[PATCH v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs

[PATCH v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs

[Faiyaz Mohammed]

alloc_calls and free_calls implementation in sysfs have two issues,
one is PAGE_SIZE limitiation of sysfs and other is it does not adhere
to "one value per file" rule.

To overcome this issues, move the alloc_calls and free_calls implemeation
to debugfs.

Debugfs cache will be created if SLAB_STORE_USER flag is set.

Rename the alloc_calls/free_calls to alloc_traces/free_traces,
to be inline with what it does.

Signed-off-by: Faiyaz Mohammed <faiyazm@codeaurora.org>
---
changes in v8:
	- Debugfs cache file will not be created if SLAB_STORE_USER flag is not set.
	- Make sure the debugfs slab release happen before slab kmem cahce release.
	
changes in V7:
	- https://lore.kernel.org/linux-mm/1621928285-751-1-git-send-email-faiyazm@codeaurora.org/

changes in v6:
	- https://lore.kernel.org/linux-mm/1621341949-26762-1-git-send-email-faiyazm@codeaurora.org/

changes in v5:
	- https://lore.kernel.org/linux-mm/1620296523-21922-1-git-send-email-faiyazm@codeaurora.org/

changes in v4:
	- https://lore.kernel.org/linux-mm/1618583239-18124-1-git-send-email-faiyazm@codeaurora.org/

changes in v3:
	- https://lore.kernel.org/linux-mm/1617712064-12264-1-git-send-email-faiyazm@codeaurora.org/

changes in v2:
	- https://lore.kernel.org/linux-mm/3ac1d3e6-6207-96ad-16a1-0f5139d8b2b5@codeaurora.org/

changes in v1
	- https://lore.kernel.org/linux-mm/1610443287-23933-1-git-send-email-faiyazm@codeaurora.org/

 include/linux/slab_def.h |   2 +
 include/linux/slub_def.h |   8 ++
 mm/slab_common.c         |   3 +
 mm/slub.c                | 306 +++++++++++++++++++++++++++++++++--------------
 4 files changed, 226 insertions(+), 93 deletions(-)

diff --git a/include/linux/slab_def.h b/include/linux/slab_def.h
index 3aa5e1e..02ca311 100644
--- a/include/linux/slab_def.h
+++ b/include/linux/slab_def.h
@@ -120,4 +120,6 @@ static inline int objs_per_slab_page(const struct kmem_cache *cache,
 	return cache->num;
 }
 
+static inline void debugfs_slab_release(struct kmem_cache *s) { }
+
 #endif	/* _LINUX_SLAB_DEF_H */
diff --git a/include/linux/slub_def.h b/include/linux/slub_def.h
index dcde82a..90d659a 100644
--- a/include/linux/slub_def.h
+++ b/include/linux/slub_def.h
@@ -159,6 +159,14 @@ static inline void sysfs_slab_release(struct kmem_cache *s)
 }
 #endif
 
+#if defined(CONFIG_DEBUG_FS) && defined(CONFIG_SLUB_DEBUG)
+void debugfs_slab_release(struct kmem_cache *);
+#else
+static inline void debugfs_slab_release(struct kmem_cache *s)
+{
+}
+#endif
+
 void object_err(struct kmem_cache *s, struct page *page,
 		u8 *object, char *reason);
 
diff --git a/mm/slab_common.c b/mm/slab_common.c
index a4a5714..ca39e84 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -449,6 +449,7 @@ static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
 	rcu_barrier();
 
 	list_for_each_entry_safe(s, s2, &to_destroy, list) {
+		debugfs_slab_release(s);
 		kfence_shutdown_cache(s);
 #ifdef SLAB_SUPPORTS_SYSFS
 		sysfs_slab_release(s);
@@ -469,6 +470,7 @@ static int shutdown_cache(struct kmem_cache *s)
 	list_del(&s->list);
 
 	if (s->flags & SLAB_TYPESAFE_BY_RCU) {
+		debugfs_slab_release(s);
 #ifdef SLAB_SUPPORTS_SYSFS
 		sysfs_slab_unlink(s);
 #endif
@@ -476,6 +478,7 @@ static int shutdown_cache(struct kmem_cache *s)
 		schedule_work(&slab_caches_to_rcu_destroy_work);
 	} else {
 		kfence_shutdown_cache(s);
+		debugfs_slab_release(s);
 #ifdef SLAB_SUPPORTS_SYSFS
 		sysfs_slab_unlink(s);
 		sysfs_slab_release(s);
diff --git a/mm/slub.c b/mm/slub.c
index 3f96e09..f7735c1 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -36,6 +36,7 @@
 #include <linux/memcontrol.h>
 #include <linux/random.h>
 
+#include <linux/debugfs.h>
 #include <trace/events/kmem.h>
 
 #include "internal.h"
@@ -225,6 +226,12 @@ static inline int sysfs_slab_alias(struct kmem_cache *s, const char *p)
 							{ return 0; }
 #endif
 
+#if defined(CONFIG_DEBUG_FS) && defined(CONFIG_SLUB_DEBUG)
+static void debugfs_slab_add(struct kmem_cache *);
+#else
+static inline void debugfs_slab_add(struct kmem_cache *s) { }
+#endif
+
 static inline void stat(const struct kmem_cache *s, enum stat_item si)
 {
 #ifdef CONFIG_SLUB_STATS
@@ -4546,6 +4553,9 @@ int __kmem_cache_create(struct kmem_cache *s, slab_flags_t flags)
 	if (err)
 		__kmem_cache_release(s);
 
+	if (s->flags & SLAB_STORE_USER)
+		debugfs_slab_add(s);
+
 	return err;
 }
 
@@ -4686,6 +4696,8 @@ static long validate_slab_cache(struct kmem_cache *s)
 
 	return count;
 }
+
+#ifdef CONFIG_DEBUG_FS
 /*
  * Generate lists of code addresses where slabcache objects are allocated
  * and freed.
@@ -4709,6 +4721,9 @@ struct loc_track {
 	struct location *loc;
 };
 
