From: Luis Henriques <lhenriques@suse.de>
To: Jeff Layton <jlayton@kernel.org>
Cc: ceph-devel@vger.kernel.org, linux-fscrypt@vger.kernel.org,
linux-fsdevel@vger.kernel.org
Subject: Re: [RFC PATCH v5 19/19] ceph: add fscrypt ioctls
Date: Tue, 6 Apr 2021 17:24:05 +0100 [thread overview]
Message-ID: <YGyLJcqhpU5gGjsW@suse.de> (raw)
In-Reply-To: <ee49d17b2087d0f52c38931f13e648ee7a762b4f.camel@kernel.org>
On Tue, Apr 06, 2021 at 12:03:27PM -0400, Jeff Layton wrote:
> On Tue, 2021-04-06 at 16:38 +0100, Luis Henriques wrote:
> > Hi Jeff!
> >
> > On Fri, Mar 26, 2021 at 01:32:27PM -0400, Jeff Layton wrote:
> > > We gate most of the ioctls on MDS feature support. The exception is the
> > > key removal and status functions that we still want to work if the MDS's
> > > were to (inexplicably) lose the feature.
> > >
> > > For the set_policy ioctl, we take Fcx caps to ensure that nothing can
> > > create files in the directory while the ioctl is running. That should
> > > be enough to ensure that the "empty_dir" check is reliable.
> > >
> > > Signed-off-by: Jeff Layton <jlayton@kernel.org>
> > > ---
> > > fs/ceph/ioctl.c | 94 +++++++++++++++++++++++++++++++++++++++++++++++++
> > > 1 file changed, 94 insertions(+)
> > >
> > > diff --git a/fs/ceph/ioctl.c b/fs/ceph/ioctl.c
> > > index 6e061bf62ad4..34b85bcfcfc7 100644
> > > --- a/fs/ceph/ioctl.c
> > > +++ b/fs/ceph/ioctl.c
> > > @@ -6,6 +6,7 @@
> > > #include "mds_client.h"
> > > #include "ioctl.h"
> > > #include <linux/ceph/striper.h>
> > > +#include <linux/fscrypt.h>
> > >
> > > /*
> > > * ioctls
> > > @@ -268,8 +269,56 @@ static long ceph_ioctl_syncio(struct file *file)
> > > return 0;
> > > }
> > >
> > > +static int vet_mds_for_fscrypt(struct file *file)
> > > +{
> > > + int i, ret = -EOPNOTSUPP;
> > > + struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(file_inode(file)->i_sb);
> > > +
> > > + mutex_lock(&mdsc->mutex);
> > > + for (i = 0; i < mdsc->max_sessions; i++) {
> > > + struct ceph_mds_session *s = mdsc->sessions[i];
> > > +
> > > + if (!s)
> > > + continue;
> > > + if (test_bit(CEPHFS_FEATURE_ALTERNATE_NAME, &s->s_features))
> > > + ret = 0;
> > > + break;
> > > + }
> > > + mutex_unlock(&mdsc->mutex);
> > > + return ret;
> > > +}
> > > +
> > > +static long ceph_set_encryption_policy(struct file *file, unsigned long arg)
> > > +{
> > > + int ret, got = 0;
> > > + struct page *page = NULL;
> > > + struct inode *inode = file_inode(file);
> > > + struct ceph_inode_info *ci = ceph_inode(inode);
> > > +
> > > + ret = vet_mds_for_fscrypt(file);
> > > + if (ret)
> > > + return ret;
> > > +
> > > + /*
> > > + * Ensure we hold these caps so that we _know_ that the rstats check
> > > + * in the empty_dir check is reliable.
