Received-SPF: pass (google.com: domain of linux-kernel+bounces-101660-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Date: Wed, 13 Mar 2024 15:04:17 +0000
From: Matthew Wilcox <willy@infradead.org>
To: Suren Baghdasaryan <surenb@google.com>
Cc: akpm@linux-foundation.org, kent.overstreet@linux.dev, mhocko@suse.com,
	vbabka@suse.cz, hannes@cmpxchg.org, roman.gushchin@linux.dev,
	mgorman@suse.de, dave@stgolabs.net, liam.howlett@oracle.com,
	penguin-kernel@i-love.sakura.ne.jp, corbet@lwn.net,
	void@manifault.com, peterz@infradead.org, juri.lelli@redhat.com,
	catalin.marinas@arm.com, will@kernel.org, arnd@arndb.de,
	tglx@linutronix.de, mingo@redhat.com, dave.hansen@linux.intel.com,
	x86@kernel.org, peterx@redhat.com, david@redhat.com,
	axboe@kernel.dk, mcgrof@kernel.org, masahiroy@kernel.org,
	nathan@kernel.org, dennis@kernel.org, jhubbard@nvidia.com,
	tj@kernel.org, muchun.song@linux.dev, rppt@kernel.org,
	paulmck@kernel.org, pasha.tatashin@soleen.com,
	yosryahmed@google.com, yuzhao@google.com, dhowells@redhat.com,
	hughd@google.com, andreyknvl@gmail.com, keescook@chromium.org,
	ndesaulniers@google.com, vvvvvv@google.com,
	gregkh@linuxfoundation.org, ebiggers@google.com, ytcoode@gmail.com,
	vincent.guittot@linaro.org, dietmar.eggemann@arm.com,
	rostedt@goodmis.org, bsegall@google.com, bristot@redhat.com,
	vschneid@redhat.com, cl@linux.com, penberg@kernel.org,
	iamjoonsoo.kim@lge.com, 42.hyeyoo@gmail.com, glider@google.com,
	elver@google.com, dvyukov@google.com, shakeelb@google.com,
	songmuchun@bytedance.com, jbaron@akamai.com, aliceryhl@google.com,
	rientjes@google.com, minchan@google.com, kaleshsingh@google.com,
	kernel-team@android.com, linux-doc@vger.kernel.org,
	linux-kernel@vger.kernel.org, iommu@lists.linux.dev,
	linux-arch@vger.kernel.org, linux-fsdevel@vger.kernel.org,
	linux-mm@kvack.org, linux-modules@vger.kernel.org,
	kasan-dev@googlegroups.com, cgroups@vger.kernel.org
Subject: Re: [PATCH v5 20/37] mm: fix non-compound multi-order memory
 accounting in __free_pages
Message-ID: <ZfHAcVwJ6w9b1x0Z@casper.infradead.org>
References: <20240306182440.2003814-1-surenb@google.com>
 <20240306182440.2003814-21-surenb@google.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20240306182440.2003814-21-surenb@google.com>

On Wed, Mar 06, 2024 at 10:24:18AM -0800, Suren Baghdasaryan wrote:
> When a non-compound multi-order page is freed, it is possible that a
> speculative reference keeps the page pinned. In this case we free all
> pages except for the first page, which will be freed later by the last
> put_page(). However put_page() ignores the order of the page being freed,
> treating it as a 0-order page. This creates a memory accounting imbalance
> because the pages freed in __free_pages() do not have their own alloc_tag
> and their memory was accounted to the first page. To fix this the first
> page should adjust its allocation size counter when "tail" pages are freed.

It's not "ignored".  It's not available!

Better wording:

However the page passed to put_page() is indisinguishable from an
order-0 page, so it cannot do the accounting, just as it cannot free
the subsequent pages.  Do the accounting here, where we free the pages.

(I'm sure further improvements are possible)

> +static inline void pgalloc_tag_sub_bytes(struct alloc_tag *tag, unsigned int order)
> +{
> +	if (mem_alloc_profiling_enabled() && tag)
> +		this_cpu_sub(tag->counters->bytes, PAGE_SIZE << order);
> +}

This is a terribly named function.  And it's not even good for what we
want to use it for.

static inline void pgalloc_tag_sub_pages(struct alloc_tag *tag, unsigned int nr)
{
	if (mem_alloc_profiling_enabled() && tag)
		this_cpu_sub(tag->counters->bytes, PAGE_SIZE * nr);
}

> +++ b/mm/page_alloc.c
> @@ -4697,12 +4697,21 @@ void __free_pages(struct page *page, unsigned int order)
>  {
>  	/* get PageHead before we drop reference */
>  	int head = PageHead(page);
> +	struct alloc_tag *tag = pgalloc_tag_get(page);
>  
>  	if (put_page_testzero(page))
>  		free_the_page(page, order);
>  	else if (!head)
> -		while (order-- > 0)
> +		while (order-- > 0) {
>  			free_the_page(page + (1 << order), order);
> +			/*
> +			 * non-compound multi-order page accounts all allocations
> +			 * to the first page (just like compound one), therefore
> +			 * we need to adjust the allocation size of the first
> +			 * page as its order is ignored when put_page() frees it.
> +			 */
> +			pgalloc_tag_sub_bytes(tag, order);

-	else if (!head
+	else if (!head) {
+		pgalloc_tag_sub_pages(1 << order - 1);
		while (order-- > 0)
			free_the_page(page + (1 << order), order);
+	}

It doesn't need a comment, it's obvious what you're doing.