[PATCH v4 0/4] hugetlb memcg accounting

[PATCH v4 0/4] hugetlb memcg accounting

[Nhat Pham]

Changelog:
v4:
	* Add another prep patch to clean up memory controller migration
	  logic.
	* Fix an issue in hugetlb folio migration where the new folio
	  is not properly charged (patch 3) (reported by Mike Kravetz)
	  (suggested by Johannes Weiner).
v3:
	* Add a prep patch at the start of the series to extend the memory
	  controller interface with new helper functions for hugetlb
	  accounting.
	* Do not account hugetlb memory for memcontroller in cgroup v1
	  (patch 2) (suggested by Johannes Weiner).
	* Change the gfp flag passed to mem cgroup charging (patch 2)
	  (suggested by Michal Hocko).
	* Add caveats to cgroup admin guide and commit changelog
	  (patch 2) (suggested by Michal Hocko).
v2:
	* Add a cgroup mount option to enable/disable the new hugetlb memcg
	  accounting behavior (patch 1) (suggested by Johannes Weiner).
	* Add a couple more ksft_print_msg() on error to aid debugging when
	  the selftest fails. (patch 2)

Currently, hugetlb memory usage is not acounted for in the memory
controller, which could lead to memory overprotection for cgroups with
hugetlb-backed memory. This has been observed in our production system.

For instance, here is one of our usecases: suppose there are two 32G
containers. The machine is booted with hugetlb_cma=6G, and each 
container may or may not use up to 3 gigantic page, depending on the 
workload within it. The rest is anon, cache, slab, etc. We can set the
hugetlb cgroup limit of each cgroup to 3G to enforce hugetlb fairness. 
But it is very difficult to configure memory.max to keep overall 
consumption, including anon, cache, slab etc. fair.
 
What we have had to resort to is to constantly poll hugetlb usage and
readjust memory.max. Similar procedure is done to other memory limits
(memory.low for e.g). However, this is rather cumbersome and buggy.
Furthermore, when there is a delay in memory limits correction, (for
e.g when hugetlb usage changes within consecutive runs of the userspace
agent), the system could be in an over/underprotected state.

This patch series rectifies this issue by charging the memcg when the
hugetlb folio is allocated, and uncharging when the folio is freed. In
addition, a new selftest is added to demonstrate and verify this new
behavior.

Nhat Pham (4):
  memcontrol: add helpers for hugetlb memcg accounting
  memcontrol: only transfer the memcg data for migration
  hugetlb: memcg: account hugetlb-backed memory in memory controller
  selftests: add a selftest to verify hugetlb usage in memcg

 Documentation/admin-guide/cgroup-v2.rst       |  29 +++
 MAINTAINERS                                   |   2 +
 include/linux/cgroup-defs.h                   |   5 +
 include/linux/memcontrol.h                    |  37 +++
 kernel/cgroup/cgroup.c                        |  15 +-
 mm/filemap.c                                  |   2 +-
 mm/hugetlb.c                                  |  35 ++-
 mm/memcontrol.c                               | 139 +++++++++--
 mm/migrate.c                                  |   3 +-
 tools/testing/selftests/cgroup/.gitignore     |   1 +
 tools/testing/selftests/cgroup/Makefile       |   2 +
 .../selftests/cgroup/test_hugetlb_memcg.c     | 234 ++++++++++++++++++
 12 files changed, 478 insertions(+), 26 deletions(-)
 create mode 100644 tools/testing/selftests/cgroup/test_hugetlb_memcg.c

-- 
2.34.1

[PATCH v4 1/4] memcontrol: add helpers for hugetlb memcg accounting

[Nhat Pham]

This patch exposes charge committing and cancelling as parts of the
memory controller interface. These functionalities are useful when the
try_charge() and commit_charge() stages have to be separated by other
actions in between (which can fail). One such example is the new hugetlb
accounting behavior in the following patch.

The patch also adds a helper function to obtain a reference to the
current task's memcg.

Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
---
 include/linux/memcontrol.h | 21 ++++++++++++++
 mm/memcontrol.c            | 59 ++++++++++++++++++++++++++++++--------
 2 files changed, 68 insertions(+), 12 deletions(-)

diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index e0cfab58ab71..42bf7e9b1a2f 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -653,6 +653,8 @@ static inline bool mem_cgroup_below_min(struct mem_cgroup *target,
 		page_counter_read(&memcg->memory);
 }
 
+void mem_cgroup_commit_charge(struct folio *folio, struct mem_cgroup *memcg);
+
 int __mem_cgroup_charge(struct folio *folio, struct mm_struct *mm, gfp_t gfp);
 
 /**
@@ -704,6 +706,8 @@ static inline void mem_cgroup_uncharge_list(struct list_head *page_list)
 	__mem_cgroup_uncharge_list(page_list);
 }
 
+void mem_cgroup_cancel_charge(struct mem_cgroup *memcg, unsigned int nr_pages);
+
 void mem_cgroup_migrate(struct folio *old, struct folio *new);
 
 /**
@@ -760,6 +764,8 @@ struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p);
 
 struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm);
 
+struct mem_cgroup *get_mem_cgroup_from_current(void);
+
 struct lruvec *folio_lruvec_lock(struct folio *folio);
 struct lruvec *folio_lruvec_lock_irq(struct folio *folio);
 struct lruvec *folio_lruvec_lock_irqsave(struct folio *folio,
@@ -1245,6 +1251,11 @@ static inline bool mem_cgroup_below_min(struct mem_cgroup *target,
 	return false;
 }
 
