZER;
+ } else
+ s = CLOSURE_done;
+
+ sleeper = new & CLOSURE_SLEEPING ? cl->sleeper : NULL;
+ new &= ~CLOSURE_SLEEPING;
+ }
+
+ closure_val_checks(cl, new, -v);
+ } while (!atomic_try_cmpxchg_release(&cl->remaining, &old, new));
+
+ if (s == CLOSURE_normal_put)
+ return;
+
+ if (sleeper) {
+ smp_mb();
+ wake_up_process(sleeper);
+ return;
+ }
+
+ if (s == CLOSURE_requeue) {
+ closure_queue(cl);
+ } else {
+ struct closure *parent = cl->parent;
+ closure_fn *destructor = cl->fn;
+
+ closure_debug_destroy(cl);
+
+ if (destructor)
+ destructor(&cl->work);
+
+ if (parent)
+ closure_put(parent);
+ }
+}
+
+/*
+ * closure_wake_up - wake up all closures on a wait list, without memory barrier
+ */
+void __bch2_closure_wake_up(struct closure_waitlist *wait_list)
+{
+ struct llist_node *list;
+ struct closure *cl, *t;
+ struct llist_node *reverse = NULL;
+
+ list = llist_del_all(&wait_list->list);
+
+ /* We first reverse the list to preserve FIFO ordering and fairness */
+ reverse = llist_reverse_order(list);
+
+ /* Then do the wakeups */
+ llist_for_each_entry_safe(cl, t, reverse, list) {
+ closure_set_waiting(cl, 0);
+ bch2_closure_sub(cl, CLOSURE_WAITING + 1);
+ }
+}
+
+/**
+ * closure_wait - add a closure to a waitlist
+ * @waitlist: will own a ref on @cl, which will be released when
+ * closure_wake_up() is called on @waitlist.
+ * @cl: closure pointer.
+ *
+ */
+bool bch2_closure_wait(struct closure_waitlist *waitlist, struct closure *cl)
+{
+ if (atomic_read(&cl->remaining) & CLOSURE_WAITING)
+ return false;
+
+ closure_set_waiting(cl, _RET_IP_);
+ unsigned r = atomic_add_return(CLOSURE_WAITING + 1, &cl->remaining);
+ closure_val_checks(cl, r, CLOSURE_WAITING + 1);
+
+ llist_add(&cl->list, &waitlist->list);
+
+ return true;
+}
+
+void __sched __bch2_closure_sync(struct closure *cl)
+{
+ cl->sleeper = current;
+ bch2_closure_sub(cl,
+ CLOSURE_REMAINING_INITIALIZER -
+ CLOSURE_SLEEPING);
+
+ while (1) {
+ set_current_state(TASK_UNINTERRUPTIBLE);
+ if (!(atomic_read(&cl->remaining) & CLOSURE_SLEEPING))
+ break;
+ schedule();
+ }
+
+ __set_current_state(TASK_RUNNING);
+}
+
+/*
+ * closure_return_sync - finish running a closure, synchronously (i.e. waiting
+ * for outstanding get()s to finish) and returning once closure refcount is 0.
+ *
+ * Unlike closure_sync() this doesn't reinit the ref to 1; subsequent
+ * closure_get_not_zero() calls will fail.
+ */
+void __sched bch2_closure_return_sync(struct closure *cl)
+{
+ cl->sleeper = current;
+ bch2_closure_sub(cl,
+ CLOSURE_REMAINING_INITIALIZER -
+ CLOSURE_DESTRUCTOR -
+ CLOSURE_SLEEPING);
+
+ while (1) {
+ set_current_state(TASK_UNINTERRUPTIBLE);
+ if (!(atomic_read(&cl->remaining) & CLOSURE_SLEEPING))
+ break;
+ schedule();
+ }
+
+ __set_current_state(TASK_RUNNING);
+
+ if (cl->parent)
+ closure_put(cl->parent);
+}
+
+int __sched __bch2_closure_sync_timeout(struct closure *cl, unsigned long timeout)
+{
+ int ret = 0;
+
+ cl->sleeper = current;
+ bch2_closure_sub(cl,
+ CLOSURE_REMAINING_INITIALIZER -
+ CLOSURE_SLEEPING);
+
+ while (1) {
+ set_current_state(TASK_UNINTERRUPTIBLE);
+ /*
+ * Carefully undo the continue_at() - but only if it
+ * hasn't completed, i.e. the final closure_put() hasn't
+ * happened yet:
+ */
+ unsigned old = atomic_read(&cl->remaining), new;
+ if (!(old & CLOSURE_SLEEPING))
+ goto success;
+
+ if (!timeout) {
+ do {
+ if (!(old & CLOSURE_SLEEPING))
+ goto success;
+
+ new = old + CLOSURE_REMAINING_INITIALIZER - CLOSURE_SLEEPING;
+ closure_val_checks(cl, new, CLOSURE_REMAINING_INITIALIZER - CLOSURE_SLEEPING);
+ } while (!atomic_try_cmpxchg(&cl->remaining, &old, new));
+
+ ret = -ETIME;
+ break;
+ }
+
+ timeout = schedule_timeout(timeout);
+ }
+success:
+ __set_current_state(TASK_RUNNING);
+ return ret;
+}
diff --git a/fs/bcachefs/vendor/closure.h b/fs/bcachefs/vendor/closure.h
new file mode 100644
index 000000000000..79112efe30a7
--- /dev/null
+++ b/fs/bcachefs/vendor/closure.h
@@ -0,0 +1,490 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_CLOSURE_H
+#define _LINUX_CLOSURE_H
+
+#include <linux/llist.h>
+#include <linux/sched.h>
+#include <linux/sched/task_stack.h>
+#include <linux/workqueue.h>
+
+/*
+ * Closure is perhaps the most overused and abused term in computer science, but
+ * since I've been unable to come up with anything better you're stuck with it
+ * again.
