1 files changed, 17 insertions, 153 deletions

diff --git a/fs/erofs/zutil.c b/fs/erofs/zutil.c
index b80f612867c2..55ff2ab5128e 100644
--- a/fs/erofs/zutil.c
+++ b/fs/erofs/zutil.c
@@ -2,6 +2,7 @@
 /*
  * Copyright (C) 2018 HUAWEI, Inc.
  *             https://www.huawei.com/
+ * Copyright (C) 2024 Alibaba Cloud
  */
 #include "internal.h"
 
@@ -19,13 +20,12 @@ static unsigned int z_erofs_gbuf_count, z_erofs_gbuf_nrpages,
 module_param_named(global_buffers, z_erofs_gbuf_count, uint, 0444);
 module_param_named(reserved_pages, z_erofs_rsv_nrpages, uint, 0444);
 
-static atomic_long_t erofs_global_shrink_cnt;	/* for all mounted instances */
-/* protected by 'erofs_sb_list_lock' */
-static unsigned int shrinker_run_no;
+atomic_long_t erofs_global_shrink_cnt;	/* for all mounted instances */
 
-/* protects the mounted 'erofs_sb_list' */
+/* protects `erofs_sb_list_lock` and the mounted `erofs_sb_list` */
 static DEFINE_SPINLOCK(erofs_sb_list_lock);
 static LIST_HEAD(erofs_sb_list);
+static unsigned int shrinker_run_no;
 static struct shrinker *erofs_shrinker_info;
 
 static unsigned int z_erofs_gbuf_id(void)
@@ -38,11 +38,13 @@ void *z_erofs_get_gbuf(unsigned int requiredpages)
 {
 	struct z_erofs_gbuf *gbuf;
 
+	migrate_disable();
 	gbuf = &z_erofs_gbufpool[z_erofs_gbuf_id()];
 	spin_lock(&gbuf->lock);
 	/* check if the buffer is too small */
 	if (requiredpages > gbuf->nrpages) {
 		spin_unlock(&gbuf->lock);
+		migrate_enable();
 		/* (for sparse checker) pretend gbuf->lock is still taken */
 		__acquire(gbuf->lock);
 		return NULL;
@@ -57,6 +59,7 @@ void z_erofs_put_gbuf(void *ptr) __releases(gbuf->lock)
 	gbuf = &z_erofs_gbufpool[z_erofs_gbuf_id()];
 	DBG_BUGON(gbuf->ptr != ptr);
 	spin_unlock(&gbuf->lock);
+	migrate_enable();
 }
 
 int z_erofs_gbuf_growsize(unsigned int nrpages)
@@ -84,8 +87,8 @@ int z_erofs_gbuf_growsize(unsigned int nrpages)
 			tmp_pages[j] = gbuf->pages[j];
 		do {
 			last = j;
-			j = alloc_pages_bulk_array(GFP_KERNEL, nrpages,
-						   tmp_pages);
+			j = alloc_pages_bulk(GFP_KERNEL, nrpages,
+					     tmp_pages);
 			if (last == j)
 				goto out;
 		} while (j != nrpages);
@@ -108,7 +111,8 @@ int z_erofs_gbuf_growsize(unsigned int nrpages)
 out:
 	if (i < z_erofs_gbuf_count && tmp_pages) {
 		for (j = 0; j < nrpages; ++j)
-			if (tmp_pages[j] && tmp_pages[j] != gbuf->pages[j])
+			if (tmp_pages[j] && (j >= gbuf->nrpages ||
+					     tmp_pages[j] != gbuf->pages[j]))
 				__free_page(tmp_pages[j]);
 		kfree(tmp_pages);
 	}
@@ -210,145 +214,6 @@ void erofs_release_pages(struct page **pagepool)
 	}
 }
 