+static inline void mem_cgroup_commit_charge(struct folio *folio,
+		struct mem_cgroup *memcg)
+{
+}
+
 static inline int mem_cgroup_charge(struct folio *folio,
 		struct mm_struct *mm, gfp_t gfp)
 {
@@ -1269,6 +1280,11 @@ static inline void mem_cgroup_uncharge_list(struct list_head *page_list)
 {
 }
 
+static inline void mem_cgroup_cancel_charge(struct mem_cgroup *memcg,
+		unsigned int nr_pages)
+{
+}
+
 static inline void mem_cgroup_migrate(struct folio *old, struct folio *new)
 {
 }
@@ -1306,6 +1322,11 @@ static inline struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm)
 	return NULL;
 }
 
+static inline struct mem_cgroup *get_mem_cgroup_from_current(void)
+{
+	return NULL;
+}
+
 static inline
 struct mem_cgroup *mem_cgroup_from_css(struct cgroup_subsys_state *css)
 {
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index d1a322a75172..0219befeae38 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1086,6 +1086,27 @@ struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm)
 }
 EXPORT_SYMBOL(get_mem_cgroup_from_mm);
 
+/**
+ * get_mem_cgroup_from_current - Obtain a reference on current task's memcg.
+ */
+struct mem_cgroup *get_mem_cgroup_from_current(void)
+{
+	struct mem_cgroup *memcg;
+
+	if (mem_cgroup_disabled())
+		return NULL;
+
+again:
+	rcu_read_lock();
+	memcg = mem_cgroup_from_task(current);
+	if (!css_tryget(&memcg->css)) {
+		rcu_read_unlock();
+		goto again;
+	}
+	rcu_read_unlock();
+	return memcg;
+}
+
 static __always_inline bool memcg_kmem_bypass(void)
 {
 	/* Allow remote memcg charging from any context. */
@@ -2873,7 +2894,12 @@ static inline int try_charge(struct mem_cgroup *memcg, gfp_t gfp_mask,
 	return try_charge_memcg(memcg, gfp_mask, nr_pages);
 }
 
-static inline void cancel_charge(struct mem_cgroup *memcg, unsigned int nr_pages)
+/**
+ * mem_cgroup_cancel_charge() - cancel an uncommitted try_charge() call.
+ * @memcg: memcg previously charged.
+ * @nr_pages: number of pages previously charged.
+ */
+void mem_cgroup_cancel_charge(struct mem_cgroup *memcg, unsigned int nr_pages)
 {
 	if (mem_cgroup_is_root(memcg))
 		return;
@@ -2898,6 +2924,22 @@ static void commit_charge(struct folio *folio, struct mem_cgroup *memcg)
 	folio->memcg_data = (unsigned long)memcg;
 }
 
+/**
+ * mem_cgroup_commit_charge - commit a previously successful try_charge().
+ * @folio: folio to commit the charge to.
+ * @memcg: memcg previously charged.
+ */
+void mem_cgroup_commit_charge(struct folio *folio, struct mem_cgroup *memcg)
+{
+	css_get(&memcg->css);
+	commit_charge(folio, memcg);
+
+	local_irq_disable();
+	mem_cgroup_charge_statistics(memcg, folio_nr_pages(folio));
+	memcg_check_events(memcg, folio_nid(folio));
+	local_irq_enable();
+}
+
 #ifdef CONFIG_MEMCG_KMEM
 /*
  * The allocated objcg pointers array is not accounted directly.
@@ -6105,7 +6147,7 @@ static void __mem_cgroup_clear_mc(void)
 
 	/* we must uncharge all the leftover precharges from mc.to */
 	if (mc.precharge) {
-		cancel_charge(mc.to, mc.precharge);
+		mem_cgroup_cancel_charge(mc.to, mc.precharge);
 		mc.precharge = 0;
 	}
 	/*
@@ -6113,7 +6155,7 @@ static void __mem_cgroup_clear_mc(void)
 	 * we must uncharge here.
 	 */
 	if (mc.moved_charge) {
-		cancel_charge(mc.from, mc.moved_charge);
+		mem_cgroup_cancel_charge(mc.from, mc.moved_charge);
 		mc.moved_charge = 0;
 	}
 	/* we must fixup refcnts and charges */
@@ -7020,20 +7062,13 @@ void mem_cgroup_calculate_protection(struct mem_cgroup *root,
 static int charge_memcg(struct folio *folio, struct mem_cgroup *memcg,
 			gfp_t gfp)
 {
-	long nr_pages = folio_nr_pages(folio);
 	int ret;
 
-	ret = try_charge(memcg, gfp, nr_pages);
+	ret = try_charge(memcg, gfp, folio_nr_pages(folio));
 	if (ret)
 		goto out;
 
-	css_get(&memcg->css);
-	commit_charge(folio, memcg);
-
-	local_irq_disable();
-	mem_cgroup_charge_statistics(memcg, nr_pages);
-	memcg_check_events(memcg, folio_nid(folio));
-	local_irq_enable();
+	mem_cgroup_commit_charge(folio, memcg);
 out:
 	return ret;
 }
-- 
2.34.1

[PATCH v4 2/4] memcontrol: only transfer the memcg data for migration

[Nhat Pham]

For most migration use cases, only transfer the memcg data from the old
folio to the new folio, and clear the old folio's memcg data. No
charging and uncharging will be done.