+ *
+ * What are closures?
+ *
+ * They embed a refcount. The basic idea is they count "things that are in
+ * progress" - in flight bios, some other thread that's doing something else -
+ * anything you might want to wait on.
+ *
+ * The refcount may be manipulated with closure_get() and closure_put().
+ * closure_put() is where many of the interesting things happen, when it causes
+ * the refcount to go to 0.
+ *
+ * Closures can be used to wait on things both synchronously and asynchronously,
+ * and synchronous and asynchronous use can be mixed without restriction. To
+ * wait synchronously, use closure_sync() - you will sleep until your closure's
+ * refcount hits 1.
+ *
+ * To wait asynchronously, use
+ * continue_at(cl, next_function, workqueue);
+ *
+ * passing it, as you might expect, the function to run when nothing is pending
+ * and the workqueue to run that function out of.
+ *
+ * continue_at() also, critically, requires a 'return' immediately following the
+ * location where this macro is referenced, to return to the calling function.
+ * There's good reason for this.
+ *
+ * To use safely closures asynchronously, they must always have a refcount while
+ * they are running owned by the thread that is running them. Otherwise, suppose
+ * you submit some bios and wish to have a function run when they all complete:
+ *
+ * foo_endio(struct bio *bio)
+ * {
+ * closure_put(cl);
+ * }
+ *
+ * closure_init(cl);
+ *
+ * do_stuff();
+ * closure_get(cl);
+ * bio1->bi_endio = foo_endio;
+ * bio_submit(bio1);
+ *
+ * do_more_stuff();
+ * closure_get(cl);
+ * bio2->bi_endio = foo_endio;
+ * bio_submit(bio2);
+ *
+ * continue_at(cl, complete_some_read, system_wq);
+ *
+ * If closure's refcount started at 0, complete_some_read() could run before the
+ * second bio was submitted - which is almost always not what you want! More
+ * importantly, it wouldn't be possible to say whether the original thread or
+ * complete_some_read()'s thread owned the closure - and whatever state it was
+ * associated with!
+ *
+ * So, closure_init() initializes a closure's refcount to 1 - and when a
+ * closure_fn is run, the refcount will be reset to 1 first.
+ *
+ * Then, the rule is - if you got the refcount with closure_get(), release it
+ * with closure_put() (i.e, in a bio->bi_endio function). If you have a refcount
+ * on a closure because you called closure_init() or you were run out of a
+ * closure - _always_ use continue_at(). Doing so consistently will help
+ * eliminate an entire class of particularly pernicious races.
+ *
+ * Lastly, you might have a wait list dedicated to a specific event, and have no
+ * need for specifying the condition - you just want to wait until someone runs
+ * closure_wake_up() on the appropriate wait list. In that case, just use
+ * closure_wait(). It will return either true or false, depending on whether the
+ * closure was already on a wait list or not - a closure can only be on one wait
+ * list at a time.
+ *
+ * Parents:
+ *
+ * closure_init() takes two arguments - it takes the closure to initialize, and
+ * a (possibly null) parent.
+ *
+ * If parent is non null, the new closure will have a refcount for its lifetime;
+ * a closure is considered to be "finished" when its refcount hits 0 and the
+ * function to run is null. Hence
+ *
+ * continue_at(cl, NULL, NULL);
+ *
+ * returns up the (spaghetti) stack of closures, precisely like normal return
+ * returns up the C stack. continue_at() with non null fn is better thought of
+ * as doing a tail call.
