On 04/30/2017 05:22 AM, Xiubo Li wrote:
> On 2017年04月30日 13:48, Mike Christie wrote:
>> On 04/26/2017 01:25 AM, lixiubo@xxxxxxxxxxxxxxxxxxxx wrote:
>>> for_each_sg(data_sg, sg, data_nents, i) {
>>> @@ -275,22 +371,26 @@ static void alloc_and_scatter_data_area(struct
>>> tcmu_dev *udev,
>>> from = kmap_atomic(sg_page(sg)) + sg->offset;
>>> while (sg_remaining > 0) {
>>> if (block_remaining == 0) {
>>> - block = find_first_zero_bit(udev->data_bitmap,
>>> - DATA_BLOCK_BITS);
>>> block_remaining = DATA_BLOCK_SIZE;
>>> - set_bit(block, udev->data_bitmap);
>>> + dbi = tcmu_get_empty_block(udev, &to);
>>> + if (dbi < 0)
>>
>> I know it you fixed the missing kunmap_atomic here and missing unlock in
>> tcmu_queue_cmd_ring in the next patch, but I think normally people
>> prefer that one patch does not add a bug, then the next patch fixes it.
> Do you mean the following kmap_atomic() ?
>
> from = kmap_atomic(sg_page(sg)) + sg->offset;
>
> In this patch there has no new kmap/kunmap introduced. This is the old
> code and
> the kunmap is at the end of aasda().
You added a new return in the error path in this patch in the if case
above, but did not add a kunmap_atomic.