-static bool erofs_workgroup_get(struct erofs_workgroup *grp)
-{
-	if (lockref_get_not_zero(&grp->lockref))
-		return true;
-
-	spin_lock(&grp->lockref.lock);
-	if (__lockref_is_dead(&grp->lockref)) {
-		spin_unlock(&grp->lockref.lock);
-		return false;
-	}
-
-	if (!grp->lockref.count++)
-		atomic_long_dec(&erofs_global_shrink_cnt);
-	spin_unlock(&grp->lockref.lock);
-	return true;
-}
-
-struct erofs_workgroup *erofs_find_workgroup(struct super_block *sb,
-					     pgoff_t index)
-{
-	struct erofs_sb_info *sbi = EROFS_SB(sb);
-	struct erofs_workgroup *grp;
-
-repeat:
-	rcu_read_lock();
-	grp = xa_load(&sbi->managed_pslots, index);
-	if (grp) {
-		if (!erofs_workgroup_get(grp)) {
-			/* prefer to relax rcu read side */
-			rcu_read_unlock();
-			goto repeat;
-		}
-
-		DBG_BUGON(index != grp->index);
-	}
-	rcu_read_unlock();
-	return grp;
-}
-
-struct erofs_workgroup *erofs_insert_workgroup(struct super_block *sb,
-					       struct erofs_workgroup *grp)
-{
-	struct erofs_sb_info *const sbi = EROFS_SB(sb);
-	struct erofs_workgroup *pre;
-
-	DBG_BUGON(grp->lockref.count < 1);
-repeat:
-	xa_lock(&sbi->managed_pslots);
-	pre = __xa_cmpxchg(&sbi->managed_pslots, grp->index,
-			   NULL, grp, GFP_KERNEL);
-	if (pre) {
-		if (xa_is_err(pre)) {
-			pre = ERR_PTR(xa_err(pre));
-		} else if (!erofs_workgroup_get(pre)) {
-			/* try to legitimize the current in-tree one */
-			xa_unlock(&sbi->managed_pslots);
-			cond_resched();
-			goto repeat;
-		}
-		grp = pre;
-	}
-	xa_unlock(&sbi->managed_pslots);
-	return grp;
-}
-
-static void  __erofs_workgroup_free(struct erofs_workgroup *grp)
-{
-	atomic_long_dec(&erofs_global_shrink_cnt);
-	erofs_workgroup_free_rcu(grp);
-}
-
-void erofs_workgroup_put(struct erofs_workgroup *grp)
-{
-	if (lockref_put_or_lock(&grp->lockref))
-		return;
-
-	DBG_BUGON(__lockref_is_dead(&grp->lockref));
-	if (grp->lockref.count == 1)
-		atomic_long_inc(&erofs_global_shrink_cnt);
-	--grp->lockref.count;
-	spin_unlock(&grp->lockref.lock);
-}
-
-static bool erofs_try_to_release_workgroup(struct erofs_sb_info *sbi,
-					   struct erofs_workgroup *grp)
-{
-	int free = false;
-
-	spin_lock(&grp->lockref.lock);
-	if (grp->lockref.count)
-		goto out;
-
-	/*
-	 * Note that all cached pages should be detached before deleted from
-	 * the XArray. Otherwise some cached pages could be still attached to
-	 * the orphan old workgroup when the new one is available in the tree.
-	 */
-	if (erofs_try_to_free_all_cached_folios(sbi, grp))
-		goto out;
-
-	/*
-	 * It's impossible to fail after the workgroup is freezed,
-	 * however in order to avoid some race conditions, add a
-	 * DBG_BUGON to observe this in advance.
-	 */
-	DBG_BUGON(__xa_erase(&sbi->managed_pslots, grp->index) != grp);
-
-	lockref_mark_dead(&grp->lockref);
-	free = true;
-out:
-	spin_unlock(&grp->lockref.lock);
-	if (free)
-		__erofs_workgroup_free(grp);
-	return free;
-}
-
-static unsigned long erofs_shrink_workstation(struct erofs_sb_info *sbi,
-					      unsigned long nr_shrink)
-{
-	struct erofs_workgroup *grp;
-	unsigned int freed = 0;
-	unsigned long index;
-
-	xa_lock(&sbi->managed_pslots);
-	xa_for_each(&sbi->managed_pslots, index, grp) {
-		/* try to shrink each valid workgroup */
-		if (!erofs_try_to_release_workgroup(sbi, grp))
-			continue;
-		xa_unlock(&sbi->managed_pslots);
-
-		++freed;
-		if (!--nr_shrink)
-			return freed;
-		xa_lock(&sbi->managed_pslots);
-	}
-	xa_unlock(&sbi->managed_pslots);
-	return freed;
-}
-
 void erofs_shrinker_register(struct super_block *sb)
 {
 	struct erofs_sb_info *sbi = EROFS_SB(sb);
@@ -365,9 +230,10 @@ void erofs_shrinker_unregister(struct super_block *sb)
 	struct erofs_sb_info *const sbi = EROFS_SB(sb);
 
 	mutex_lock(&sbi->umount_mutex);
-	/* clean up all remaining workgroups in memory */
-	erofs_shrink_workstation(sbi, ~0UL);
-
+	while (!xa_empty(&sbi->managed_pslots)) {
+		z_erofs_shrink_scan(sbi, ~0UL);
+		cond_resched();
+	}
 	spin_lock(&erofs_sb_list_lock);
 	list_del(&sbi->list);
 	spin_unlock(&erofs_sb_list_lock);
@@ -377,7 +243,7 @@ void erofs_shrinker_unregister(struct super_block *sb)
 static unsigned long erofs_shrink_count(struct shrinker *shrink,
 					struct shrink_control *sc)
 {
-	return atomic_long_read(&erofs_global_shrink_cnt);
+	return atomic_long_read(&erofs_global_shrink_cnt) ?: SHRINK_EMPTY;
 }
 
 static unsigned long erofs_shrink_scan(struct shrinker *shrink,
@@ -414,9 +280,7 @@ static unsigned long erofs_shrink_scan(struct shrinker *shrink,
 
 		spin_unlock(&erofs_sb_list_lock);
 		sbi->shrinker_run_no = run_no;
-
-		freed += erofs_shrink_workstation(sbi, nr - freed);
-
+		freed += z_erofs_shrink_scan(sbi, nr - freed);
 		spin_lock(&erofs_sb_list_lock);
 		/* Get the next list element before we move this one */
 		p = p->next;