+ *
+ * All this implies that a closure should typically be embedded in a particular
+ * struct (which its refcount will normally control the lifetime of), and that
+ * struct can very much be thought of as a stack frame.
+ */
+
+struct closure;
+struct closure_syncer;
+typedef void (closure_fn) (struct work_struct *);
+extern struct dentry *bcache_debug;
+
+struct closure_waitlist {
+ struct llist_head list;
+};
+
+enum closure_state {
+ /*
+ * CLOSURE_WAITING: Set iff the closure is on a waitlist. Must be set by
+ * the thread that owns the closure, and cleared by the thread that's
+ * waking up the closure.
+ *
+ * The rest are for debugging and don't affect behaviour:
+ *
+ * CLOSURE_RUNNING: Set when a closure is running (i.e. by
+ * closure_init() and when closure_put() runs then next function), and
+ * must be cleared before remaining hits 0. Primarily to help guard
+ * against incorrect usage and accidentally transferring references.
+ * continue_at() and closure_return() clear it for you, if you're doing
+ * something unusual you can use closure_set_dead() which also helps
+ * annotate where references are being transferred.
+ */
+
+ CLOSURE_BITS_START = (1U << 24),
+ CLOSURE_DESTRUCTOR = (1U << 24),
+ CLOSURE_SLEEPING = (1U << 26),
+ CLOSURE_WAITING = (1U << 28),
+ CLOSURE_RUNNING = (1U << 30),
+};
+
+#define CLOSURE_GUARD_MASK \
+ (((CLOSURE_DESTRUCTOR|CLOSURE_SLEEPING|CLOSURE_WAITING|CLOSURE_RUNNING) << 1)|(CLOSURE_BITS_START >> 1))
+
+#define CLOSURE_REMAINING_MASK (CLOSURE_BITS_START - 1)
+#define CLOSURE_REMAINING_INITIALIZER (1|CLOSURE_RUNNING)
+
+struct closure {
+ union {
+ struct {
+ struct workqueue_struct *wq;
+ struct task_struct *sleeper;
+ struct llist_node list;
+ closure_fn *fn;
+ };
+ struct work_struct work;
+ };
+
+ struct closure *parent;
+
+ atomic_t remaining;
+
+#ifdef CONFIG_DEBUG_CLOSURES
+#define CLOSURE_MAGIC_DEAD 0xc054dead
+#define CLOSURE_MAGIC_ALIVE 0xc054a11e
+#define CLOSURE_MAGIC_STACK 0xc05451cc
+
+ unsigned int magic;
+ struct list_head all;
+ unsigned long ip;
+ unsigned long waiting_on;
+#endif
+};
+
+void bch2_closure_sub(struct closure *cl, int v);
+void __bch2_closure_wake_up(struct closure_waitlist *list);
+bool bch2_closure_wait(struct closure_waitlist *list, struct closure *cl);
+void __bch2_closure_sync(struct closure *cl);
+
+/*
+ * closure_put - decrement a closure's refcount
+ */
+static inline void closure_put(struct closure *cl)
+{
+ bch2_closure_sub(cl, 1);
+}
+
+static inline unsigned closure_nr_remaining(struct closure *cl)
+{
+ return atomic_read(&cl->remaining) & CLOSURE_REMAINING_MASK;
+}
+
+/**
+ * closure_sync - sleep until a closure a closure has nothing left to wait on
+ *
+ * Sleeps until the refcount hits 1 - the thread that's running the closure owns
+ * the last refcount.
