Re: [PATCH V3 0/7] f2fs: new mount API conversion

[Jaegeuk Kim <jaegeuk@kernel.org>]

Hmm, I had to drop the series at the moment, since it seems needing more
work to deal with default_options(), which breaks my device setup.
For example, set_opt(sbi, READ_EXTENT_CACHE) in default_options is not propagating
to the below logics. In this case, do we need ctx_set_opt() if user doesn't set?

On 04/23, Eric Sandeen wrote:
> V3:
> - Rebase onto git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git
>   dev branch
> - Fix up some 0day robot warnings
> 
> This is a forward-port of Hongbo's original f2fs mount API conversion,
> posted last August at 
> https://lore.kernel.org/linux-f2fs-devel/20240814023912.3959299-1-lihongbo22@huawei.com/
> 
> I had been trying to approach this with a little less complexity,
> but in the end I realized that Hongbo's approach (which follows
> the ext4 approach) was a good one, and I was not making any progrss
> myself. 😉
> 
> In addition to the forward-port, I have also fixed a couple bugs I found
> during testing, and some improvements / style choices as well. Hongbo and
> I have discussed most of this off-list already, so I'm presenting the
> net result here.
> 
> This does pass my typical testing which does a large number of random
> mounts/remounts with valid and invalid option sets, on f2fs filesystem
> images with various features in the on-disk superblock. (I was not able
> to test all of this completely, as some options or features require
> hardware I dn't have.)
> 
> Thanks,
> -Eric
> 
> (A recap of Hongbo's original cover letter is below, edited slightly for
> this series:)
> 
> Since many filesystems have done the new mount API conversion,
> we introduce the new mount API conversion in f2fs.
> 
> The series can be applied on top of the current mainline tree
> and the work is based on the patches from Lukas Czerner (has
> done this in ext4[1]). His patch give me a lot of ideas.
> 
> Here is a high level description of the patchset:
> 
> 1. Prepare the f2fs mount parameters required by the new mount
> API and use it for parsing, while still using the old API to
> get mount options string. Split the parameter parsing and
> validation of the parse_options helper into two separate
> helpers.
> 
>   f2fs: Add fs parameter specifications for mount options
>   f2fs: move the option parser into handle_mount_opt
> 
> 2. Remove the use of sb/sbi structure of f2fs from all the
> parsing code, because with the new mount API the parsing is
> going to be done before we even get the super block. In this
> part, we introduce f2fs_fs_context to hold the temporary
> options when parsing. For the simple options check, it has
> to be done during parsing by using f2fs_fs_context structure.
> For the check which needs sb/sbi, we do this during super
> block filling.
> 
>   f2fs: Allow sbi to be NULL in f2fs_printk
>   f2fs: Add f2fs_fs_context to record the mount options
>   f2fs: separate the options parsing and options checking
> 
> 3. Switch the f2fs to use the new mount API for mount and
> remount.
> 
>   f2fs: introduce fs_context_operation structure
>   f2fs: switch to the new mount api
> 
> [1] https://lore.kernel.org/all/20211021114508.21407-1-lczerner@redhat.com/