This shaves off some work on the migration path, and avoids the
temporary double charging of a folio during its migration.

The only exception is replace_page_cache_folio(), which will use the old
mem_cgroup_migrate() (now renamed to mem_cgroup_replace_folio). In that
context, the isolation of the old page isn't quite as thorough as with
migration, so we cannot use our new implementation directly.

This patch is the result of the following discussion on the new hugetlb
memcg accounting behavior:

https://lore.kernel.org/lkml/20231003171329.GB314430@monkey/

Suggested-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
---
 include/linux/memcontrol.h |  7 +++++++
 mm/filemap.c               |  2 +-
 mm/memcontrol.c            | 40 +++++++++++++++++++++++++++++++++++---
 3 files changed, 45 insertions(+), 4 deletions(-)

diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index 42bf7e9b1a2f..5daf14da3759 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -708,6 +708,8 @@ static inline void mem_cgroup_uncharge_list(struct list_head *page_list)
 
 void mem_cgroup_cancel_charge(struct mem_cgroup *memcg, unsigned int nr_pages);
 
+void mem_cgroup_replace_folio(struct folio *old, struct folio *new);
+
 void mem_cgroup_migrate(struct folio *old, struct folio *new);
 
 /**
@@ -1285,6 +1287,11 @@ static inline void mem_cgroup_cancel_charge(struct mem_cgroup *memcg,
 {
 }
 
+static inline void mem_cgroup_replace_folio(struct folio *old,
+		struct folio *new)
+{
+}
+
 static inline void mem_cgroup_migrate(struct folio *old, struct folio *new)
 {
 }
diff --git a/mm/filemap.c b/mm/filemap.c
index 9481ffaf24e6..673745219c82 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -819,7 +819,7 @@ void replace_page_cache_folio(struct folio *old, struct folio *new)
 	new->mapping = mapping;
 	new->index = offset;
 
-	mem_cgroup_migrate(old, new);
+	mem_cgroup_replace_folio(old, new);
 
 	xas_lock_irq(&xas);
 	xas_store(&xas, new);
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 0219befeae38..b9c479d768e2 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -7281,16 +7281,17 @@ void __mem_cgroup_uncharge_list(struct list_head *page_list)
 }
 
 /**
- * mem_cgroup_migrate - Charge a folio's replacement.
+ * mem_cgroup_replace_folio - Charge a folio's replacement.
  * @old: Currently circulating folio.
  * @new: Replacement folio.
  *
  * Charge @new as a replacement folio for @old. @old will
- * be uncharged upon free.
+ * be uncharged upon free. This is only used by the page cache
+ * (in replace_page_cache_folio()).
  *
  * Both folios must be locked, @new->mapping must be set up.
  */
-void mem_cgroup_migrate(struct folio *old, struct folio *new)
+void mem_cgroup_replace_folio(struct folio *old, struct folio *new)
 {
 	struct mem_cgroup *memcg;
 	long nr_pages = folio_nr_pages(new);
@@ -7329,6 +7330,39 @@ void mem_cgroup_migrate(struct folio *old, struct folio *new)
 	local_irq_restore(flags);
 }
 
+/**
+ * mem_cgroup_migrate - Transfer the memcg data from the old to the new folio.
+ * @old: Currently circulating folio.
+ * @new: Replacement folio.
+ *
+ * Transfer the memcg data from the old folio to the new folio for migration.
+ * The old folio's data info will be cleared. Note that the memory counters
+ * will remain unchanged throughout the process.
+ *
+ * Both folios must be locked, @new->mapping must be set up.
+ */
+void mem_cgroup_migrate(struct folio *old, struct folio *new)
+{
+	struct mem_cgroup *memcg;
+
+	VM_BUG_ON_FOLIO(!folio_test_locked(old), old);
+	VM_BUG_ON_FOLIO(!folio_test_locked(new), new);
+	VM_BUG_ON_FOLIO(folio_test_anon(old) != folio_test_anon(new), new);
+	VM_BUG_ON_FOLIO(folio_nr_pages(old) != folio_nr_pages(new), new);
+
+	if (mem_cgroup_disabled())
+		return;
+
+	memcg = folio_memcg(old);
+	VM_WARN_ON_ONCE_FOLIO(!memcg, old);
+	if (!memcg)
+		return;
+
+	/* Transfer the charge and the css ref */
+	commit_charge(new, memcg);
+	old->memcg_data = 0;
+}
+
 DEFINE_STATIC_KEY_FALSE(memcg_sockets_enabled_key);
 EXPORT_SYMBOL(memcg_sockets_enabled_key);
 
-- 
2.34.1

[PATCH v4 3/4] hugetlb: memcg: account hugetlb-backed memory in memory controller

[Nhat Pham]

Currently, hugetlb memory usage is not acounted for in the memory
controller, which could lead to memory overprotection for cgroups with
hugetlb-backed memory. This has been observed in our production system.