+static struct dentry *slab_debugfs_root;
+struct loc_track t = { 0, 0, NULL };
+
 static void free_loc_track(struct loc_track *t)
 {
 	if (t->max)
@@ -4825,82 +4840,7 @@ static void process_slab(struct loc_track *t, struct kmem_cache *s,
 			add_location(t, s, get_track(s, p, alloc));
 	put_map(map);
 }
-
-static int list_locations(struct kmem_cache *s, char *buf,
-			  enum track_item alloc)
-{
-	int len = 0;
-	unsigned long i;
-	struct loc_track t = { 0, 0, NULL };
-	int node;
-	struct kmem_cache_node *n;
-
-	if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location),
-			     GFP_KERNEL)) {
-		return sysfs_emit(buf, "Out of memory\n");
-	}
-	/* Push back cpu slabs */
-	flush_all(s);
-
-	for_each_kmem_cache_node(s, node, n) {
-		unsigned long flags;
-		struct page *page;
-
-		if (!atomic_long_read(&n->nr_slabs))
-			continue;
-
-		spin_lock_irqsave(&n->list_lock, flags);
-		list_for_each_entry(page, &n->partial, slab_list)
-			process_slab(&t, s, page, alloc);
-		list_for_each_entry(page, &n->full, slab_list)
-			process_slab(&t, s, page, alloc);
-		spin_unlock_irqrestore(&n->list_lock, flags);
-	}
-
-	for (i = 0; i < t.count; i++) {
-		struct location *l = &t.loc[i];
-
-		len += sysfs_emit_at(buf, len, "%7ld ", l->count);
-
-		if (l->addr)
-			len += sysfs_emit_at(buf, len, "%pS", (void *)l->addr);
-		else
-			len += sysfs_emit_at(buf, len, "<not-available>");
-
-		if (l->sum_time != l->min_time)
-			len += sysfs_emit_at(buf, len, " age=%ld/%ld/%ld",
-					     l->min_time,
-					     (long)div_u64(l->sum_time,
-							   l->count),
-					     l->max_time);
-		else
-			len += sysfs_emit_at(buf, len, " age=%ld", l->min_time);
-
-		if (l->min_pid != l->max_pid)
-			len += sysfs_emit_at(buf, len, " pid=%ld-%ld",
-					     l->min_pid, l->max_pid);
-		else
-			len += sysfs_emit_at(buf, len, " pid=%ld",
-					     l->min_pid);
-
-		if (num_online_cpus() > 1 &&
-		    !cpumask_empty(to_cpumask(l->cpus)))
-			len += sysfs_emit_at(buf, len, " cpus=%*pbl",
-					     cpumask_pr_args(to_cpumask(l->cpus)));
-
-		if (nr_online_nodes > 1 && !nodes_empty(l->nodes))
-			len += sysfs_emit_at(buf, len, " nodes=%*pbl",
-					     nodemask_pr_args(&l->nodes));
-
-		len += sysfs_emit_at(buf, len, "\n");
-	}
-
-	free_loc_track(&t);
-	if (!t.count)
-		len += sysfs_emit_at(buf, len, "No data\n");
-
-	return len;
-}
+#endif  /* CONFIG_DEBUG_FS   */
 #endif	/* CONFIG_SLUB_DEBUG */
 
 #ifdef SLUB_RESILIENCY_TEST
@@ -5350,21 +5290,6 @@ static ssize_t validate_store(struct kmem_cache *s,
 }
 SLAB_ATTR(validate);
 
-static ssize_t alloc_calls_show(struct kmem_cache *s, char *buf)
-{
-	if (!(s->flags & SLAB_STORE_USER))
-		return -ENOSYS;
-	return list_locations(s, buf, TRACK_ALLOC);
-}
-SLAB_ATTR_RO(alloc_calls);
-
-static ssize_t free_calls_show(struct kmem_cache *s, char *buf)
-{
-	if (!(s->flags & SLAB_STORE_USER))
-		return -ENOSYS;
-	return list_locations(s, buf, TRACK_FREE);
-}
-SLAB_ATTR_RO(free_calls);
 #endif /* CONFIG_SLUB_DEBUG */
 
 #ifdef CONFIG_FAILSLAB
@@ -5528,8 +5453,6 @@ static struct attribute *slab_attrs[] = {
 	&poison_attr.attr,
 	&store_user_attr.attr,
 	&validate_attr.attr,
-	&alloc_calls_attr.attr,
-	&free_calls_attr.attr,
 #endif
 #ifdef CONFIG_ZONE_DMA
 	&cache_dma_attr.attr,
@@ -5818,6 +5741,203 @@ static int __init slab_sysfs_init(void)
 __initcall(slab_sysfs_init);
 #endif /* CONFIG_SYSFS */
 
+#if defined(CONFIG_SLUB_DEBUG) && defined(CONFIG_DEBUG_FS)
+static int slab_debugfs_show(struct seq_file *seq, void *v)
+{
+
+	struct location *l;
+	unsigned int idx = *(unsigned int *)v;
+
+	if (idx < t.count) {
+		l = &t.loc[idx];
+
+		seq_printf(seq, "%7ld ", l->count);
+
+		if (l->addr)
+			seq_printf(seq, "%pS", (void *)l->addr);
+		else
+			seq_puts(seq, "<not-available>");
+
+		if (l->sum_time != l->min_time) {
+			seq_printf(seq, " age=%ld/%ld/%ld",
+				l->min_time,
+				(long)div_u64(l->sum_time, l->count),
+				l->max_time);
+		} else
+			seq_printf(seq, " age=%ld",
+				l->min_time);
+
+		if (l->min_pid != l->max_pid)
+			seq_printf(seq, " pid=%ld-%ld",
+				l->min_pid, l->max_pid);
+		else
+			seq_printf(seq, " pid=%ld",
+				l->min_pid);
+
+		if (num_online_cpus() > 1 &&
+				!cpumask_empty(to_cpumask(l->cpus)))
+			seq_printf(seq, " cpus=%*pbl",
+				 cpumask_pr_args(to_cpumask(l->cpus)));
+
+		if (nr_online_nodes > 1 && !nodes_empty(l->nodes))
+			seq_printf(seq, " nodes=%*pbl",
+				 nodemask_pr_args(&l->nodes));
+
+		seq_puts(seq, "\n");
+	}
+
+	if (t.count == 0)
+		seq_puts(seq, "No data\n");
+
+	return 0;
+}
+
+static void slab_debugfs_stop(struct seq_file *seq, void *v)
+{
+	if (!v && t.max) {
+		free_loc_track(&t);
+		t.max = 0;
+	}
+}
+
+static void *slab_debugfs_next(struct seq_file *seq, void *v, loff_t *ppos)
+{
+	loff_t *spos = v;
+
+	if (*ppos < t.count) {
+		*spos = *spos + 1;
+		*ppos = *spos;
+		return spos;
+	}
+
+	return NULL;
+}
+
+static void *slab_debugfs_start(struct seq_file *seq, loff_t *ppos)
+{
+	struct kmem_cache_node *n;
+	struct kmem_cache *s;
+	enum track_item alloc;
+	int node;
+	loff_t *spos = kmalloc(sizeof(loff_t), GFP_KERNEL);
+
+	s = seq->file->f_inode->i_private;
+
+	if (!spos)
+		return NULL;
+
+	if (*ppos == 0) {
+
+		t.count = 0;
+		t.max = 0;
+		t.loc = NULL;
+		if (strcmp(seq->file->f_path.dentry->d_name.name, "alloc_traces") == 0)
+			alloc =  TRACK_ALLOC;
+		else
+			alloc =  TRACK_FREE;
+
+		if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location),
+			     GFP_KERNEL)) {
+			seq_puts(seq, "Out of memory\n");
+			kfree(spos);
+			return ERR_PTR(-ENOMEM);
+		}
+		/* Push back cpu slabs */
+		flush_all(s);
+
+		for_each_kmem_cache_node(s, node, n) {
+			unsigned long flags;
+			struct page *page;
+
+			if (!atomic_long_read(&n->nr_slabs))
+				continue;
+
+			spin_lock_irqsave(&n->list_lock, flags);
+			list_for_each_entry(page, &n->partial, slab_list)
+				process_slab(&t, s, page, alloc);
+			list_for_each_entry(page, &n->full, slab_list)
+				process_slab(&t, s, page, alloc);
+			spin_unlock_irqrestore(&n->list_lock, flags);
+		}
+	}
+
+	if (*ppos < t.count) {
+		*spos = *ppos;
+		return spos;
+	}
+
+	kfree(spos);
+	return NULL;
+}
+
+static const struct seq_operations slab_debugfs_sops = {
+	.start  = slab_debugfs_start,
+	.next   = slab_debugfs_next,
+	.stop   = slab_debugfs_stop,
+	.show   = slab_debugfs_show
+};
+DEFINE_SEQ_ATTRIBUTE(slab_debugfs);
+
+static void debugfs_slab_add(struct kmem_cache *s)
+{
+	const char *name;
+	struct dentry *slab_cache_dir;
+	int unmergeable = slab_unmergeable(s);
+
+	if (unlikely(!slab_debugfs_root))
+		return;
+
+	if (!unmergeable && disable_higher_order_debug &&
+			(slub_debug & DEBUG_METADATA_FLAGS))
+		unmergeable = 1;
+
+	if (unmergeable) {
+		/*
+		 * Slabcache can never be merged so we can use the name proper.
+		 * This is typically the case for debug situations. In that
+		 * case we can catch duplicate names easily.
+		 */
+		slab_cache_dir = debugfs_lookup(s->name, slab_debugfs_root);
+		debugfs_remove_recursive(slab_cache_dir);
+		name = s->name;
+	} else {
+		/*
+		 * Create a unique name for the slab as a target
+		 * for the symlinks.
+		 */
+		name = create_unique_id(s);
+	}
+
+	slab_cache_dir = debugfs_create_dir(name, slab_debugfs_root);
+
+	debugfs_create_file("alloc_traces", 0400,
+		slab_cache_dir, s, &slab_debugfs_fops);
+
+	debugfs_create_file("free_traces", 0400,
+		slab_cache_dir, s, &slab_debugfs_fops);
+}
+
+void debugfs_slab_release(struct kmem_cache *s)
+{
+	debugfs_remove_recursive(debugfs_lookup(s->name,
+					slab_debugfs_root));
+}
+
+static int __init slab_debugfs_init(void)
+{
+	struct kmem_cache *s;
+
+	slab_debugfs_root = debugfs_create_dir("slab", NULL);
+
+	list_for_each_entry(s, &slab_caches, list)
+		if (s->flags & SLAB_STORE_USER)
+			debugfs_slab_add(s);
+
+	return 0;
+
+}
+__initcall(slab_debugfs_init);
+#endif
 /*
  * The /proc/slabinfo ABI
  */
-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a
member of the Code Aurora Forum, hosted by The Linux Foundation

