Re: [f2fs-dev] [PATCH v6 0/9] Support negative dentries on case-insensitive ext4 and f2fs

From: Gabriel Krisman Bertazi <krisman@suse.de>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Christian Brauner <brauner@kernel.org>,
	 viro@zeniv.linux.org.uk, tytso@mit.edu,
	 linux-f2fs-devel@lists.sourceforge.net, ebiggers@kernel.org,
	 linux-fsdevel@vger.kernel.org,  jaegeuk@kernel.org,
	linux-ext4@vger.kernel.org
Subject: Re: [f2fs-dev] [PATCH v6 0/9] Support negative dentries on case-insensitive ext4 and f2fs
Date: Tue, 21 Nov 2023 22:30:48 -0500	[thread overview]
Message-ID: <87zfz6jwgn.fsf@> (raw)
In-Reply-To: <CAHk-=whTCWwfmSzv3uVLN286_WZ6coN-GNw=4DWja7NZzp5ytg@mail.gmail.com> (Linus Torvalds's message of "Mon, 20 Nov 2023 10:07:51 -0800")

Linus Torvalds <torvalds@linux-foundation.org> writes:

> I dislike case folding with a passion - it's about the worst design
> decision a filesystem can ever do - but the other side of that is that
> if you have to have case folding, the last thing you want to do is to
> have each filesystem deal with that sh*t-for-brains decision itself.

Thanks for pitching in.

We all agree it is a horrible feature that we support for very specific
use cases.  I'm the one who added the code, yes, but I was never
under the illusion it's a good feature.  It solely exists to let Linux
handle the bad decisions made elsewhere.

> So moving more support for case folding into the VFS so that the
> horrid thing at least gets better support is something I'm perfectly
> fine with despite my dislike of it.

Yes. The entire implementation was meant to stay as far away as possible
from the fast lookup path (didn't want to displease Viro).  The negative
dentry is the only exception that required more changes to vfs to
address the issue he found of dangling negative dentries when turning a
directory case-insensitive.

But, fyi, there is work in progress to add support to more filesystems.
This is why I really want to get all of this done first. There is a use
case to enable it in shmem because of containerized environments
running wine; I was recently cc'ed on a bcachefs implementation; and
there are people working on adding it to btrfs (to support wine in
specific products).

> Of course, "do it in shared generic code" doesn't tend to really fix
> the braindamage, but at least it's now shared braindamage and not
> spread out all over. I'm looking at things like
> generic_ci_d_compare(), and it hurts to see the mindless "let's do
> lookups and compares one utf8 character at a time". What a disgrace.
> Somebody either *really* didn't care, or was a Unicode person who
> didn't understand the point of UTF-8.

Yes. I saw the rest of the thread and you are obviously correct here.
It needs to be fixed.  I will follow up with patches.

> The patches look fine to me. Al - do you even care about them?

I saw that Al Viro answered. Thank you, Al. So I'll wait for either his
review or the merge window.

Thanks,

-- 
Gabriel Krisman Bertazi