For instance, here is one of our usecases: suppose there are two 32G
containers. The machine is booted with hugetlb_cma=6G, and each
container may or may not use up to 3 gigantic page, depending on the
workload within it. The rest is anon, cache, slab, etc. We can set the
hugetlb cgroup limit of each cgroup to 3G to enforce hugetlb fairness.
But it is very difficult to configure memory.max to keep overall
consumption, including anon, cache, slab etc. fair.

What we have had to resort to is to constantly poll hugetlb usage and
readjust memory.max. Similar procedure is done to other memory limits
(memory.low for e.g). However, this is rather cumbersome and buggy.
Furthermore, when there is a delay in memory limits correction, (for e.g
when hugetlb usage changes within consecutive runs of the userspace
agent), the system could be in an over/underprotected state.

This patch rectifies this issue by charging the memcg when the hugetlb
folio is utilized, and uncharging when the folio is freed (analogous to
the hugetlb controller). Note that we do not charge when the folio is
allocated to the hugetlb pool, because at this point it is not owned by
any memcg.

Some caveats to consider:
  * This feature is only available on cgroup v2.
  * There is no hugetlb pool management involved in the memory
    controller. As stated above, hugetlb folios are only charged towards
    the memory controller when it is used. Host overcommit management
    has to consider it when configuring hard limits.
  * Failure to charge towards the memcg results in SIGBUS. This could
    happen even if the hugetlb pool still has pages (but the cgroup
    limit is hit and reclaim attempt fails).
  * When this feature is enabled, hugetlb pages contribute to memory
    reclaim protection. low, min limits tuning must take into account
    hugetlb memory.
  * Hugetlb pages utilized while this option is not selected will not
    be tracked by the memory controller (even if cgroup v2 is remounted
    later on).

Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
---
 Documentation/admin-guide/cgroup-v2.rst | 29 +++++++++++++++++
 include/linux/cgroup-defs.h             |  5 +++
 include/linux/memcontrol.h              |  9 ++++++
 kernel/cgroup/cgroup.c                  | 15 ++++++++-
 mm/hugetlb.c                            | 35 ++++++++++++++++-----
 mm/memcontrol.c                         | 42 ++++++++++++++++++++++++-
 mm/migrate.c                            |  3 +-
 7 files changed, 127 insertions(+), 11 deletions(-)

diff --git a/Documentation/admin-guide/cgroup-v2.rst b/Documentation/admin-guide/cgroup-v2.rst
index 622a7f28db1f..606b2e0eac4b 100644
--- a/Documentation/admin-guide/cgroup-v2.rst
+++ b/Documentation/admin-guide/cgroup-v2.rst
@@ -210,6 +210,35 @@ cgroup v2 currently supports the following mount options.
         relying on the original semantics (e.g. specifying bogusly
         high 'bypass' protection values at higher tree levels).
 
+  memory_hugetlb_accounting
+        Count HugeTLB memory usage towards the cgroup's overall
+        memory usage for the memory controller (for the purpose of
+        statistics reporting and memory protetion). This is a new
+        behavior that could regress existing setups, so it must be
+        explicitly opted in with this mount option.
+
+        A few caveats to keep in mind:
+
+        * There is no HugeTLB pool management involved in the memory
+          controller. The pre-allocated pool does not belong to anyone.
+          Specifically, when a new HugeTLB folio is allocated to
+          the pool, it is not accounted for from the perspective of the
+          memory controller. It is only charged to a cgroup when it is
+          actually used (for e.g at page fault time). Host memory
+          overcommit management has to consider this when configuring
+          hard limits. In general, HugeTLB pool management should be
+          done via other mechanisms (such as the HugeTLB controller).
+        * Failure to charge a HugeTLB folio to the memory controller
+          results in SIGBUS. This could happen even if the HugeTLB pool
+          still has pages available (but the cgroup limit is hit and
+          reclaim attempt fails).
+        * Charging HugeTLB memory towards the memory controller affects
+          memory protection and reclaim dynamics. Any userspace tuning
+          (of low, min limits for e.g) needs to take this into account.
+        * HugeTLB pages utilized while this option is not selected
+          will not be tracked by the memory controller (even if cgroup
+          v2 is remounted later on).
+
 
 Organizing Processes and Threads
 --------------------------------
diff --git a/include/linux/cgroup-defs.h b/include/linux/cgroup-defs.h
index f1b3151ac30b..8641f4320c98 100644
--- a/include/linux/cgroup-defs.h
+++ b/include/linux/cgroup-defs.h
@@ -115,6 +115,11 @@ enum {
 	 * Enable recursive subtree protection
 	 */
 	CGRP_ROOT_MEMORY_RECURSIVE_PROT = (1 << 18),
+
+	/*
+	 * Enable hugetlb accounting for the memory controller.
+	 */
+	 CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING = (1 << 19),
 };
 
 /* cftype->flags */
diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index 5daf14da3759..e3eaa123256b 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -679,6 +679,9 @@ static inline int mem_cgroup_charge(struct folio *folio, struct mm_struct *mm,
 	return __mem_cgroup_charge(folio, mm, gfp);
 }
 
+int mem_cgroup_hugetlb_try_charge(struct mem_cgroup *memcg, gfp_t gfp,
+		long nr_pages);
+
 int mem_cgroup_swapin_charge_folio(struct folio *folio, struct mm_struct *mm,
 				  gfp_t gfp, swp_entry_t entry);
 void mem_cgroup_swapin_uncharge_swap(swp_entry_t entry);
@@ -1264,6 +1267,12 @@ static inline int mem_cgroup_charge(struct folio *folio,
 	return 0;
 }
 