> > > + */
> > > + ret = ceph_get_caps(file, CEPH_CAP_FILE_SHARED, 0, -1, &got, &page);
> > > + if (ret)
> > > + return ret;
> > > + if (page)
> > > + put_page(page);
> > > + ret = fscrypt_ioctl_set_policy(file, (const void __user *)arg);
> > > + if (got)
> > > + ceph_put_cap_refs(ci, got);
> > > + return ret;
> > > +}
> > > +
> > > long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
> > > {
> > > + int ret;
> > > + struct ceph_inode_info *ci = ceph_inode(file_inode(file));
> > > +
> > > dout("ioctl file %p cmd %u arg %lu\n", file, cmd, arg);
> > > switch (cmd) {
> > > case CEPH_IOC_GET_LAYOUT:
> > > @@ -289,6 +338,51 @@ long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
> > >
> > > case CEPH_IOC_SYNCIO:
> > > return ceph_ioctl_syncio(file);
> > > +
> > > + case FS_IOC_SET_ENCRYPTION_POLICY:
> > > + return ceph_set_encryption_policy(file, arg);
> > > +
> > > + case FS_IOC_GET_ENCRYPTION_POLICY:
> > > + ret = vet_mds_for_fscrypt(file);
> > > + if (ret)
> > > + return ret;
> > > + return fscrypt_ioctl_get_policy(file, (void __user *)arg);
> > > +
> > > + case FS_IOC_GET_ENCRYPTION_POLICY_EX:
> > > + ret = vet_mds_for_fscrypt(file);
> > > + if (ret)
> > > + return ret;
> > > + return fscrypt_ioctl_get_policy_ex(file, (void __user *)arg);
> > > +
> > > + case FS_IOC_ADD_ENCRYPTION_KEY:
> > > + ret = vet_mds_for_fscrypt(file);
> > > + if (ret)
> > > + return ret;
> > > + atomic_inc(&ci->i_shared_gen);
> >
> > I've spent a few hours already looking at the bug I reported before, and I
> > can't really understand this code. What does it mean to increment
> > ->i_shared_gen at this point?
> >
> > The reason I'm asking is because it looks like the problem I'm seeing goes
> > away if I remove this code. Here's what I'm doing/seeing:
> >
> > # mount ...
> > # fscrypt unlock d
> >
> > -> 'd' dentry is eventually pruned at this point *if* ->i_shared_gen was
> > incremented by the line above.
> >
> > # cat d/f
> >
> > -> when ceph_fill_inode() is executed, 'd' isn't *not* set as encrypted
> > because both ci->i_xattrs.version and info->xattr_version are both
> > set to 0.
> >
>
> Interesting. That sounds like it might be the bug right there. "d"
> should clearly have a fscrypt context in its xattrs at that point. If
> the MDS isn't passing that back, then that could be a problem.
>
> I had a concern about that when I was developing this, and I *thought*
> Zheng had assured us that the MDS will always pass along the xattr blob
> in a trace. Maybe that's not correct?
Hmm, that's what I thought too. I was hoping not having to go look at the
MDS, but seems like I'll have to :-)
> > cat: d/f: No such file or directory
> >
> > I'm not sure anymore if the issue is on the client or on the MDS side.
> > Before digging deeper, I wonder if this ring any bell. ;-)
> >
> >
>
> No, this is not something I've seen before.
>
> Dentries that live in a directory have a copy of the i_shared_gen of the
> directory when they are instantiated. Bumping that value on a directory
> should basically ensure that its child dentries end up invalidated,
> which is what we want once we add the key to the directory. Once we add
> a key, any old dentries in that directory are no longer valid.
>
> That said, I could certainly have missed some subtlety here.
Great, thanks for clarifying. This should help me investigate a little
bit more.
