On Thu, Jun 29, 2023 at 11:44:35AM +0800, Yangtao Li wrote:

On 2023/6/29 0:46, Matthew Wilcox wrote:

quoted

On Wed, Jun 28, 2023 at 05:34:54PM +0800, Yangtao Li wrote:

quoted

Introduce queue_logical_block_mask() and bdev_logical_block_mask()
to simplify code, which replace (queue_logical_block_size(q) - 1)
and (bdev_logical_block_size(bdev) - 1).

The thing is that I know what queue_logical_block_size - 1 does.
That's the low bits.  _Which_ bits are queue_logical_block_mask?
The high bits or the low bits?  And before you say "It's obviously",
we have both ways round in the kernel today.

I guess for this you mentioned, can we name it bdev_logical_block_lmask and
queue_logical_block_lmask?

{bdev,queue}_offset_in_lba() ?

--D

Thx,

quoted

I am not in favour of this change.  I might be in favour of bool
queue_logical_block_aligned(q, x), but even then it doesn't seem worth
the bits.