+static inline int mem_cgroup_hugetlb_try_charge(struct mem_cgroup *memcg,
+		gfp_t gfp, long nr_pages)
+{
+	return 0;
+}
+
 static inline int mem_cgroup_swapin_charge_folio(struct folio *folio,
 			struct mm_struct *mm, gfp_t gfp, swp_entry_t entry)
 {
diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
index 1fb7f562289d..f11488b18ceb 100644
--- a/kernel/cgroup/cgroup.c
+++ b/kernel/cgroup/cgroup.c
@@ -1902,6 +1902,7 @@ enum cgroup2_param {
 	Opt_favordynmods,
 	Opt_memory_localevents,
 	Opt_memory_recursiveprot,
+	Opt_memory_hugetlb_accounting,
 	nr__cgroup2_params
 };
 
@@ -1910,6 +1911,7 @@ static const struct fs_parameter_spec cgroup2_fs_parameters[] = {
 	fsparam_flag("favordynmods",		Opt_favordynmods),
 	fsparam_flag("memory_localevents",	Opt_memory_localevents),
 	fsparam_flag("memory_recursiveprot",	Opt_memory_recursiveprot),
+	fsparam_flag("memory_hugetlb_accounting", Opt_memory_hugetlb_accounting),
 	{}
 };
 
@@ -1936,6 +1938,9 @@ static int cgroup2_parse_param(struct fs_context *fc, struct fs_parameter *param
 	case Opt_memory_recursiveprot:
 		ctx->flags |= CGRP_ROOT_MEMORY_RECURSIVE_PROT;
 		return 0;
+	case Opt_memory_hugetlb_accounting:
+		ctx->flags |= CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING;
+		return 0;
 	}
 	return -EINVAL;
 }
@@ -1960,6 +1965,11 @@ static void apply_cgroup_root_flags(unsigned int root_flags)
 			cgrp_dfl_root.flags |= CGRP_ROOT_MEMORY_RECURSIVE_PROT;
 		else
 			cgrp_dfl_root.flags &= ~CGRP_ROOT_MEMORY_RECURSIVE_PROT;
+
+		if (root_flags & CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING)
+			cgrp_dfl_root.flags |= CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING;
+		else
+			cgrp_dfl_root.flags &= ~CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING;
 	}
 }
 
@@ -1973,6 +1983,8 @@ static int cgroup_show_options(struct seq_file *seq, struct kernfs_root *kf_root
 		seq_puts(seq, ",memory_localevents");
 	if (cgrp_dfl_root.flags & CGRP_ROOT_MEMORY_RECURSIVE_PROT)
 		seq_puts(seq, ",memory_recursiveprot");
+	if (cgrp_dfl_root.flags & CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING)
+		seq_puts(seq, ",memory_hugetlb_accounting");
 	return 0;
 }
 
@@ -7050,7 +7062,8 @@ static ssize_t features_show(struct kobject *kobj, struct kobj_attribute *attr,
 			"nsdelegate\n"
 			"favordynmods\n"
 			"memory_localevents\n"
-			"memory_recursiveprot\n");
+			"memory_recursiveprot\n"
+			"memory_hugetlb_accounting\n");
 }
 static struct kobj_attribute cgroup_features_attr = __ATTR_RO(features);
 
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index de220e3ff8be..74472e911b0a 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -1902,6 +1902,7 @@ void free_huge_folio(struct folio *folio)
 				     pages_per_huge_page(h), folio);
 	hugetlb_cgroup_uncharge_folio_rsvd(hstate_index(h),
 					  pages_per_huge_page(h), folio);
+	mem_cgroup_uncharge(folio);
 	if (restore_reserve)
 		h->resv_huge_pages++;
 
@@ -3009,11 +3010,20 @@ struct folio *alloc_hugetlb_folio(struct vm_area_struct *vma,
 	struct hugepage_subpool *spool = subpool_vma(vma);
 	struct hstate *h = hstate_vma(vma);
 	struct folio *folio;
-	long map_chg, map_commit;
+	long map_chg, map_commit, nr_pages = pages_per_huge_page(h);
 	long gbl_chg;
-	int ret, idx;
+	int memcg_charge_ret, ret, idx;
 	struct hugetlb_cgroup *h_cg = NULL;
+	struct mem_cgroup *memcg;
 	bool deferred_reserve;
+	gfp_t gfp = htlb_alloc_mask(h) | __GFP_RETRY_MAYFAIL;
+
+	memcg = get_mem_cgroup_from_current();
+	memcg_charge_ret = mem_cgroup_hugetlb_try_charge(memcg, gfp, nr_pages);
+	if (memcg_charge_ret == -ENOMEM) {
+		mem_cgroup_put(memcg);
+		return ERR_PTR(-ENOMEM);
+	}
 
 	idx = hstate_index(h);
 	/*
@@ -3022,8 +3032,12 @@ struct folio *alloc_hugetlb_folio(struct vm_area_struct *vma,
 	 * code of zero indicates a reservation exists (no change).
 	 */
 	map_chg = gbl_chg = vma_needs_reservation(h, vma, addr);
-	if (map_chg < 0)
+	if (map_chg < 0) {
+		if (!memcg_charge_ret)
+			mem_cgroup_cancel_charge(memcg, nr_pages);
+		mem_cgroup_put(memcg);
 		return ERR_PTR(-ENOMEM);
+	}
 