+ */
+static inline void closure_sync(struct closure *cl)
+{
+ if (closure_nr_remaining(cl) > 1)
+ __bch2_closure_sync(cl);
+}
+
+int __bch2_closure_sync_timeout(struct closure *cl, unsigned long timeout);
+
+static inline int closure_sync_timeout(struct closure *cl, unsigned long timeout)
+{
+ return closure_nr_remaining(cl) > 1
+ ? __bch2_closure_sync_timeout(cl, timeout)
+ : 0;
+}
+
+//#ifdef CONFIG_DEBUG_CLOSURES
+#if 0
+
+void bch2_closure_debug_create(struct closure *cl);
+void closure_debug_destroy(struct closure *cl);
+
+#else
+
+static inline void bch2_closure_debug_create(struct closure *cl) {}
+static inline void closure_debug_destroy(struct closure *cl) {}
+
+#endif
+
+static inline void closure_set_ip(struct closure *cl)
+{
+#ifdef CONFIG_DEBUG_CLOSURES
+ cl->ip = _THIS_IP_;
+#endif
+}
+
+static inline void closure_set_ret_ip(struct closure *cl)
+{
+#ifdef CONFIG_DEBUG_CLOSURES
+ cl->ip = _RET_IP_;
+#endif
+}
+
+static inline void closure_set_waiting(struct closure *cl, unsigned long f)
+{
+#ifdef CONFIG_DEBUG_CLOSURES
+ cl->waiting_on = f;
+#endif
+}
+
+static inline void closure_set_stopped(struct closure *cl)
+{
+ atomic_sub(CLOSURE_RUNNING, &cl->remaining);
+}
+
+static inline void set_closure_fn(struct closure *cl, closure_fn *fn,
+ struct workqueue_struct *wq)
+{
+ closure_set_ip(cl);
+ cl->fn = fn;
+ cl->wq = wq;
+}
+
+static inline void closure_queue(struct closure *cl)
+{
+ struct workqueue_struct *wq = cl->wq;
+ /**
+ * Changes made to closure, work_struct, or a couple of other structs
+ * may cause work.func not pointing to the right location.
+ */
+ BUILD_BUG_ON(offsetof(struct closure, fn)
+ != offsetof(struct work_struct, func));
+
+ if (wq) {
+ INIT_WORK(&cl->work, cl->work.func);
+ BUG_ON(!queue_work(wq, &cl->work));
+ } else
+ cl->fn(&cl->work);
+}
+
+/**
+ * closure_get - increment a closure's refcount
+ */
+static inline void closure_get(struct closure *cl)
+{
+#ifdef CONFIG_DEBUG_CLOSURES
+ BUG_ON((atomic_inc_return(&cl->remaining) &
+ CLOSURE_REMAINING_MASK) <= 1);
+#else
+ atomic_inc(&cl->remaining);
+#endif
+}
+
+/**
+ * closure_get_not_zero
+ */
+static inline bool closure_get_not_zero(struct closure *cl)
+{
+ unsigned old = atomic_read(&cl->remaining);
+ do {
+ if (!(old & CLOSURE_REMAINING_MASK))
+ return false;
+
+ } while (!atomic_try_cmpxchg_acquire(&cl->remaining, &old, old + 1));
+
+ return true;
+}
+
+/**
+ * closure_init - Initialize a closure, setting the refcount to 1
+ * @cl: closure to initialize
+ * @parent: parent of the new closure. cl will take a refcount on it for its
+ * lifetime; may be NULL.
+ */
+static inline void closure_init(struct closure *cl, struct closure *parent)
+{
+ cl->fn = NULL;
+ cl->parent = parent;
+ if (parent)
+ closure_get(parent);
+
+ atomic_set(&cl->remaining, CLOSURE_REMAINING_INITIALIZER);
+
+ bch2_closure_debug_create(cl);
+ closure_set_ip(cl);
+}
+
+static inline void closure_init_stack(struct closure *cl)
+{
+ memset(cl, 0, sizeof(struct closure));
+ atomic_set(&cl->remaining, CLOSURE_REMAINING_INITIALIZER);
+#ifdef CONFIG_DEBUG_CLOSURES
+ cl->magic = CLOSURE_MAGIC_STACK;
+#endif
+}
+
+static inline void closure_init_stack_release(struct closure *cl)
+{
+ memset(cl, 0, sizeof(struct closure));
+ atomic_set_release(&cl->remaining, CLOSURE_REMAINING_INITIALIZER);
+#ifdef CONFIG_DEBUG_CLOSURES
+ cl->magic = CLOSURE_MAGIC_STACK;
+#endif
+}
+
+/**
+ * closure_wake_up - wake up all closures on a wait list,
+ * with memory barrier
+ */
+static inline void closure_wake_up(struct closure_waitlist *list)
+{
+ /* Memory barrier for the wait list */
+ smp_mb();
+ __bch2_closure_wake_up(list);
+}
+
+#define CLOSURE_CALLBACK(name) void name(struct work_struct *ws)
+#define closure_type(name, type, member) \
+ struct closure *cl = container_of(ws, struct closure, work); \
+ type *name = container_of(cl, type, member)
+
+/**
+ * continue_at - jump to another function with barrier
+ *
+ * After @cl is no longer waiting on anything (i.e. all outstanding refs have
+ * been dropped with closure_put()), it will resume execution at @fn running out
+ * of @wq (or, if @wq is NULL, @fn will be called by closure_put() directly).
