Received-SPF: pass (google.com: domain of linux-kernel+bounces-171836-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Date: Tue, 7 May 2024 17:49:00 +0200
From: Claudio Imbrenda <imbrenda@linux.ibm.com>
To: David Hildenbrand <david@redhat.com>
Cc: linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
        linux-s390@vger.kernel.org, Heiko Carstens <hca@linux.ibm.com>,
        Vasily
 Gorbik <gor@linux.ibm.com>,
        Alexander Gordeev <agordeev@linux.ibm.com>,
        Christian Borntraeger <borntraeger@linux.ibm.com>,
        Sven Schnelle
 <svens@linux.ibm.com>,
        Janosch Frank <frankja@linux.ibm.com>,
        Gerald
 Schaefer <gerald.schaefer@linux.ibm.com>,
        Matthew Wilcox
 <willy@infradead.org>, Thomas Huth <thuth@redhat.com>
Subject: Re: [PATCH v2 04/10] s390/uv: convert PG_arch_1 users to only work
 on small folios
Message-ID: <20240507174900.0b05791a@p-imbrenda.boeblingen.de.ibm.com>
In-Reply-To: <20240412142120.220087-5-david@redhat.com>
References: <20240412142120.220087-1-david@redhat.com>
	<20240412142120.220087-5-david@redhat.com>
Organization: IBM
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Fri, 12 Apr 2024 16:21:14 +0200
David Hildenbrand <david@redhat.com> wrote:

> Now that make_folio_secure() may only set PG_arch_1 for small folios,
> let's convert relevant remaining UV code to only work on (small) folios
> and simply reject large folios early. This way, we'll never end up
> touching PG_arch_1 on tail pages of a large folio in UV code.
> 
> The folio_get()/folio_put() for functions that are documented to already
> hold a folio reference look weird; likely they are required to make
> concurrent gmap_make_secure() back off because the caller might only hold
> an implicit reference due to the page mapping. So leave that alone for now.
> 
> Signed-off-by: David Hildenbrand <david@redhat.com>

Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>

> ---
>  arch/s390/include/asm/page.h |  2 ++
>  arch/s390/kernel/uv.c        | 41 ++++++++++++++++++++++--------------
>  2 files changed, 27 insertions(+), 16 deletions(-)
> 
> diff --git a/arch/s390/include/asm/page.h b/arch/s390/include/asm/page.h
> index 9381879f7ecf..b64384872c0f 100644
> --- a/arch/s390/include/asm/page.h
> +++ b/arch/s390/include/asm/page.h
> @@ -214,7 +214,9 @@ static inline unsigned long __phys_addr(unsigned long x, bool is_31bit)
>  #define pfn_to_phys(pfn)	((pfn) << PAGE_SHIFT)
>  
>  #define phys_to_page(phys)	pfn_to_page(phys_to_pfn(phys))
> +#define phys_to_folio(phys)	page_folio(phys_to_page(phys))
>  #define page_to_phys(page)	pfn_to_phys(page_to_pfn(page))
> +#define folio_to_phys(page)	pfn_to_phys(folio_pfn(folio))
>  
>  static inline void *pfn_to_virt(unsigned long pfn)
>  {
> diff --git a/arch/s390/kernel/uv.c b/arch/s390/kernel/uv.c
> index 3c6d86e3e828..914dcec27329 100644
> --- a/arch/s390/kernel/uv.c
> +++ b/arch/s390/kernel/uv.c
> @@ -135,14 +135,18 @@ static int uv_destroy_page(unsigned long paddr)
>   */
>  int uv_destroy_owned_page(unsigned long paddr)
>  {
> -	struct page *page = phys_to_page(paddr);
> +	struct folio *folio = phys_to_folio(paddr);
>  	int rc;
>  
> -	get_page(page);
> +	/* See gmap_make_secure(): large folios cannot be secure */
> +	if (unlikely(folio_test_large(folio)))
> +		return 0;
> +
> +	folio_get(folio);
>  	rc = uv_destroy_page(paddr);
>  	if (!rc)
> -		clear_bit(PG_arch_1, &page->flags);
> -	put_page(page);
> +		clear_bit(PG_arch_1, &folio->flags);
> +	folio_put(folio);
>  	return rc;
>  }
>  
> @@ -170,14 +174,18 @@ int uv_convert_from_secure(unsigned long paddr)
>   */
>  int uv_convert_owned_from_secure(unsigned long paddr)
>  {
> -	struct page *page = phys_to_page(paddr);
> +	struct folio *folio = phys_to_folio(paddr);
>  	int rc;
>  
> -	get_page(page);
> +	/* See gmap_make_secure(): large folios cannot be secure */
> +	if (unlikely(folio_test_large(folio)))
> +		return 0;
> +
> +	folio_get(folio);
>  	rc = uv_convert_from_secure(paddr);
>  	if (!rc)
> -		clear_bit(PG_arch_1, &page->flags);
> -	put_page(page);
> +		clear_bit(PG_arch_1, &folio->flags);
> +	folio_put(folio);
>  	return rc;
>  }
>  
> @@ -479,33 +487,34 @@ EXPORT_SYMBOL_GPL(gmap_destroy_page);
>   */
>  int arch_make_page_accessible(struct page *page)
>  {
> +	struct folio *folio = page_folio(page);
>  	int rc = 0;
>  
> -	/* Hugepage cannot be protected, so nothing to do */
> -	if (PageHuge(page))
> +	/* See gmap_make_secure(): large folios cannot be secure */
> +	if (unlikely(folio_test_large(folio)))
>  		return 0;
>  
>  	/*
>  	 * PG_arch_1 is used in 3 places:
>  	 * 1. for kernel page tables during early boot
>  	 * 2. for storage keys of huge pages and KVM
> -	 * 3. As an indication that this page might be secure. This can
> +	 * 3. As an indication that this small folio might be secure. This can
>  	 *    overindicate, e.g. we set the bit before calling
>  	 *    convert_to_secure.
>  	 * As secure pages are never huge, all 3 variants can co-exists.
>  	 */
> -	if (!test_bit(PG_arch_1, &page->flags))
> +	if (!test_bit(PG_arch_1, &folio->flags))
>  		return 0;
>  
> -	rc = uv_pin_shared(page_to_phys(page));
> +	rc = uv_pin_shared(folio_to_phys(folio));
>  	if (!rc) {
> -		clear_bit(PG_arch_1, &page->flags);
> +		clear_bit(PG_arch_1, &folio->flags);
>  		return 0;
>  	}
>  
> -	rc = uv_convert_from_secure(page_to_phys(page));
> +	rc = uv_convert_from_secure(folio_to_phys(folio));
>  	if (!rc) {
> -		clear_bit(PG_arch_1, &page->flags);
> +		clear_bit(PG_arch_1, &folio->flags);
>  		return 0;
>  	}
>