Re: [PATCH v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs

[Greg KH]

On Tue, Jun 01, 2021 at 03:37:37PM +0530, Faiyaz Mohammed wrote:
> alloc_calls and free_calls implementation in sysfs have two issues,
> one is PAGE_SIZE limitiation of sysfs and other is it does not adhere
> to "one value per file" rule.
> 
> To overcome this issues, move the alloc_calls and free_calls implemeation
> to debugfs.
> 
> Debugfs cache will be created if SLAB_STORE_USER flag is set.
> 
> Rename the alloc_calls/free_calls to alloc_traces/free_traces,
> to be inline with what it does.
> 
> Signed-off-by: Faiyaz Mohammed <faiyazm@codeaurora.org>

Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Re: [PATCH v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs

[kernel test robot]

[-- Attachment #1: Type: text/plain, Size: 3009 bytes --]

Hi Faiyaz,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on linus/master]
[also build test ERROR on v5.13-rc4]
[cannot apply to hnaz-linux-mm/master next-20210601]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
base:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git c2131f7e73c9e9365613e323d65c7b9e5b910f56
config: i386-tinyconfig (attached as .config)
compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
reproduce (this is a W=1 build):
        # https://github.com/0day-ci/linux/commit/9540acc4691d680b7124d8daa1a2eb98a97ee19a
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
        git checkout 9540acc4691d680b7124d8daa1a2eb98a97ee19a
        # save the attached .config to linux build tree
        make W=1 ARCH=i386 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   mm/slab_common.c: In function 'slab_caches_to_rcu_destroy_workfn':
>> mm/slab_common.c:452:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror=implicit-function-declaration]
     452 |   debugfs_slab_release(s);
         |   ^~~~~~~~~~~~~~~~~~~~
   cc1: some warnings being treated as errors


vim +/debugfs_slab_release +452 mm/slab_common.c

   427	
   428	static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
   429	{
   430		LIST_HEAD(to_destroy);
   431		struct kmem_cache *s, *s2;
   432	
   433		/*
   434		 * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the
   435		 * @slab_caches_to_rcu_destroy list.  The slab pages are freed
   436		 * through RCU and the associated kmem_cache are dereferenced
   437		 * while freeing the pages, so the kmem_caches should be freed only
   438		 * after the pending RCU operations are finished.  As rcu_barrier()
   439		 * is a pretty slow operation, we batch all pending destructions
   440		 * asynchronously.
   441		 */
   442		mutex_lock(&slab_mutex);
   443		list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy);
   444		mutex_unlock(&slab_mutex);
   445	
   446		if (list_empty(&to_destroy))
   447			return;
   448	
   449		rcu_barrier();
   450	
   451		list_for_each_entry_safe(s, s2, &to_destroy, list) {
 > 452			debugfs_slab_release(s);
   453			kfence_shutdown_cache(s);
   454	#ifdef SLAB_SUPPORTS_SYSFS
   455			sysfs_slab_release(s);
   456	#else
   457			slab_kmem_cache_release(s);
   458	#endif
   459		}
   460	}
   461	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 7416 bytes --]

Re: [PATCH v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs

[Vlastimil Babka]

On 6/1/21 1:08 PM, kernel test robot wrote:
> Hi Faiyaz,
> 
> Thank you for the patch! Yet something to improve:
> 
> [auto build test ERROR on linus/master]
> [also build test ERROR on v5.13-rc4]
> [cannot apply to hnaz-linux-mm/master next-20210601]
> [If your patch is applied to the wrong git tree, kindly drop us a note.
> And when submitting patch, we suggest to use '--base' as documented in
> https://git-scm.com/docs/git-format-patch]
> 
> url:    https://github.com/0day-ci/linux/commits/Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
> base:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git c2131f7e73c9e9365613e323d65c7b9e5b910f56
> config: i386-tinyconfig (attached as .config)
> compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
> reproduce (this is a W=1 build):
>         # https://github.com/0day-ci/linux/commit/9540acc4691d680b7124d8daa1a2eb98a97ee19a
>         git remote add linux-review https://github.com/0day-ci/linux
>         git fetch --no-tags linux-review Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
>         git checkout 9540acc4691d680b7124d8daa1a2eb98a97ee19a
>         # save the attached .config to linux build tree
>         make W=1 ARCH=i386 
> 
> If you fix the issue, kindly add following tag as appropriate
> Reported-by: kernel test robot <lkp@intel.com>
> 
> All errors (new ones prefixed by >>):
> 
>    mm/slab_common.c: In function 'slab_caches_to_rcu_destroy_workfn':
>>> mm/slab_common.c:452:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror=implicit-function-declaration]
>      452 |   debugfs_slab_release(s);
>          |   ^~~~~~~~~~~~~~~~~~~~
>    cc1: some warnings being treated as errors

Oh, right, CONFIG_SLOB.
How about moving the ifdef-ed declaration from slub-def.h to mm/slab.h and
discarding the slab-def.h empty one. That will take care of SLOB too.