+ *
+ * This is because after calling continue_at() you no longer have a ref on @cl,
+ * and whatever @cl owns may be freed out from under you - a running closure fn
+ * has a ref on its own closure which continue_at() drops.
+ *
+ * Note you are expected to immediately return after using this macro.
+ */
+#define continue_at(_cl, _fn, _wq) \
+do { \
+ set_closure_fn(_cl, _fn, _wq); \
+ bch2_closure_sub(_cl, CLOSURE_RUNNING + 1); \
+} while (0)
+
+/**
+ * closure_return - finish execution of a closure
+ *
+ * This is used to indicate that @cl is finished: when all outstanding refs on
+ * @cl have been dropped @cl's ref on its parent closure (as passed to
+ * closure_init()) will be dropped, if one was specified - thus this can be
+ * thought of as returning to the parent closure.
+ */
+#define closure_return(_cl) continue_at((_cl), NULL, NULL)
+
+void bch2_closure_return_sync(struct closure *cl);
+
+/**
+ * continue_at_nobarrier - jump to another function without barrier
+ *
+ * Causes @fn to be executed out of @cl, in @wq context (or called directly if
+ * @wq is NULL).
+ *
+ * The ref the caller of continue_at_nobarrier() had on @cl is now owned by @fn,
+ * thus it's not safe to touch anything protected by @cl after a
+ * continue_at_nobarrier().
+ */
+#define continue_at_nobarrier(_cl, _fn, _wq) \
+do { \
+ set_closure_fn(_cl, _fn, _wq); \
+ closure_queue(_cl); \
+} while (0)
+
+/**
+ * closure_return_with_destructor - finish execution of a closure,
+ * with destructor
+ *
+ * Works like closure_return(), except @destructor will be called when all
+ * outstanding refs on @cl have been dropped; @destructor may be used to safely
+ * free the memory occupied by @cl, and it is called with the ref on the parent
+ * closure still held - so @destructor could safely return an item to a
+ * freelist protected by @cl's parent.
+ */
+#define closure_return_with_destructor(_cl, _destructor) \
+do { \
+ set_closure_fn(_cl, _destructor, NULL); \
+ bch2_closure_sub(_cl, CLOSURE_RUNNING - CLOSURE_DESTRUCTOR + 1); \
+} while (0)
+
+/**
+ * closure_call - execute @fn out of a new, uninitialized closure
+ *
+ * Typically used when running out of one closure, and we want to run @fn
+ * asynchronously out of a new closure - @parent will then wait for @cl to
+ * finish.
+ */
+static inline void closure_call(struct closure *cl, closure_fn fn,
+ struct workqueue_struct *wq,
+ struct closure *parent)
+{
+ closure_init(cl, parent);
+ continue_at_nobarrier(cl, fn, wq);
+}
+
+#define __closure_wait_event(waitlist, _cond) \
+do { \
+ struct closure cl; \
+ \
+ closure_init_stack(&cl); \
+ \
+ while (1) { \
+ bch2_closure_wait(waitlist, &cl); \
+ if (_cond) \
+ break; \
+ closure_sync(&cl); \
+ } \
+ closure_wake_up(waitlist); \
+ closure_sync(&cl); \
+} while (0)
+
+#define closure_wait_event(waitlist, _cond) \
+do { \
+ if (!(_cond)) \
+ __closure_wait_event(waitlist, _cond); \
+} while (0)
+
+#define __closure_wait_event_timeout(waitlist, _cond, _until) \
+({ \
+ struct closure cl; \
+ long _t; \
+ \
+ closure_init_stack(&cl); \
+ \
+ while (1) { \
+ bch2_closure_wait(waitlist, &cl); \
+ if (_cond) { \
+ _t = max_t(long, 1L, _until - jiffies); \
+ break; \
+ } \
+ _t = max_t(long, 0L, _until - jiffies); \
+ if (!_t) \
+ break; \
+ closure_sync_timeout(&cl, _t); \
+ } \
+ closure_wake_up(waitlist); \
+ closure_sync(&cl); \
+ _t; \
+})
+
+/*
+ * Returns 0 if timeout expired, remaining time in jiffies (at least 1) if
+ * condition became true
+ */
+#define closure_wait_event_timeout(waitlist, _cond, _timeout) \
+({ \
+ unsigned long _until = jiffies + _timeout; \
+ (_cond) \
+ ? max_t(long, 1L, _until - jiffies) \
+ : __closure_wait_event_timeout(waitlist, _cond, _until);\
+})
+
+#endif /* _LINUX_CLOSURE_H */
generated by cgit v1.2.3 (git 2.51.0) at 2025-09-20 01:48:33 +0000