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

[Gabriel Krisman Bertazi <krisman@suse.de>]

Christian Brauner <brauner@kernel.org> writes:

> On Wed, 16 Aug 2023 01:07:54 -0400, Gabriel Krisman Bertazi wrote:
>> This is v6 of the negative dentry on case-insensitive directories.
>> Thanks Eric for the review of the last iteration.  This version
>> drops the patch to expose the helper to check casefolding directories,
>> since it is not necessary in ecryptfs and it might be going away.  It
>> also addresses some documentation details, fix a build bot error and
>> simplifies the commit messages.  See the changelog in each patch for
>> more details.
>> 
>> [...]
>
> Ok, let's put it into -next so it sees some testing.
> So it's too late for v6.7. Seems we forgot about this series.
> Sorry about that.

Christian,

We are approaching -rc2 and, until last Friday, it didn't shown up in
linux-next. So, to avoid turning a 6 month delay into 9 months, I pushed
your signed tag to linux-next myself.

That obviously uncovered a merge conflict: in v6.6, ceph added fscrypt,
and the caller had to be updated.  I fixed it and pushed again to
linux-next to get more testing.

Now, I don't want to send it to Linus myself. This is 100% VFS/FS code,
I'm not the maintainer and it will definitely raise eyebrows.  Can you
please requeue and make sure it goes through this time?  I'm happy to
drop my branch from linux-next once yours shows up.

  https://git.kernel.org/pub/scm/linux/kernel/git/krisman/unicode.git/log/?h=negative-dentries

This branch has the latest version with the ceph conflict folded in.  I
did it this way because I'd consider it was never picked up and there is
no point in making the history complex by adding a fix on top of your
signed tag, since it already fails to build ceph.

I can send it as a v7; but I prefer you just pull from the branch
above. Or you can ack and I'll send to Linus.

This is the diff from you signed tag:

diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 629d8fb31d8f..21278a9d9baa 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -1869,7 +1869,7 @@ static int ceph_d_revalidate(struct dentry *dentry, const struct qstr *name,
        struct inode *dir, *inode;
        struct ceph_mds_client *mdsc;
 
-       valid = fscrypt_d_revalidate(dentry, flags);
+       valid = fscrypt_d_revalidate(dentry, name, flags);
        if (valid <= 0)
                return valid;
 
diff --git a/fs/ecryptfs/dentry.c b/fs/ecryptfs/dentry.c
index 56093648d838..ce86891a1711 100644
--- a/fs/ecryptfs/dentry.c
+++ b/fs/ecryptfs/dentry.c
@@ -18,6 +18,7 @@
 /**
  * ecryptfs_d_revalidate - revalidate an ecryptfs dentry
  * @dentry: The ecryptfs dentry
+ * @name: The name under lookup
  * @flags: lookup flags
  *
  * Called when the VFS needs to revalidate a dentry. This
diff --git a/fs/gfs2/dentry.c b/fs/gfs2/dentry.c
index 3dd93d36aaf2..5e4910e016a8 100644
--- a/fs/gfs2/dentry.c
+++ b/fs/gfs2/dentry.c
@@ -22,6 +22,7 @@
 /**
  * gfs2_drevalidate - Check directory lookup consistency
  * @dentry: the mapping to check
+ * @name: The name under lookup
  * @flags: lookup flags
  *
  * Check to make sure the lookup necessary to arrive at this inode from its

-- 
Gabriel Krisman Bertazi