Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <ad5e7f08-5211-0885-04d9-96fd741ccbe2@suse.cz>
Date:   Wed, 2 Nov 2022 13:35:04 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.4.0
Subject: Re: [RFC PATCH] mm: introduce accounting of page_owner via backtrace
Content-Language: en-US
To:     "zhaoyang.huang" <zhaoyang.huang@unisoc.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Zhaoyang Huang <huangzhaoyang@gmail.com>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, ke.wang@unisoc.com,
        steve.kang@unisoc.com, Oscar Salvador <osalvador@suse.de>,
        Michal Hocko <mhocko@suse.com>
References: <1667390781-17515-1-git-send-email-zhaoyang.huang@unisoc.com>
From:   Vlastimil Babka <vbabka@suse.cz>
In-Reply-To: <1667390781-17515-1-git-send-email-zhaoyang.huang@unisoc.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 11/2/22 13:06, zhaoyang.huang wrote:
> From: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
> 
> Page_owner could be accounted and sort via its backtrace, which could
> simplify the output.
> 
> Signed-off-by: Zhaoyang Huang <zhaoyang.huang@unisoc.com>

At first glance this looks like the same goal as Oscar's (CC'd) patchset:
https://lore.kernel.org/linux-mm/20220905031012.4450-1-osalvador@suse.de/


> ---
>  include/linux/page_owner.h |   7 +++
>  mm/page_owner.c            | 113 +++++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 120 insertions(+)
> 
> diff --git a/include/linux/page_owner.h b/include/linux/page_owner.h
> index 119a0c9..c86a342 100644
> --- a/include/linux/page_owner.h
> +++ b/include/linux/page_owner.h
> @@ -3,6 +3,13 @@
>  #define __LINUX_PAGE_OWNER_H
>  
>  #include <linux/jump_label.h>
> +#include <linux/stackdepot.h>
> +
> +struct hash_object {
> +	int count;
> +	depot_stack_handle_t trace_hash;
> +	struct rb_node rb_node;
> +};
>  
>  #ifdef CONFIG_PAGE_OWNER
>  extern struct static_key_false page_owner_inited;
> diff --git a/mm/page_owner.c b/mm/page_owner.c
> index e4c6f3f..3d014fd 100644
> --- a/mm/page_owner.c
> +++ b/mm/page_owner.c
> @@ -42,6 +42,12 @@ struct page_owner {
>  static depot_stack_handle_t early_handle;
>  
>  static void init_early_allocated_pages(void);
> +static void store_user_hash(depot_stack_handle_t trace_hash, int count);
> +static DEFINE_SPINLOCK(stack_hash_lock);
> +static struct hash_object ho_pool[8192];
> +static int ho_pool_free_count = ARRAY_SIZE(ho_pool);
> +struct rb_root user_hash_root = RB_ROOT;
> +
>  
>  static int __init early_page_owner_param(char *buf)
>  {
> @@ -152,6 +158,8 @@ void __reset_page_owner(struct page *page, unsigned short order)
>  		page_owner->free_handle = handle;
>  		page_owner->free_ts_nsec = free_ts_nsec;
>  		page_ext = page_ext_next(page_ext);
> +		if (!i)
> +			store_user_hash(page_owner->handle, -(1 << order));
>  	}
>  }
>  
> @@ -190,6 +198,7 @@ noinline void __set_page_owner(struct page *page, unsigned short order,
>  		return;
>  
>  	handle = save_stack(gfp_mask);
> +	store_user_hash(handle, 1 << order);
>  	__set_page_owner_handle(page_ext, handle, order, gfp_mask);
>  }
>  
> @@ -570,6 +579,105 @@ void __dump_page_owner(const struct page *page)
>  	return 0;
>  }
>  
> +static void store_user_hash(depot_stack_handle_t trace_hash, int count)
> +{
> +	struct rb_node *rb_parent;
> +	unsigned int trace_hash_parent;
> +	struct rb_node **link = &user_hash_root.rb_node;
> +	struct hash_object *hash_parent = NULL;
> +	struct hash_object *hash_object = NULL;
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&stack_hash_lock, flags);
> +	while (*link) {
> +		rb_parent = *link;
