Re: [PATCH v4 5/6] arm64: mm: Convert to GENERIC_IOREMAP

[Kefeng Wang <wangkefeng.wang@huawei.com>]

On 2022/6/6 15:54, Christoph Hellwig wrote:
>> +#define ioremap_wc(addr, size)		ioremap_prot((addr), (size), PROT_NORMAL_NC)
>> +#define ioremap_np(addr, size)		ioremap_prot((addr), (size), PROT_DEVICE_nGnRnE)
> Please avoid the overly long lines here.  Independt of that having
> a non-trivial body on a separate line tends to generlly be a lot more
> readable anyway.

Hi Christoph，

As commit bdc48fa11e46  ("checkpatch/coding-style: deprecate 80-column 
warning") increased

the limit to 100 columns，so I don't get warning when using checkpatch, 
and it is not a hard limit,

but if this is a mandatory requirement, I will resend them with break lines.

>> +#define ioremap_cache(addr, size) ({							\
>> +	pfn_is_map_memory(__phys_to_pfn(addr)) ?					\
>> +	(void __iomem *)__phys_to_virt(addr) : ioremap_prot(addr, size, PROT_NORMAL);	\
>> +})
> And this really should be an inline function.

We still need a define, see kernel/iomem.c,

#ifndef ioremap_cache
__weak void __iomem *ioremap_cache(resource_size_t offset, unsigned long 
size)
{
         return ioremap(offset, size);
}
#endif

>
>> +int iounmap_allowed(void __iomem *addr)
>>   {
>>   	/*
>>   	 * We could get an address outside vmalloc range in case
>>   	 * of ioremap_cache() reusing a RAM mapping.
>>   	 */
>> +	return is_vmalloc_addr(addr) ? 0 : -EINVAL;
> As the generic ioremap only returns vmalloc addresses, this check
> really should go into common code.

Good point, will move into generic ioremap, thanks.

> .