Re: [PATCH 1/5] staging: zsmalloc: zsmalloc memory allocation library

[Seth Jennings <sjenning@linux.vnet.ibm.com>]

On 01/26/2012 01:12 PM, Dave Hansen wrote:
> On 01/09/2012 02:51 PM, Seth Jennings wrote:
>> +	area = &get_cpu_var(zs_map_area);
>> +	if (off + class->size <= PAGE_SIZE) {
>> +		/* this object is contained entirely within a page */
>> +		area->vm_addr = kmap_atomic(page);
>> +	} else {
>> +		/* this object spans two pages */
>> +		struct page *nextp;
>> +
>> +		nextp = get_next_page(page);
>> +		BUG_ON(!nextp);
>> +
>> +
>> +		set_pte(area->vm_ptes[0], mk_pte(page, PAGE_KERNEL));
>> +		set_pte(area->vm_ptes[1], mk_pte(nextp, PAGE_KERNEL));
>> +
>> +		/* We pre-allocated VM area so mapping can never fail */
>> +		area->vm_addr = area->vm->addr;
>> +	}
> 
> This bit appears to be trying to make kmap_atomic() variant that can map
> two pages in to contigious virtual addresses.  Instead of open-coding it
> in a non-portable way like this, should we just make a new kmap_atomic()
> variant that does this?
> 
> From the way it's implemented, I _think_ you're guaranteed to get two
> contiguous addresses if you do two adjacent kmap_atomics() on the same CPU:
> 
> void *kmap_atomic_prot(struct page *page, pgprot_t prot)
> {
> ...
>         type = kmap_atomic_idx_push();
>         idx = type + KM_TYPE_NR*smp_processor_id();
>         vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
> 
> I think if you do a get_cpu()/put_cpu() or just a preempt_disable()
> across the operations you'll be guaranteed to get two contiguous addresses.

I'm not quite following here.  kmap_atomic() only does this for highmem pages.
For normal pages (all pages for 64-bit), it doesn't do any mapping at all.  It
just returns the virtual address of the page since it is in the kernel's address
space.

For this design, the pages _must_ be mapped, even if the pages are directly
reachable in the address space, because they must be virtually contiguous.

--
Seth