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;
Date:   Thu, 4 Aug 2022 18:02:12 +0200
From:   Alexander Gordeev <agordeev@linux.ibm.com>
To:     Baoquan He <bhe@redhat.com>
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        akpm@linux-foundation.org, hch@infradead.org,
        wangkefeng.wang@huawei.com, linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH 02/11] mm: ioremap: fixup the physical address
Message-ID: <YuvthDzuPlAwD/LA@li-4a3a4a4c-28e5-11b2-a85c-a8d192c6f089.ibm.com>
References: <20220801144029.57829-1-bhe@redhat.com>
 <20220801144029.57829-3-bhe@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220801144029.57829-3-bhe@redhat.com>
Precedence: bulk

On Mon, Aug 01, 2022 at 10:40:20PM +0800, Baoquan He wrote:
> This is a preparation patch, no functionality change.

There is, please see below.

> @@ -3,11 +3,17 @@
>  #include <linux/mm.h>
>  #include <linux/io.h>
>  
> -void __iomem *ioremap_allowed(phys_addr_t phys_addr, size_t size, unsigned long prot)
> +void __iomem *
> +ioremap_allowed(phys_addr_t *paddr, size_t size, unsigned long *prot_val)
>  {
> -	unsigned long last_addr = phys_addr + size - 1;
> +	unsigned long last_addr, offset, phys_addr = *paddr;
>  	int ret = -EINVAL;
>  
> +	offset = phys_addr & (~PAGE_MASK);
> +	phys_addr -= offset;

FWIW, phys_addr &= PAGE_MASK looks much more usual.

> @@ -11,13 +11,20 @@
>  #include <linux/io.h>
>  #include <linux/export.h>
>  
> -void __iomem *ioremap_prot(phys_addr_t phys_addr, size_t size,
> +void __iomem *ioremap_prot(phys_addr_t paddr, size_t size,
>  			   unsigned long prot)
>  {
>  	unsigned long offset, vaddr;
> -	phys_addr_t last_addr;
> +	phys_addr_t last_addr, phys_addr = paddr;
>  	struct vm_struct *area;
>  	void __iomem *base;
> +	unsigned long prot_val = prot;

Why prot_val is needed?

> +	base = ioremap_allowed(&phys_addr, size, &prot_val);
> +	if (IS_ERR(base))
> +		return NULL;
> +	else if (base)
> +		return base;

By moving ioremap_allowed() here you allow it to be called
before the wrap-around check, including architectures that
do not do fixups.

And now ioremap_allowed() semantics, prototype and name turn
less than obvious. Why not introduce a separate fixup callback?

>  	/* Disallow wrap-around or zero size */
>  	last_addr = phys_addr + size - 1;
> @@ -29,12 +36,6 @@ void __iomem *ioremap_prot(phys_addr_t phys_addr, size_t size,
>  	phys_addr -= offset;
>  	size = PAGE_ALIGN(size + offset);
>  
> -	base = ioremap_allowed(phys_addr, size, prot);
> -	if (IS_ERR(base))
> -		return NULL;
> -	else if (base)
> -		return base;
> -
>  	area = get_vm_area_caller(size, VM_IOREMAP,
>  			__builtin_return_address(0));
>  	if (!area)