 	/*
 	 * Processes that did not create the mapping will have no
@@ -3034,10 +3048,8 @@ struct folio *alloc_hugetlb_folio(struct vm_area_struct *vma,
 	 */
 	if (map_chg || avoid_reserve) {
 		gbl_chg = hugepage_subpool_get_pages(spool, 1);
-		if (gbl_chg < 0) {
-			vma_end_reservation(h, vma, addr);
-			return ERR_PTR(-ENOSPC);
-		}
+		if (gbl_chg < 0)
+			goto out_end_reservation;
 
 		/*
 		 * Even though there was no reservation in the region/reserve
@@ -3119,6 +3131,11 @@ struct folio *alloc_hugetlb_folio(struct vm_area_struct *vma,
 			hugetlb_cgroup_uncharge_folio_rsvd(hstate_index(h),
 					pages_per_huge_page(h), folio);
 	}
+
+	if (!memcg_charge_ret)
+		mem_cgroup_commit_charge(folio, memcg);
+	mem_cgroup_put(memcg);
+
 	return folio;
 
 out_uncharge_cgroup:
@@ -3130,7 +3147,11 @@ struct folio *alloc_hugetlb_folio(struct vm_area_struct *vma,
 out_subpool_put:
 	if (map_chg || avoid_reserve)
 		hugepage_subpool_put_pages(spool, 1);
+out_end_reservation:
 	vma_end_reservation(h, vma, addr);
+	if (!memcg_charge_ret)
+		mem_cgroup_cancel_charge(memcg, nr_pages);
+	mem_cgroup_put(memcg);
 	return ERR_PTR(-ENOSPC);
 }
 
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index b9c479d768e2..a3adfecf5977 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -7085,6 +7085,41 @@ int __mem_cgroup_charge(struct folio *folio, struct mm_struct *mm, gfp_t gfp)
 	return ret;
 }
 
+/**
+ * mem_cgroup_hugetlb_try_charge - try to charge the memcg for a hugetlb folio
+ * @memcg: memcg to charge.
+ * @gfp: reclaim mode.
+ * @nr_pages: number of pages to charge.
+ *
+ * This function is called when allocating a huge page folio to determine if
+ * the memcg has the capacity for it. It does not commit the charge yet,
+ * as the hugetlb folio itself has not been obtained from the hugetlb pool.
+ *
+ * Once we have obtained the hugetlb folio, we can call
+ * mem_cgroup_commit_charge() to commit the charge. If we fail to obtain the
+ * folio, we should instead call mem_cgroup_cancel_charge() to undo the effect
+ * of try_charge().
+ *
+ * Returns 0 on success. Otherwise, an error code is returned.
+ */
+int mem_cgroup_hugetlb_try_charge(struct mem_cgroup *memcg, gfp_t gfp,
+			long nr_pages)
+{
+	/*
+	 * If hugetlb memcg charging is not enabled, do not fail hugetlb allocation,
+	 * but do not attempt to commit charge later (or cancel on error) either.
+	 */
+	if (mem_cgroup_disabled() || !memcg ||
+		!cgroup_subsys_on_dfl(memory_cgrp_subsys) ||
+		!(cgrp_dfl_root.flags & CGRP_ROOT_MEMORY_HUGETLB_ACCOUNTING))
+		return -EOPNOTSUPP;
+
+	if (try_charge(memcg, gfp, nr_pages))
+		return -ENOMEM;
+
+	return 0;
+}
+
 /**
  * mem_cgroup_swapin_charge_folio - Charge a newly allocated folio for swapin.
  * @folio: folio to charge.
@@ -7354,7 +7389,12 @@ void mem_cgroup_migrate(struct folio *old, struct folio *new)
 		return;
 
 	memcg = folio_memcg(old);
-	VM_WARN_ON_ONCE_FOLIO(!memcg, old);
+	/*
+	 * Note that it is normal to see !memcg for a hugetlb folio.
+	 * For e.g, itt could have been allocated when memory_hugetlb_accounting
+	 * was not selected.
+	 */
+	VM_WARN_ON_ONCE_FOLIO(!folio_test_hugetlb(old) && !memcg, old);
 	if (!memcg)
 		return;
 
diff --git a/mm/migrate.c b/mm/migrate.c
index 7d1804c4a5d9..6034c7ed1d65 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -633,8 +633,7 @@ void folio_migrate_flags(struct folio *newfolio, struct folio *folio)
 
 	folio_copy_owner(newfolio, folio);
 
-	if (!folio_test_hugetlb(folio))
-		mem_cgroup_migrate(folio, newfolio);
+	mem_cgroup_migrate(folio, newfolio);
 }
 EXPORT_SYMBOL(folio_migrate_flags);
 
-- 
2.34.1

[PATCH v4 4/4] selftests: add a selftest to verify hugetlb usage in memcg

[Nhat Pham]

This patch add a new kselftest to demonstrate and verify the new
hugetlb memcg accounting behavior.

