Re: [PATCH 3/5] block: remove support for the host aware zone model

[Damien Le Moal]

On 12/19/23 17:08, Ed Tsai (蔡宗軒) wrote:
> On Tue, 2023-12-19 at 07:16 +0000, Naohiro Aota wrote:
>>  	 
>> External email : Please do not click links or open attachments until
>> you have verified the sender or the content.
>>  On Mon, Dec 18, 2023 at 08:21:22AM +0000, Ed Tsai (蔡宗軒) wrote:
>>> On Mon, 2023-12-18 at 15:53 +0900, Damien Le Moal wrote:
>>>>  On 2023/12/18 15:15, Ed Tsai (蔡宗軒) wrote:
>>>>> Hi Christoph,
>>>>>
>>>>> some minor suggestions:
>>>>>
>>>>> On Sun, 2023-12-17 at 17:53 +0100, Christoph Hellwig wrote:
>>>>>>
>>>>>> diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
>>>>>> index 198d38b53322c1..260b5b8f2b0d7e 100644
>>>>>> --- a/drivers/md/dm-table.c
>>>>>> +++ b/drivers/md/dm-table.c
>>>>>> @@ -1579,21 +1579,18 @@ bool
>> dm_table_has_no_data_devices(struct
>>>>>> dm_table *t)
>>>>>>  return true;
>>>>>>  }
>>>>>>  
>>>>>> -static int device_not_zoned_model(struct dm_target *ti,
>> struct
>>>>>> dm_dev *dev,
>>>>>> -  sector_t start, sector_t len, void
>>>>>> *data)
>>>>>> +static int device_not_zoned(struct dm_target *ti, struct
>> dm_dev
>>>>>> *dev,
>>>>>> +    sector_t start, sector_t len, void *data)
>>>>>>  {
>>>>>> -struct request_queue *q = bdev_get_queue(dev->bdev);
>>>>>> -enum blk_zoned_model *zoned_model = data;
>>>>>> +bool *zoned = data;
>>>>>>  
>>>>>> -return blk_queue_zoned_model(q) != *zoned_model;
>>>>>> +return bdev_is_zoned(dev->bdev) != *zoned;
>>>>>>  }
>>>>>>  
>>>>>>  static int device_is_zoned_model(struct dm_target *ti, struct
>>>> dm_dev
>>>>>> *dev,
>>>>>>   sector_t start, sector_t len, void
>>>>>> *data)
>>>>>
>>>>> Seems like the word "model" should also be remove here.
>>>>>
>>>>>>  {
>>>>>> -struct request_queue *q = bdev_get_queue(dev->bdev);
>>>>>> -
>>>>>> -return blk_queue_zoned_model(q) != BLK_ZONED_NONE;
>>>>>> +return bdev_is_zoned(dev->bdev);
>>>>>>  }
>>>>>>  
>>>>>>  /*
>>>>>> @@ -1603,8 +1600,7 @@ static int device_is_zoned_model(struct
>>>>>> dm_target *ti, struct dm_dev *dev,
>>>>>>   * has the DM_TARGET_MIXED_ZONED_MODEL feature set, the
>> devices
>>>> can
>>>>>> have any
>>>>>>   * zoned model with all zoned devices having the same zone
>> size.
>>>>>>   */
>>>>>> -static bool dm_table_supports_zoned_model(struct dm_table *t,
>>>>>> -  enum blk_zoned_model
>>>>>> zoned_model)
>>>>>> +static bool dm_table_supports_zoned(struct dm_table *t, bool
>>>> zoned)
>>>>>>  {
>>>>>>  for (unsigned int i = 0; i < t->num_targets; i++) {
>>>>>>  struct dm_target *ti = dm_table_get_target(t, i);
>>>>>> @@ -1623,11 +1619,11 @@ static bool
>>>>>> dm_table_supports_zoned_model(struct dm_table *t,
>>>>>>  
>>>>>>  if (dm_target_supports_zoned_hm(ti->type)) {
>>>>>>  if (!ti->type->iterate_devices ||
>>>>>> -    ti->type->iterate_devices(ti,
>>>>>> device_not_zoned_model,
>>>>>> -      &zoned_model))
>>>>>> +    ti->type->iterate_devices(ti,
>>>>>> device_not_zoned,
>>>>>> +      &zoned))
>>>>>>  return false;
>>>>>>  } else if (!dm_target_supports_mixed_zoned_model(ti-
>>>>>>> type)) {
>>>>>> -if (zoned_model == BLK_ZONED_HM)
>>>>>> +if (zoned)
>>>>>>  return false;
>>>>>>  }
>>>>>>  }
>>>>>
>>>>> The parameter "bool zoned" is redundant. It should be removed
>> from
>>>> the
>>>>> above 3 functions
>>>
>>> The two func, is zoned and not zoned, are essentially the same.
>> They
>>> can be simplified into one function.
>>
>> Both functions are used for iterate_devices's callback in
>> dm_table_supports_zoned_model(). As shown in raid_iterate_devices(),
>> iterate_devices() returns 0 if the callback func calls on all the
>> devices
>> returns 0, or returns a non-zero result early otherwise. So, the
>> iterate_devices() call returns "true" if any one of the underlying
>> devices
>> is (zoned|not zoned).
>>
>> Since we cannot create lambda as in other fancy languages, we need
>> two
>> functions...
> 
> Not really, there is a "void *data" can be used.
> 
> The device_is_zoned_model() is just the same as the device_not_zoned()
> with (bool *)data = false.
> 
> It's very minor, so is okay to ignore my preference.

Send a patch on top of Christoph's series if you want to clean this up.

-- 
Damien Le Moal
Western Digital Research