> vim +/debugfs_slab_release +452 mm/slab_common.c
> 
>    427	
>    428	static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
>    429	{
>    430		LIST_HEAD(to_destroy);
>    431		struct kmem_cache *s, *s2;
>    432	
>    433		/*
>    434		 * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the
>    435		 * @slab_caches_to_rcu_destroy list.  The slab pages are freed
>    436		 * through RCU and the associated kmem_cache are dereferenced
>    437		 * while freeing the pages, so the kmem_caches should be freed only
>    438		 * after the pending RCU operations are finished.  As rcu_barrier()
>    439		 * is a pretty slow operation, we batch all pending destructions
>    440		 * asynchronously.
>    441		 */
>    442		mutex_lock(&slab_mutex);
>    443		list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy);
>    444		mutex_unlock(&slab_mutex);
>    445	
>    446		if (list_empty(&to_destroy))
>    447			return;
>    448	
>    449		rcu_barrier();
>    450	
>    451		list_for_each_entry_safe(s, s2, &to_destroy, list) {
>  > 452			debugfs_slab_release(s);
>    453			kfence_shutdown_cache(s);
>    454	#ifdef SLAB_SUPPORTS_SYSFS
>    455			sysfs_slab_release(s);
>    456	#else
>    457			slab_kmem_cache_release(s);
>    458	#endif
>    459		}
>    460	}
>    461	
> 
> ---
> 0-DAY CI Kernel Test Service, Intel Corporation
> https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
> 

Re: [PATCH v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs

[Faiyaz Mohammed]

On 6/1/2021 5:44 PM, Vlastimil Babka wrote:
> On 6/1/21 1:08 PM, kernel test robot wrote:
>> Hi Faiyaz,
>>
>> Thank you for the patch! Yet something to improve:
>>
>> [auto build test ERROR on linus/master]
>> [also build test ERROR on v5.13-rc4]
>> [cannot apply to hnaz-linux-mm/master next-20210601]
>> [If your patch is applied to the wrong git tree, kindly drop us a note.
>> And when submitting patch, we suggest to use '--base' as documented in
>> https://git-scm.com/docs/git-format-patch]
>>
>> url:    https://github.com/0day-ci/linux/commits/Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
>> base:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git c2131f7e73c9e9365613e323d65c7b9e5b910f56
>> config: i386-tinyconfig (attached as .config)
>> compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
>> reproduce (this is a W=1 build):
>>         # https://github.com/0day-ci/linux/commit/9540acc4691d680b7124d8daa1a2eb98a97ee19a
>>         git remote add linux-review https://github.com/0day-ci/linux
>>         git fetch --no-tags linux-review Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
>>         git checkout 9540acc4691d680b7124d8daa1a2eb98a97ee19a
>>         # save the attached .config to linux build tree
>>         make W=1 ARCH=i386 
>>
>> If you fix the issue, kindly add following tag as appropriate
>> Reported-by: kernel test robot <lkp@intel.com>
>>
>> All errors (new ones prefixed by >>):
>>
>>    mm/slab_common.c: In function 'slab_caches_to_rcu_destroy_workfn':
>>>> mm/slab_common.c:452:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror=implicit-function-declaration]
>>      452 |   debugfs_slab_release(s);
>>          |   ^~~~~~~~~~~~~~~~~~~~
>>    cc1: some warnings being treated as errors
> 
> Oh, right, CONFIG_SLOB.
> How about moving the ifdef-ed declaration from slub-def.h to mm/slab.h and
> discarding the slab-def.h empty one. That will take care of SLOB too.
> 

Yes, doing the same.
>> vim +/debugfs_slab_release +452 mm/slab_common.c
>>
>>    427	
>>    428	static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
>>    429	{
>>    430		LIST_HEAD(to_destroy);
>>    431		struct kmem_cache *s, *s2;
>>    432	
>>    433		/*
>>    434		 * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the
>>    435		 * @slab_caches_to_rcu_destroy list.  The slab pages are freed
>>    436		 * through RCU and the associated kmem_cache are dereferenced
>>    437		 * while freeing the pages, so the kmem_caches should be freed only
>>    438		 * after the pending RCU operations are finished.  As rcu_barrier()
>>    439		 * is a pretty slow operation, we batch all pending destructions
>>    440		 * asynchronously.
>>    441		 */
>>    442		mutex_lock(&slab_mutex);
>>    443		list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy);
>>    444		mutex_unlock(&slab_mutex);
>>    445	
>>    446		if (list_empty(&to_destroy))
>>    447			return;
>>    448	
>>    449		rcu_barrier();
>>    450	
>>    451		list_for_each_entry_safe(s, s2, &to_destroy, list) {
>>  > 452			debugfs_slab_release(s);
>>    453			kfence_shutdown_cache(s);
>>    454	#ifdef SLAB_SUPPORTS_SYSFS
>>    455			sysfs_slab_release(s);
>>    456	#else
>>    457			slab_kmem_cache_release(s);
>>    458	#endif
>>    459		}
>>    460	}
>>    461	
>>
>> ---
>> 0-DAY CI Kernel Test Service, Intel Corporation
>> https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
>>
> 

Re: [PATCH v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs

[Vlastimil Babka]

On 6/1/21 12:07 PM, Faiyaz Mohammed wrote:
> alloc_calls and free_calls implementation in sysfs have two issues,
> one is PAGE_SIZE limitiation of sysfs and other is it does not adhere
> to "one value per file" rule.
> 
> To overcome this issues, move the alloc_calls and free_calls implemeation
> to debugfs.
> 
> Debugfs cache will be created if SLAB_STORE_USER flag is set.
> 
> Rename the alloc_calls/free_calls to alloc_traces/free_traces,
> to be inline with what it does.
> 
> Signed-off-by: Faiyaz Mohammed <faiyazm@codeaurora.org>

When reading the files I'm getting dmesg errors like this:

[   37.998152] seq_file: buggy .next function slab_debugfs_next did not update
position index

Don't recall this from previous versions?

> diff --git a/mm/slab_common.c b/mm/slab_common.c
> index a4a5714..ca39e84 100644
> --- a/mm/slab_common.c
> +++ b/mm/slab_common.c
> @@ -449,6 +449,7 @@ static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
>  	rcu_barrier();
>  
>  	list_for_each_entry_safe(s, s2, &to_destroy, list) {
> +		debugfs_slab_release(s);

You should remove this one, you already do the release when queueing the work
below. This is not like sysfs where we distinguish _unlink() when queueing and
_release when processing the work.