Signed-off-by: Nhat Pham <nphamcs@gmail.com>
---
 MAINTAINERS                                   |   2 +
 tools/testing/selftests/cgroup/.gitignore     |   1 +
 tools/testing/selftests/cgroup/Makefile       |   2 +
 .../selftests/cgroup/test_hugetlb_memcg.c     | 234 ++++++++++++++++++
 4 files changed, 239 insertions(+)
 create mode 100644 tools/testing/selftests/cgroup/test_hugetlb_memcg.c

diff --git a/MAINTAINERS b/MAINTAINERS
index bf0f54c24f81..ce9f40bcc2ba 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -5269,6 +5269,7 @@ S:	Maintained
 F:	mm/memcontrol.c
 F:	mm/swap_cgroup.c
 F:	tools/testing/selftests/cgroup/memcg_protection.m
+F:	tools/testing/selftests/cgroup/test_hugetlb_memcg.c
 F:	tools/testing/selftests/cgroup/test_kmem.c
 F:	tools/testing/selftests/cgroup/test_memcontrol.c
 
@@ -9652,6 +9653,7 @@ F:	include/linux/hugetlb.h
 F:	mm/hugetlb.c
 F:	mm/hugetlb_vmemmap.c
 F:	mm/hugetlb_vmemmap.h
+F:	tools/testing/selftests/cgroup/test_hugetlb_memcg.c
 
 HVA ST MEDIA DRIVER
 M:	Jean-Christophe Trotin <jean-christophe.trotin@foss.st.com>
diff --git a/tools/testing/selftests/cgroup/.gitignore b/tools/testing/selftests/cgroup/.gitignore
index af8c3f30b9c1..2732e0b29271 100644
--- a/tools/testing/selftests/cgroup/.gitignore
+++ b/tools/testing/selftests/cgroup/.gitignore
@@ -7,4 +7,5 @@ test_kill
 test_cpu
 test_cpuset
 test_zswap
+test_hugetlb_memcg
 wait_inotify
diff --git a/tools/testing/selftests/cgroup/Makefile b/tools/testing/selftests/cgroup/Makefile
index c27f05f6ce9b..00b441928909 100644
--- a/tools/testing/selftests/cgroup/Makefile
+++ b/tools/testing/selftests/cgroup/Makefile
@@ -14,6 +14,7 @@ TEST_GEN_PROGS += test_kill
 TEST_GEN_PROGS += test_cpu
 TEST_GEN_PROGS += test_cpuset
 TEST_GEN_PROGS += test_zswap
+TEST_GEN_PROGS += test_hugetlb_memcg
 
