Re: [f2fs-dev] [PATCH v3] f2fs: fix race of pending_pages in decompression

[Chao Yu]

On 2020/12/7 15:28, Daeho Jeong wrote:
> > It looks like it will be better to move this into merge condition?
> >
> >                  if (bio && (!page_is_mergeable(sbi, bio,
> >                                          *last_block_in_bio, blkaddr) ||
> >                      !f2fs_crypt_mergeable_bio(bio, inode, page->index, NULL) ||
> >                          f2fs_verify_mergeable_bio())) {
>
> I tried this for the first time, but it requires unnecessary checks
> within the compression cluster.

We only need to check f2fs_verify_mergeable_bio for i == 0 case? something like:

static bool f2fs_verify_mergeable_bio(struct bio *bio, bool verify, bool first_page)
{
	if (!first_page)
		return false;
	if (!verify)
		return false;

	ctx = bio->bi_private;
	if (!(ctx->enabled_steps & (1 << STEP_VERITY)))
		return true;
}

Thoughts?

> I wanted to just check one time in the beginning of the cluster.
> What do you think?

It's trivial, but I'm think about the readability... at least, one line comment
is needed to describe why we submit previous bio. :)

Thanks,

> .