> +static struct dentry *slab_debugfs_root;
> +struct loc_track t = { 0, 0, NULL };

Ooops, this can't be a global variable, especially when shared without any lock
protection - but even with a protection it would be very suboptimal. You'll need
to make it a seq_file private, see e.g. here for an example:
https://lore.kernel.org/linux-mm/20210521121127.24653-3-glittao@gmail.com/

>  static void free_loc_track(struct loc_track *t)
>  {
>  	if (t->max)
> @@ -4825,82 +4840,7 @@ static void process_slab(struct loc_track *t, struct kmem_cache *s,
>  			add_location(t, s, get_track(s, p, alloc));
>  	put_map(map);
>  }
> -
> -static int list_locations(struct kmem_cache *s, char *buf,
> -			  enum track_item alloc)
> -{
> -	int len = 0;
> -	unsigned long i;
> -	struct loc_track t = { 0, 0, NULL };
> -	int node;
> -	struct kmem_cache_node *n;
> -
> -	if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location),
> -			     GFP_KERNEL)) {
> -		return sysfs_emit(buf, "Out of memory\n");
> -	}
> -	/* Push back cpu slabs */
> -	flush_all(s);
> -
> -	for_each_kmem_cache_node(s, node, n) {
> -		unsigned long flags;
> -		struct page *page;
> -
> -		if (!atomic_long_read(&n->nr_slabs))
> -			continue;
> -
> -		spin_lock_irqsave(&n->list_lock, flags);
> -		list_for_each_entry(page, &n->partial, slab_list)
> -			process_slab(&t, s, page, alloc);
> -		list_for_each_entry(page, &n->full, slab_list)
> -			process_slab(&t, s, page, alloc);
> -		spin_unlock_irqrestore(&n->list_lock, flags);
> -	}
> -
> -	for (i = 0; i < t.count; i++) {
> -		struct location *l = &t.loc[i];
> -
> -		len += sysfs_emit_at(buf, len, "%7ld ", l->count);
> -
> -		if (l->addr)
> -			len += sysfs_emit_at(buf, len, "%pS", (void *)l->addr);
> -		else
> -			len += sysfs_emit_at(buf, len, "<not-available>");
> -
> -		if (l->sum_time != l->min_time)
> -			len += sysfs_emit_at(buf, len, " age=%ld/%ld/%ld",
> -					     l->min_time,
> -					     (long)div_u64(l->sum_time,
> -							   l->count),
> -					     l->max_time);
> -		else
> -			len += sysfs_emit_at(buf, len, " age=%ld", l->min_time);
> -
> -		if (l->min_pid != l->max_pid)
> -			len += sysfs_emit_at(buf, len, " pid=%ld-%ld",
> -					     l->min_pid, l->max_pid);
> -		else
> -			len += sysfs_emit_at(buf, len, " pid=%ld",
> -					     l->min_pid);
> -
> -		if (num_online_cpus() > 1 &&
> -		    !cpumask_empty(to_cpumask(l->cpus)))
> -			len += sysfs_emit_at(buf, len, " cpus=%*pbl",
> -					     cpumask_pr_args(to_cpumask(l->cpus)));
> -
> -		if (nr_online_nodes > 1 && !nodes_empty(l->nodes))
> -			len += sysfs_emit_at(buf, len, " nodes=%*pbl",
> -					     nodemask_pr_args(&l->nodes));
> -
> -		len += sysfs_emit_at(buf, len, "\n");
> -	}
> -
> -	free_loc_track(&t);
> -	if (!t.count)
> -		len += sysfs_emit_at(buf, len, "No data\n");
> -
> -	return len;
> -}
> +#endif  /* CONFIG_DEBUG_FS   */
>  #endif	/* CONFIG_SLUB_DEBUG */
>  
>  #ifdef SLUB_RESILIENCY_TEST
> @@ -5350,21 +5290,6 @@ static ssize_t validate_store(struct kmem_cache *s,
>  }
>  SLAB_ATTR(validate);
>  
> -static ssize_t alloc_calls_show(struct kmem_cache *s, char *buf)
> -{
> -	if (!(s->flags & SLAB_STORE_USER))
> -		return -ENOSYS;
> -	return list_locations(s, buf, TRACK_ALLOC);
> -}
> -SLAB_ATTR_RO(alloc_calls);
> -
> -static ssize_t free_calls_show(struct kmem_cache *s, char *buf)
> -{
> -	if (!(s->flags & SLAB_STORE_USER))
> -		return -ENOSYS;
> -	return list_locations(s, buf, TRACK_FREE);
> -}
> -SLAB_ATTR_RO(free_calls);
>  #endif /* CONFIG_SLUB_DEBUG */
>  
>  #ifdef CONFIG_FAILSLAB
> @@ -5528,8 +5453,6 @@ static struct attribute *slab_attrs[] = {
>  	&poison_attr.attr,
>  	&store_user_attr.attr,
>  	&validate_attr.attr,
> -	&alloc_calls_attr.attr,
> -	&free_calls_attr.attr,
>  #endif
>  #ifdef CONFIG_ZONE_DMA
>  	&cache_dma_attr.attr,
> @@ -5818,6 +5741,203 @@ static int __init slab_sysfs_init(void)
>  __initcall(slab_sysfs_init);
>  #endif /* CONFIG_SYSFS */
>  
> +#if defined(CONFIG_SLUB_DEBUG) && defined(CONFIG_DEBUG_FS)
> +static int slab_debugfs_show(struct seq_file *seq, void *v)
> +{
> +
> +	struct location *l;
> +	unsigned int idx = *(unsigned int *)v;
> +
> +	if (idx < t.count) {
> +		l = &t.loc[idx];
> +
> +		seq_printf(seq, "%7ld ", l->count);
> +
> +		if (l->addr)
> +			seq_printf(seq, "%pS", (void *)l->addr);
> +		else
> +			seq_puts(seq, "<not-available>");
> +
> +		if (l->sum_time != l->min_time) {
> +			seq_printf(seq, " age=%ld/%ld/%ld",
> +				l->min_time,
> +				(long)div_u64(l->sum_time, l->count),
> +				l->max_time);
> +		} else
> +			seq_printf(seq, " age=%ld",
> +				l->min_time);
> +
> +		if (l->min_pid != l->max_pid)
> +			seq_printf(seq, " pid=%ld-%ld",
> +				l->min_pid, l->max_pid);
> +		else
> +			seq_printf(seq, " pid=%ld",
> +				l->min_pid);
> +
> +		if (num_online_cpus() > 1 &&
> +				!cpumask_empty(to_cpumask(l->cpus)))
> +			seq_printf(seq, " cpus=%*pbl",
> +				 cpumask_pr_args(to_cpumask(l->cpus)));
> +
> +		if (nr_online_nodes > 1 && !nodes_empty(l->nodes))
> +			seq_printf(seq, " nodes=%*pbl",
> +				 nodemask_pr_args(&l->nodes));
> +
> +		seq_puts(seq, "\n");
> +	}
> +
> +	if (t.count == 0)
> +		seq_puts(seq, "No data\n");
> +
> +	return 0;
> +}
> +
> +static void slab_debugfs_stop(struct seq_file *seq, void *v)
> +{
> +	if (!v && t.max) {
> +		free_loc_track(&t);
> +		t.max = 0;
> +	}
> +}
> +
> +static void *slab_debugfs_next(struct seq_file *seq, void *v, loff_t *ppos)
> +{
> +	loff_t *spos = v;
> +
> +	if (*ppos < t.count) {
> +		*spos = *spos + 1;
> +		*ppos = *spos;
> +		return spos;
> +	}
> +
> +	return NULL;
> +}
> +
> +static void *slab_debugfs_start(struct seq_file *seq, loff_t *ppos)
> +{
> +	struct kmem_cache_node *n;
> +	struct kmem_cache *s;
> +	enum track_item alloc;
> +	int node;
> +	loff_t *spos = kmalloc(sizeof(loff_t), GFP_KERNEL);
> +
> +	s = seq->file->f_inode->i_private;
> +
> +	if (!spos)
> +		return NULL;
> +
> +	if (*ppos == 0) {
> +
> +		t.count = 0;
> +		t.max = 0;
> +		t.loc = NULL;
> +		if (strcmp(seq->file->f_path.dentry->d_name.name, "alloc_traces") == 0)
> +			alloc =  TRACK_ALLOC;
> +		else
> +			alloc =  TRACK_FREE;
> +
> +		if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location),
> +			     GFP_KERNEL)) {
> +			seq_puts(seq, "Out of memory\n");
> +			kfree(spos);
> +			return ERR_PTR(-ENOMEM);
> +		}
> +		/* Push back cpu slabs */
> +		flush_all(s);
> +
> +		for_each_kmem_cache_node(s, node, n) {
> +			unsigned long flags;
> +			struct page *page;
> +
> +			if (!atomic_long_read(&n->nr_slabs))
> +				continue;
> +
> +			spin_lock_irqsave(&n->list_lock, flags);
> +			list_for_each_entry(page, &n->partial, slab_list)
> +				process_slab(&t, s, page, alloc);
> +			list_for_each_entry(page, &n->full, slab_list)
> +				process_slab(&t, s, page, alloc);
> +			spin_unlock_irqrestore(&n->list_lock, flags);
> +		}
> +	}
> +
> +	if (*ppos < t.count) {
> +		*spos = *ppos;
> +		return spos;
> +	}
> +
> +	kfree(spos);
> +	return NULL;
> +}
> +
> +static const struct seq_operations slab_debugfs_sops = {
> +	.start  = slab_debugfs_start,
> +	.next   = slab_debugfs_next,
> +	.stop   = slab_debugfs_stop,
> +	.show   = slab_debugfs_show
> +};
> +DEFINE_SEQ_ATTRIBUTE(slab_debugfs);
> +
> +static void debugfs_slab_add(struct kmem_cache *s)
> +{
> +	const char *name;
> +	struct dentry *slab_cache_dir;
> +	int unmergeable = slab_unmergeable(s);
> +
> +	if (unlikely(!slab_debugfs_root))
> +		return;
> +
> +	if (!unmergeable && disable_higher_order_debug &&
> +			(slub_debug & DEBUG_METADATA_FLAGS))
> +		unmergeable = 1;
> +
> +	if (unmergeable) {
> +		/*
> +		 * Slabcache can never be merged so we can use the name proper.
> +		 * This is typically the case for debug situations. In that
> +		 * case we can catch duplicate names easily.
> +		 */
> +		slab_cache_dir = debugfs_lookup(s->name, slab_debugfs_root);
> +		debugfs_remove_recursive(slab_cache_dir);
> +		name = s->name;
> +	} else {
> +		/*
> +		 * Create a unique name for the slab as a target
> +		 * for the symlinks.
> +		 */
> +		name = create_unique_id(s);
> +	}
> +
> +	slab_cache_dir = debugfs_create_dir(name, slab_debugfs_root);
> +
> +	debugfs_create_file("alloc_traces", 0400,
> +		slab_cache_dir, s, &slab_debugfs_fops);
> +
> +	debugfs_create_file("free_traces", 0400,
> +		slab_cache_dir, s, &slab_debugfs_fops);
> +}
> +
> +void debugfs_slab_release(struct kmem_cache *s)
> +{
> +	debugfs_remove_recursive(debugfs_lookup(s->name,
> +					slab_debugfs_root));
> +}
> +
> +static int __init slab_debugfs_init(void)
> +{
> +	struct kmem_cache *s;
> +
> +	slab_debugfs_root = debugfs_create_dir("slab", NULL);
> +
> +	list_for_each_entry(s, &slab_caches, list)
> +		if (s->flags & SLAB_STORE_USER)
> +			debugfs_slab_add(s);
> +
> +	return 0;
> +
> +}
> +__initcall(slab_debugfs_init);
> +#endif
>  /*
>   * The /proc/slabinfo ABI
>   */
> 