> +		hash_parent = rb_entry(rb_parent, struct hash_object, rb_node);
> +		trace_hash_parent = hash_parent->trace_hash;
> +		if (trace_hash < trace_hash_parent)
> +			link = &hash_parent->rb_node.rb_left;
> +		else if (trace_hash_parent < trace_hash)
> +			link = &hash_parent->rb_node.rb_right;
> +		else {
> +			hash_object = rb_entry(*link, struct hash_object, rb_node);
> +			hash_object->count += count;
> +			if (!RB_EMPTY_ROOT(&user_hash_root)
> +					&& !hash_object->count) {
> +				rb_erase(&hash_object->rb_node, &user_hash_root);
> +			}
> +			spin_unlock_irqrestore(&stack_hash_lock, flags);
> +			return;
> +		}
> +	}
> +	/*
> +	 * hash_object is the 1st node represent this trace_hash
> +	 * insert it to user_hash_root
> +	 */
> +	hash_object = ho_pool_free_count ? &ho_pool[--ho_pool_free_count] : NULL;
> +	if (!hash_object) {
> +		spin_unlock_irqrestore(&stack_hash_lock, flags);
> +		return;
> +	}
> +	hash_object->trace_hash = trace_hash;
> +	/* add the node to rb tree*/
> +	rb_link_node(&hash_object->rb_node, rb_parent, link);
> +	rb_insert_color(&hash_object->rb_node, &user_hash_root);
> +	spin_unlock_irqrestore(&stack_hash_lock, flags);
> +	return;
> +}
> +static ssize_t print_owner_stack(struct file *file, char __user *buf, size_t count, loff_t *ppos)
> +{
> +	struct hash_object *object;
> +	struct rb_node *rb;
> +	unsigned long *entries;
> +	unsigned int nr_entries;
> +	char *kbuf;
> +	int i, ret = 0;
> +	unsigned long flags;
> +
> +	count = min_t(size_t, count, PAGE_SIZE);
> +	kbuf = kmalloc(count, GFP_KERNEL);
> +	if (!kbuf)
> +		return -ENOMEM;
> +
> +	rb = file->private_data ? (struct rb_node *)file->private_data : rb_first(&user_hash_root);
> +	if(!rb)
> +		return 0;
> +	spin_lock_irqsave(&stack_hash_lock, flags);
> +	object = rb_entry(rb, struct hash_object, rb_node);
> +	if (object) {
> +		while (!object->trace_hash || object->count <= 0) {
> +			rb = rb_next(rb);
> +			object = rb ? rb_entry(rb, struct hash_object, rb_node) : NULL;
> +			if (!object) {
> +				spin_unlock_irqrestore(&stack_hash_lock, flags);
> +				kfree(kbuf);
> +				return 0;
> +			}
> +		}
> +		nr_entries = stack_depot_fetch(object->trace_hash, &entries);
> +		ret += snprintf(kbuf + ret, count - ret, "count %d\n", object->count);
> +		if (ret >= count)
> +			goto err;
> +		for (i = 0; i < nr_entries; i++) {
> +			void *ptr = (void *)entries[i];
> +			ret += snprintf(kbuf + ret, count - ret, "    [<%p>] %pS\n", ptr, ptr);
> +			if (ret >= count)
> +				goto err;
> +		}
> +	}
> +	file->private_data = (void *)rb_next(rb);
> +	spin_unlock_irqrestore(&stack_hash_lock, flags);
> +	if (copy_to_user(buf, kbuf, ret))
> +		ret = -EFAULT;
> +
> +	kfree(kbuf);
> +	return ret;
> +err:
> +	spin_unlock_irqrestore(&stack_hash_lock, flags);
> +	kfree(kbuf);
> +	return -ENOMEM;
> +}
> +
>  static void init_pages_in_zone(pg_data_t *pgdat, struct zone *zone)
>  {
>  	unsigned long pfn = zone->zone_start_pfn;
> @@ -661,6 +769,9 @@ static void init_early_allocated_pages(void)
>  static const struct file_operations proc_page_owner_operations = {
>  	.read		= read_page_owner,
>  };
> +static const struct file_operations proc_page_owner_simple_operations = {
> +	.read		= print_owner_stack,
> +};
>  
>  static int __init pageowner_init(void)
>  {
> @@ -671,6 +782,8 @@ static int __init pageowner_init(void)
>  
>  	debugfs_create_file("page_owner", 0400, NULL, NULL,
>  			    &proc_page_owner_operations);
> +	debugfs_create_file("page_owner_simple", 0400, NULL, NULL,
> +			    &proc_page_owner_simple_operations);
>  
>  	return 0;
>  }