On 2024/1/4 20:32, Yu Kuai wrote:
Hi, Jan!
在 2024/01/04 20:02, Jan Kara 写道:
On Thu 21-12-23 16:58:26, Yu Kuai wrote:
From: Yu Kuai <yukuai3@xxxxxxxxxx>
Avoid to access bd_inode directly, prepare to remove bd_inode from
block_device.
Signed-off-by: Yu Kuai <yukuai3@xxxxxxxxxx>
I'm not erofs maintainer but IMO this is quite ugly and grows erofs_buf
unnecessarily. I'd rather store 'sb' pointer in erofs_buf and then do the
right thing in erofs_bread() which is the only place that seems to care
about the erofs_is_fscache_mode() distinction... Also blkszbits is then
trivially sb->s_blocksize_bits so it would all seem much more
straightforward.
Thanks for your suggestion, I'll follow this unless Gao Xiang has other
suggestions.
Yes, that would be better, I'm fine with that. Yet in the future we
may support a seperate large dirblocksize more than block size, but
we could revisit later.
Thanks,
Gao Xiang
Kuai
Honza