Re: [PATCH v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs

[Faiyaz Mohammed]

On 6/1/2021 7:01 PM, Vlastimil Babka wrote:
> On 6/1/21 12:07 PM, Faiyaz Mohammed wrote:
>> alloc_calls and free_calls implementation in sysfs have two issues,
>> one is PAGE_SIZE limitiation of sysfs and other is it does not adhere
>> to "one value per file" rule.
>>
>> To overcome this issues, move the alloc_calls and free_calls implemeation
>> to debugfs.
>>
>> Debugfs cache will be created if SLAB_STORE_USER flag is set.
>>
>> Rename the alloc_calls/free_calls to alloc_traces/free_traces,
>> to be inline with what it does.
>>
>> Signed-off-by: Faiyaz Mohammed <faiyazm@codeaurora.org>
> 
> When reading the files I'm getting dmesg errors like this:
> 
> [   37.998152] seq_file: buggy .next function slab_debugfs_next did not update
> position index
> 
> Don't recall this from previous versions?
> 
slab_debugfs_next() is not updating the pos after last read, so it's
retuning null. In fs/seq_file.c if pos is not updated in ->next() fn,
it's printing the above info but I will address this in next patch version.

>> diff --git a/mm/slab_common.c b/mm/slab_common.c
>> index a4a5714..ca39e84 100644
>> --- a/mm/slab_common.c
>> +++ b/mm/slab_common.c
>> @@ -449,6 +449,7 @@ static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
>>  	rcu_barrier();
>>  
>>  	list_for_each_entry_safe(s, s2, &to_destroy, list) {
>> +		debugfs_slab_release(s);
> 
> You should remove this one, you already do the release when queueing the work
> below. This is not like sysfs where we distinguish _unlink() when queueing and
> _release when processing the work.
>
oh, yes, I will update in next patch version.

>> +static struct dentry *slab_debugfs_root;
>> +struct loc_track t = { 0, 0, NULL };
> 
> Ooops, this can't be a global variable, especially when shared without any lock
> protection - but even with a protection it would be very suboptimal. You'll need
> to make it a seq_file private, see e.g. here for an example:
> https://lore.kernel.org/linux-mm/20210521121127.24653-3-glittao@gmail.com/
>
I will address this in next patch version.

