Re: [PATCH v1 01/24] IB/core: Introduce new fast registration API

[Bart Van Assche <bart.vanassche@sandisk.com>]

On 09/24/2015 12:37 AM, Sagi Grimberg wrote:
> On 9/23/2015 12:21 AM, Bart Van Assche wrote:
>> On 09/17/2015 02:42 AM, Sagi Grimberg wrote:
>>> +            } else if (last_page_off + dma_len < mr->page_size) {
>>> +                /* chunk this fragment with the last */
>>> +                last_end_dma_addr += dma_len;
>>> +                last_page_off += dma_len;
>>> +                mr->length += dma_len;
>>> +                continue;
>>
>> Shouldn't this code update last_page_addr ?
>
> Actually I think it doesn't since it is only relevant for the else
> statement where we are passing the page_size boundary.

Hello Sagi,

Suppose that the following sg-list is passed to this function as { 
offset, length } pairs and that this list has not been coalesced by the 
DMA mapping code: [ { 0, page_size / 4 }, { page_size / 4, page_size / 4 
}, { 2 * page_size / 4, page_size / 2 } ]. I think the algorithm in 
patch 01/24 will map the above sample sg-list onto two pages. Shouldn't 
that sg-list be mapped onto one page instead ?

Thanks,

Bart.