 LOCAL_HDRS += $(selfdir)/clone3/clone3_selftests.h $(selfdir)/pidfd/pidfd.h
 
@@ -27,3 +28,4 @@ $(OUTPUT)/test_kill: cgroup_util.c
 $(OUTPUT)/test_cpu: cgroup_util.c
 $(OUTPUT)/test_cpuset: cgroup_util.c
 $(OUTPUT)/test_zswap: cgroup_util.c
+$(OUTPUT)/test_hugetlb_memcg: cgroup_util.c
diff --git a/tools/testing/selftests/cgroup/test_hugetlb_memcg.c b/tools/testing/selftests/cgroup/test_hugetlb_memcg.c
new file mode 100644
index 000000000000..f0fefeb4cc24
--- /dev/null
+++ b/tools/testing/selftests/cgroup/test_hugetlb_memcg.c
@@ -0,0 +1,234 @@
+// SPDX-License-Identifier: GPL-2.0
+#define _GNU_SOURCE
+
+#include <linux/limits.h>
+#include <sys/mman.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <fcntl.h>
+#include "../kselftest.h"
+#include "cgroup_util.h"
+
+#define ADDR ((void *)(0x0UL))
+#define FLAGS (MAP_PRIVATE | MAP_ANONYMOUS | MAP_HUGETLB)
+/* mapping 8 MBs == 4 hugepages */
+#define LENGTH (8UL*1024*1024)
+#define PROTECTION (PROT_READ | PROT_WRITE)
+
+/* borrowed from mm/hmm-tests.c */
+static long get_hugepage_size(void)
+{
+	int fd;
+	char buf[2048];
+	int len;
+	char *p, *q, *path = "/proc/meminfo", *tag = "Hugepagesize:";
+	long val;
+
+	fd = open(path, O_RDONLY);
+	if (fd < 0) {
+		/* Error opening the file */
+		return -1;
+	}
+
+	len = read(fd, buf, sizeof(buf));
+	close(fd);
+	if (len < 0) {
+		/* Error in reading the file */
+		return -1;
+	}
+	if (len == sizeof(buf)) {
+		/* Error file is too large */
+		return -1;
+	}
+	buf[len] = '\0';
+
+	/* Search for a tag if provided */
+	if (tag) {
+		p = strstr(buf, tag);
+		if (!p)
+			return -1; /* looks like the line we want isn't there */
+		p += strlen(tag);
+	} else
+		p = buf;
+
+	val = strtol(p, &q, 0);
+	if (*q != ' ') {
+		/* Error parsing the file */
+		return -1;
+	}
+
+	return val;
+}
+
+static int set_file(const char *path, long value)
+{
+	FILE *file;
+	int ret;
+
+	file = fopen(path, "w");
+	if (!file)
+		return -1;
+	ret = fprintf(file, "%ld\n", value);
+	fclose(file);
+	return ret;
+}
+
+static int set_nr_hugepages(long value)
+{
+	return set_file("/proc/sys/vm/nr_hugepages", value);
+}
+
+static unsigned int check_first(char *addr)
+{
+	return *(unsigned int *)addr;
+}
+
+static void write_data(char *addr)
+{
+	unsigned long i;
+
+	for (i = 0; i < LENGTH; i++)
+		*(addr + i) = (char)i;
+}
+
+static int hugetlb_test_program(const char *cgroup, void *arg)
+{
+	char *test_group = (char *)arg;
+	void *addr;
+	long old_current, expected_current, current;
+	int ret = EXIT_FAILURE;
+
+	old_current = cg_read_long(test_group, "memory.current");
+	set_nr_hugepages(20);
+	current = cg_read_long(test_group, "memory.current");
+	if (current - old_current >= MB(2)) {
+		ksft_print_msg(
+			"setting nr_hugepages should not increase hugepage usage.\n");
+		ksft_print_msg("before: %ld, after: %ld\n", old_current, current);
+		return EXIT_FAILURE;
+	}
+
+	addr = mmap(ADDR, LENGTH, PROTECTION, FLAGS, 0, 0);
+	if (addr == MAP_FAILED) {
+		ksft_print_msg("fail to mmap.\n");
+		return EXIT_FAILURE;
+	}
+	current = cg_read_long(test_group, "memory.current");
+	if (current - old_current >= MB(2)) {
+		ksft_print_msg("mmap should not increase hugepage usage.\n");
+		ksft_print_msg("before: %ld, after: %ld\n", old_current, current);
+		goto out_failed_munmap;
+	}
+	old_current = current;
+
+	/* read the first page */
+	check_first(addr);
+	expected_current = old_current + MB(2);
+	current = cg_read_long(test_group, "memory.current");
+	if (!values_close(expected_current, current, 5)) {
+		ksft_print_msg("memory usage should increase by around 2MB.\n");
+		ksft_print_msg(
+			"expected memory: %ld, actual memory: %ld\n",
+			expected_current, current);
+		goto out_failed_munmap;
+	}
+
+	/* write to the whole range */
+	write_data(addr);
+	current = cg_read_long(test_group, "memory.current");
+	expected_current = old_current + MB(8);
+	if (!values_close(expected_current, current, 5)) {
+		ksft_print_msg("memory usage should increase by around 8MB.\n");
+		ksft_print_msg(
+			"expected memory: %ld, actual memory: %ld\n",
+			expected_current, current);
+		goto out_failed_munmap;
+	}
+
+	/* unmap the whole range */
+	munmap(addr, LENGTH);
+	current = cg_read_long(test_group, "memory.current");
+	expected_current = old_current;
+	if (!values_close(expected_current, current, 5)) {
+		ksft_print_msg("memory usage should go back down.\n");
+		ksft_print_msg(
+			"expected memory: %ld, actual memory: %ld\n",
+			expected_current, current);
+		return ret;
+	}
+
+	ret = EXIT_SUCCESS;
+	return ret;
+
+out_failed_munmap:
+	munmap(addr, LENGTH);
+	return ret;
+}
+
+static int test_hugetlb_memcg(char *root)
+{
+	int ret = KSFT_FAIL;
+	char *test_group;
+
+	test_group = cg_name(root, "hugetlb_memcg_test");
+	if (!test_group || cg_create(test_group)) {
+		ksft_print_msg("fail to create cgroup.\n");
+		goto out;
+	}
+
+	if (cg_write(test_group, "memory.max", "100M")) {
+		ksft_print_msg("fail to set cgroup memory limit.\n");
+		goto out;
+	}
+
+	/* disable swap */
+	if (cg_write(test_group, "memory.swap.max", "0")) {
+		ksft_print_msg("fail to disable swap.\n");
+		goto out;
+	}
+
+	if (!cg_run(test_group, hugetlb_test_program, (void *)test_group))
+		ret = KSFT_PASS;
+out:
+	cg_destroy(test_group);
+	free(test_group);
+	return ret;
+}
+
+int main(int argc, char **argv)
+{
+	char root[PATH_MAX];
+	int ret = EXIT_SUCCESS, has_memory_hugetlb_acc;
+
+	has_memory_hugetlb_acc = proc_mount_contains("memory_hugetlb_accounting");
+	if (has_memory_hugetlb_acc < 0)
+		ksft_exit_skip("Failed to query cgroup mount option\n");
+	else if (!has_memory_hugetlb_acc)
+		ksft_exit_skip("memory hugetlb accounting is disabled\n");
+
+	/* Unit is kB! */
+	if (get_hugepage_size() != 2048) {
+		ksft_print_msg("test_hugetlb_memcg requires 2MB hugepages\n");
+		ksft_test_result_skip("test_hugetlb_memcg\n");
+		return ret;
+	}
+
+	if (cg_find_unified_root(root, sizeof(root)))
+		ksft_exit_skip("cgroup v2 isn't mounted\n");
+
+	switch (test_hugetlb_memcg(root)) {
+	case KSFT_PASS:
+		ksft_test_result_pass("test_hugetlb_memcg\n");
+		break;
+	case KSFT_SKIP:
+		ksft_test_result_skip("test_hugetlb_memcg\n");
+		break;
+	default:
+		ret = EXIT_FAILURE;
+		ksft_test_result_fail("test_hugetlb_memcg\n");
+		break;
+	}
+
+	return ret;
+}
-- 
2.34.1