Re: [PATCH v3 3/3] mm: cma: release trigger

[Sasha Levin <sasha.levin@oracle.com>]

On 01/29/2015 02:51 AM, Joonsoo Kim wrote:
> On Tue, Jan 27, 2015 at 09:22:57PM -0500, Sasha Levin wrote:
>> > +static int cma_free_mem(struct cma *cma, int count)
>> > +{
>> > +	struct cma_mem *mem = NULL;
>> > +
>> > +	while (count) {
>> > +		mem = cma_get_entry_from_list(cma);
>> > +		if (mem == NULL)
>> > +			return 0;
>> > +
>> > +		if (mem->n <= count) {
>> > +			cma_release(cma, mem->p, mem->n);
>> > +			count -= mem->n;
>> > +			kfree(mem);
>> > +		} else if (cma->order_per_bit == 0) {
>> > +			cma_release(cma, mem->p, count);
>> > +			mem->p += count;
>> > +			mem->n -= count;
>> > +			count = 0;
>> > +			cma_add_to_cma_mem_list(cma, mem);
>> > +		} else {
>> > +			cma_add_to_cma_mem_list(cma, mem);
>> > +			break;
>> > +		}
>> > +	}
> Hmm... I'm not sure we need to deal with "mem->m < count" case.
> It is not suitable for any cma_alloc/release usecase.

Maybe not, but the code in mm/cma.c supports that right now. If that
ever changes we can make appropriate changes here.

> And, I'd like to insert some error log to last case rather than
> silently ignoring the request.

I can add that.


Thanks,
Sasha