>>  static void free_loc_track(struct loc_track *t)
>>  {
>>  	if (t->max)
>> @@ -4825,82 +4840,7 @@ static void process_slab(struct loc_track *t, struct kmem_cache *s,
>>  			add_location(t, s, get_track(s, p, alloc));
>>  	put_map(map);
>>  }
>> -
>> -static int list_locations(struct kmem_cache *s, char *buf,
>> -			  enum track_item alloc)
>> -{
>> -	int len = 0;
>> -	unsigned long i;
>> -	struct loc_track t = { 0, 0, NULL };
>> -	int node;
>> -	struct kmem_cache_node *n;
>> -
>> -	if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location),
>> -			     GFP_KERNEL)) {
>> -		return sysfs_emit(buf, "Out of memory\n");
>> -	}
>> -	/* Push back cpu slabs */
>> -	flush_all(s);
>> -
>> -	for_each_kmem_cache_node(s, node, n) {
>> -		unsigned long flags;
>> -		struct page *page;
>> -
>> -		if (!atomic_long_read(&n->nr_slabs))
>> -			continue;
>> -
>> -		spin_lock_irqsave(&n->list_lock, flags);
>> -		list_for_each_entry(page, &n->partial, slab_list)
>> -			process_slab(&t, s, page, alloc);
>> -		list_for_each_entry(page, &n->full, slab_list)
>> -			process_slab(&t, s, page, alloc);
>> -		spin_unlock_irqrestore(&n->list_lock, flags);
>> -	}
>> -
>> -	for (i = 0; i < t.count; i++) {
>> -		struct location *l = &t.loc[i];
>> -
>> -		len += sysfs_emit_at(buf, len, "%7ld ", l->count);
>> -
>> -		if (l->addr)
>> -			len += sysfs_emit_at(buf, len, "%pS", (void *)l->addr);
>> -		else
>> -			len += sysfs_emit_at(buf, len, "<not-available>");
>> -
>> -		if (l->sum_time != l->min_time)
>> -			len += sysfs_emit_at(buf, len, " age=%ld/%ld/%ld",
>> -					     l->min_time,
>> -					     (long)div_u64(l->sum_time,
>> -							   l->count),
>> -					     l->max_time);
>> -		else
>> -			len += sysfs_emit_at(buf, len, " age=%ld", l->min_time);
>> -
>> -		if (l->min_pid != l->max_pid)
>> -			len += sysfs_emit_at(buf, len, " pid=%ld-%ld",
>> -					     l->min_pid, l->max_pid);
>> -		else
>> -			len += sysfs_emit_at(buf, len, " pid=%ld",
>> -					     l->min_pid);
>> -
>> -		if (num_online_cpus() > 1 &&
>> -		    !cpumask_empty(to_cpumask(l->cpus)))
>> -			len += sysfs_emit_at(buf, len, " cpus=%*pbl",
>> -					     cpumask_pr_args(to_cpumask(l->cpus)));
>> -
>> -		if (nr_online_nodes > 1 && !nodes_empty(l->nodes))
>> -			len += sysfs_emit_at(buf, len, " nodes=%*pbl",
>> -					     nodemask_pr_args(&l->nodes));
>> -
>> -		len += sysfs_emit_at(buf, len, "\n");
>> -	}
>> -
>> -	free_loc_track(&t);
>> -	if (!t.count)
>> -		len += sysfs_emit_at(buf, len, "No data\n");
>> -
>> -	return len;
>> -}
>> +#endif  /* CONFIG_DEBUG_FS   */
>>  #endif	/* CONFIG_SLUB_DEBUG */
>>  
>>  #ifdef SLUB_RESILIENCY_TEST
>> @@ -5350,21 +5290,6 @@ static ssize_t validate_store(struct kmem_cache *s,
>>  }
>>  SLAB_ATTR(validate);
>>  
>> -static ssize_t alloc_calls_show(struct kmem_cache *s, char *buf)
>> -{
>> -	if (!(s->flags & SLAB_STORE_USER))
>> -		return -ENOSYS;
>> -	return list_locations(s, buf, TRACK_ALLOC);
>> -}
>> -SLAB_ATTR_RO(alloc_calls);
>> -
>> -static ssize_t free_calls_show(struct kmem_cache *s, char *buf)
>> -{
>> -	if (!(s->flags & SLAB_STORE_USER))
>> -		return -ENOSYS;
>> -	return list_locations(s, buf, TRACK_FREE);
>> -}
>> -SLAB_ATTR_RO(free_calls);
>>  #endif /* CONFIG_SLUB_DEBUG */
>>  
>>  #ifdef CONFIG_FAILSLAB
>> @@ -5528,8 +5453,6 @@ static struct attribute *slab_attrs[] = {
>>  	&poison_attr.attr,
>>  	&store_user_attr.attr,
>>  	&validate_attr.attr,
>> -	&alloc_calls_attr.attr,
>> -	&free_calls_attr.attr,
>>  #endif
>>  #ifdef CONFIG_ZONE_DMA
>>  	&cache_dma_attr.attr,
>> @@ -5818,6 +5741,203 @@ static int __init slab_sysfs_init(void)
>>  __initcall(slab_sysfs_init);
>>  #endif /* CONFIG_SYSFS */
>>  
>> +#if defined(CONFIG_SLUB_DEBUG) && defined(CONFIG_DEBUG_FS)
>> +static int slab_debugfs_show(struct seq_file *seq, void *v)
>> +{
>> +
>> +	struct location *l;
>> +	unsigned int idx = *(unsigned int *)v;
>> +
>> +	if (idx < t.count) {
>> +		l = &t.loc[idx];
>> +
>> +		seq_printf(seq, "%7ld ", l->count);
>> +
>> +		if (l->addr)
>> +			seq_printf(seq, "%pS", (void *)l->addr);
>> +		else
>> +			seq_puts(seq, "<not-available>");
>> +
>> +		if (l->sum_time != l->min_time) {
>> +			seq_printf(seq, " age=%ld/%ld/%ld",
>> +				l->min_time,
>> +				(long)div_u64(l->sum_time, l->count),
>> +				l->max_time);
>> +		} else
>> +			seq_printf(seq, " age=%ld",
>> +				l->min_time);
>> +
>> +		if (l->min_pid != l->max_pid)
>> +			seq_printf(seq, " pid=%ld-%ld",
>> +				l->min_pid, l->max_pid);
>> +		else
>> +			seq_printf(seq, " pid=%ld",
>> +				l->min_pid);
>> +
>> +		if (num_online_cpus() > 1 &&
>> +				!cpumask_empty(to_cpumask(l->cpus)))
>> +			seq_printf(seq, " cpus=%*pbl",
>> +				 cpumask_pr_args(to_cpumask(l->cpus)));
>> +
>> +		if (nr_online_nodes > 1 && !nodes_empty(l->nodes))
>> +			seq_printf(seq, " nodes=%*pbl",
>> +				 nodemask_pr_args(&l->nodes));
>> +
>> +		seq_puts(seq, "\n");
>> +	}
>> +
>> +	if (t.count == 0)
>> +		seq_puts(seq, "No data\n");
>> +
>> +	return 0;
>> +}
>> +
>> +static void slab_debugfs_stop(struct seq_file *seq, void *v)
>> +{
>> +	if (!v && t.max) {
>> +		free_loc_track(&t);
>> +		t.max = 0;
>> +	}
>> +}
>> +
>> +static void *slab_debugfs_next(struct seq_file *seq, void *v, loff_t *ppos)
>> +{
>> +	loff_t *spos = v;
>> +
>> +	if (*ppos < t.count) {
>> +		*spos = *spos + 1;
>> +		*ppos = *spos;
>> +		return spos;
>> +	}
>> +
>> +	return NULL;
>> +}
>> +
>> +static void *slab_debugfs_start(struct seq_file *seq, loff_t *ppos)
>> +{
>> +	struct kmem_cache_node *n;
>> +	struct kmem_cache *s;
>> +	enum track_item alloc;
>> +	int node;
>> +	loff_t *spos = kmalloc(sizeof(loff_t), GFP_KERNEL);
>> +
>> +	s = seq->file->f_inode->i_private;
>> +
>> +	if (!spos)
>> +		return NULL;
>> +
>> +	if (*ppos == 0) {
>> +
>> +		t.count = 0;
>> +		t.max = 0;
>> +		t.loc = NULL;
>> +		if (strcmp(seq->file->f_path.dentry->d_name.name, "alloc_traces") == 0)
>> +			alloc =  TRACK_ALLOC;
>> +		else
>> +			alloc =  TRACK_FREE;
>> +
>> +		if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location),
>> +			     GFP_KERNEL)) {
>> +			seq_puts(seq, "Out of memory\n");
>> +			kfree(spos);
>> +			return ERR_PTR(-ENOMEM);
>> +		}
>> +		/* Push back cpu slabs */
>> +		flush_all(s);
>> +
>> +		for_each_kmem_cache_node(s, node, n) {
>> +			unsigned long flags;
>> +			struct page *page;
>> +
>> +			if (!atomic_long_read(&n->nr_slabs))
>> +				continue;
>> +
>> +			spin_lock_irqsave(&n->list_lock, flags);
>> +			list_for_each_entry(page, &n->partial, slab_list)
>> +				process_slab(&t, s, page, alloc);
>> +			list_for_each_entry(page, &n->full, slab_list)
>> +				process_slab(&t, s, page, alloc);
>> +			spin_unlock_irqrestore(&n->list_lock, flags);
>> +		}
>> +	}
>> +
>> +	if (*ppos < t.count) {
>> +		*spos = *ppos;
>> +		return spos;
>> +	}
>> +
>> +	kfree(spos);
>> +	return NULL;
>> +}
>> +
>> +static const struct seq_operations slab_debugfs_sops = {
>> +	.start  = slab_debugfs_start,
>> +	.next   = slab_debugfs_next,
>> +	.stop   = slab_debugfs_stop,
>> +	.show   = slab_debugfs_show
>> +};
>> +DEFINE_SEQ_ATTRIBUTE(slab_debugfs);
>> +
>> +static void debugfs_slab_add(struct kmem_cache *s)
>> +{
>> +	const char *name;
>> +	struct dentry *slab_cache_dir;
>> +	int unmergeable = slab_unmergeable(s);
>> +
>> +	if (unlikely(!slab_debugfs_root))
>> +		return;
>> +
>> +	if (!unmergeable && disable_higher_order_debug &&
>> +			(slub_debug & DEBUG_METADATA_FLAGS))
>> +		unmergeable = 1;
>> +
>> +	if (unmergeable) {
>> +		/*
>> +		 * Slabcache can never be merged so we can use the name proper.
>> +		 * This is typically the case for debug situations. In that
>> +		 * case we can catch duplicate names easily.
>> +		 */
>> +		slab_cache_dir = debugfs_lookup(s->name, slab_debugfs_root);
>> +		debugfs_remove_recursive(slab_cache_dir);
>> +		name = s->name;
>> +	} else {
>> +		/*
>> +		 * Create a unique name for the slab as a target
>> +		 * for the symlinks.
>> +		 */
>> +		name = create_unique_id(s);
>> +	}
>> +
>> +	slab_cache_dir = debugfs_create_dir(name, slab_debugfs_root);
>> +
>> +	debugfs_create_file("alloc_traces", 0400,
>> +		slab_cache_dir, s, &slab_debugfs_fops);
>> +
>> +	debugfs_create_file("free_traces", 0400,
>> +		slab_cache_dir, s, &slab_debugfs_fops);
>> +}
>> +
>> +void debugfs_slab_release(struct kmem_cache *s)
>> +{
>> +	debugfs_remove_recursive(debugfs_lookup(s->name,
>> +					slab_debugfs_root));
>> +}
>> +
>> +static int __init slab_debugfs_init(void)
>> +{
>> +	struct kmem_cache *s;
>> +
>> +	slab_debugfs_root = debugfs_create_dir("slab", NULL);
>> +
>> +	list_for_each_entry(s, &slab_caches, list)
>> +		if (s->flags & SLAB_STORE_USER)
>> +			debugfs_slab_add(s);
>> +
>> +	return 0;
>> +
>> +}
>> +__initcall(slab_debugfs_init);
>> +#endif
>>  /*
>>   * The /proc/slabinfo ABI
>>   */
>>
> 

