Re: [PATCH RFC v3 for-6.8/block 04/17] mtd: block2mtd: use bdev apis

From: Yu Kuai
Date: Thu Jan 04 2024 - 07:23:40 EST


Hi,

在 2024/01/04 19:28, Jan Kara 写道:
On Thu 21-12-23 16:56:59, Yu Kuai wrote:
From: Yu Kuai <yukuai3@xxxxxxxxxx>

On the one hand covert to use folio while reading bdev inode, on the
other hand prevent to access bd_inode directly.

Signed-off-by: Yu Kuai <yukuai3@xxxxxxxxxx>
...
+ for (p = folio_address(folio); p < max; p++)
if (*p != -1UL) {
- lock_page(page);
- memset(page_address(page), 0xff, PAGE_SIZE);
- set_page_dirty(page);
- unlock_page(page);
- balance_dirty_pages_ratelimited(mapping);
+ folio_lock(folio);
+ memset(folio_address(folio), 0xff,
+ folio_size(folio));
+ folio_mark_dirty(folio);
+ folio_unlock(folio);
+ bdev_balance_dirty_pages_ratelimited(bdev);

Rather then creating this bdev_balance_dirty_pages_ratelimited() just for
MTD perhaps we can have here (and in other functions):

...
mapping = folio_mapping(folio);
folio_unlock(folio);
if (mapping)
balance_dirty_pages_ratelimited(mapping);

What do you think? Because when we are working with the folios it is rather
natural to use their mapping for dirty balancing?

I think this is a great idea! And bdev_balance_dirty_pages_ratelimited()
can be removed as well.

Thanks,
Kuai


Honza