Re: [PATCH v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs

[kernel test robot]

[-- Attachment #1: Type: text/plain, Size: 3698 bytes --]

Hi Faiyaz,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on linus/master]
[also build test ERROR on v5.13-rc4]
[cannot apply to hnaz-linux-mm/master next-20210601]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
base:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git c2131f7e73c9e9365613e323d65c7b9e5b910f56
config: x86_64-randconfig-a001-20210601 (attached as .config)
compiler: clang version 13.0.0 (https://github.com/llvm/llvm-project db26cd30b6dd65e88d786e97a1e453af5cd48966)
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # install x86_64 cross compiling tool for clang build
        # apt-get install binutils-x86-64-linux-gnu
        # https://github.com/0day-ci/linux/commit/9540acc4691d680b7124d8daa1a2eb98a97ee19a
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
        git checkout 9540acc4691d680b7124d8daa1a2eb98a97ee19a
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross ARCH=x86_64 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

>> mm/slab_common.c:452:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror,-Wimplicit-function-declaration]
                   debugfs_slab_release(s);
                   ^
   mm/slab_common.c:473:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror,-Wimplicit-function-declaration]
                   debugfs_slab_release(s);
                   ^
   mm/slab_common.c:481:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror,-Wimplicit-function-declaration]
                   debugfs_slab_release(s);
                   ^
   3 errors generated.


vim +/debugfs_slab_release +452 mm/slab_common.c

   427	
   428	static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
   429	{
   430		LIST_HEAD(to_destroy);
   431		struct kmem_cache *s, *s2;
   432	
   433		/*
   434		 * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the
   435		 * @slab_caches_to_rcu_destroy list.  The slab pages are freed
   436		 * through RCU and the associated kmem_cache are dereferenced
   437		 * while freeing the pages, so the kmem_caches should be freed only
   438		 * after the pending RCU operations are finished.  As rcu_barrier()
   439		 * is a pretty slow operation, we batch all pending destructions
   440		 * asynchronously.
   441		 */
   442		mutex_lock(&slab_mutex);
   443		list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy);
   444		mutex_unlock(&slab_mutex);
   445	
   446		if (list_empty(&to_destroy))
   447			return;
   448	
   449		rcu_barrier();
   450	
   451		list_for_each_entry_safe(s, s2, &to_destroy, list) {
 > 452			debugfs_slab_release(s);
   453			kfence_shutdown_cache(s);
   454	#ifdef SLAB_SUPPORTS_SYSFS
   455			sysfs_slab_release(s);
   456	#else
   457			slab_kmem_cache_release(s);
   458	#endif
   459		}
   460	}
   461	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 35575 bytes --]