[ And I'm also surprised you don't see this behaviour as it's very easy to
reproduce. ]
Cheers,
--
Luís
> >
> > > + ceph_dir_clear_ordered(file_inode(file));
> > > + ceph_dir_clear_complete(file_inode(file));
> > > + return fscrypt_ioctl_add_key(file, (void __user *)arg);
> > > +
> > > + case FS_IOC_REMOVE_ENCRYPTION_KEY:
> > > + atomic_inc(&ci->i_shared_gen);
> > > + ceph_dir_clear_ordered(file_inode(file));
> > > + ceph_dir_clear_complete(file_inode(file));
> > > + return fscrypt_ioctl_remove_key(file, (void __user *)arg);
> > > +
> > > + case FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS:
> > > + atomic_inc(&ci->i_shared_gen);
> > > + ceph_dir_clear_ordered(file_inode(file));
> > > + ceph_dir_clear_complete(file_inode(file));
> > > + return fscrypt_ioctl_remove_key_all_users(file, (void __user *)arg);
> > > +
> > > + case FS_IOC_GET_ENCRYPTION_KEY_STATUS:
> > > + return fscrypt_ioctl_get_key_status(file, (void __user *)arg);
> > > +
> > > + case FS_IOC_GET_ENCRYPTION_NONCE:
> > > + ret = vet_mds_for_fscrypt(file);
> > > + if (ret)
> > > + return ret;
> > > + return fscrypt_ioctl_get_nonce(file, (void __user *)arg);
> > > }
> > >
> > > return -ENOTTY;
> > > --
> > > 2.30.2
> > >
>
> --
> Jeff Layton <jlayton@kernel.org>
>
next prev parent reply other threads:[~2021-04-06 16:22 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-26 17:32 [RFC PATCH v5 00/19] ceph+fscrypt: context, filename and symlink support Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 01/19] vfs: export new_inode_pseudo Jeff Layton
2021-04-08 1:08 ` Eric Biggers
2021-04-08 16:18 ` Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 02/19] fscrypt: export fscrypt_base64_encode and fscrypt_base64_decode Jeff Layton
2021-04-08 1:06 ` Eric Biggers
2021-04-08 16:22 ` Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 03/19] fscrypt: export fscrypt_fname_encrypt and fscrypt_fname_encrypted_size Jeff Layton
2021-04-08 1:19 ` Eric Biggers
2021-03-26 17:32 ` [RFC PATCH v5 04/19] fscrypt: add fscrypt_context_for_new_inode Jeff Layton
2021-04-08 1:21 ` Eric Biggers
2021-04-08 16:27 ` Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 05/19] ceph: crypto context handling for ceph Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 06/19] ceph: implement -o test_dummy_encryption mount option Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 07/19] ceph: preallocate inode for ops that may create one Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 08/19] ceph: add routine to create fscrypt context prior to RPC Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 09/19] ceph: make ceph_msdc_build_path use ref-walk Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 10/19] ceph: add encrypted fname handling to ceph_mdsc_build_path Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 11/19] ceph: decode alternate_name in lease info Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 12/19] ceph: send altname in MClientRequest Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 13/19] ceph: properly set DCACHE_NOKEY_NAME flag in lookup Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 14/19] ceph: make d_revalidate call fscrypt revalidator for encrypted dentries Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 15/19] ceph: add helpers for converting names for userland presentation Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 16/19] ceph: add fscrypt support to ceph_fill_trace Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 17/19] ceph: add support to readdir for encrypted filenames Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 18/19] ceph: create symlinks with encrypted and base64-encoded targets Jeff Layton
2021-03-26 17:32 ` [RFC PATCH v5 19/19] ceph: add fscrypt ioctls Jeff Layton
2021-04-06 15:38 ` Luis Henriques
2021-04-06 16:03 ` Jeff Layton
2021-04-06 16:24 ` Luis Henriques [this message]
2021-04-06 17:27 ` Jeff Layton
2021-04-06 18:04 ` Luis Henriques
2021-04-07 12:47 ` Jeff Layton
2021-03-26 18:38 ` [RFC PATCH v5 00/19] ceph+fscrypt: context, filename and symlink support Jeff Layton
2021-03-31 20:35 ` [RFC PATCH v5 20/19] ceph: make ceph_get_name decrypt filenames Jeff Layton
2021-04-01 11:14 ` Luis Henriques
2021-04-01 12:15 ` Jeff Layton
2021-04-01 13:05 ` Luis Henriques
2021-04-01 13:12 ` Jeff Layton
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=YGyLJcqhpU5gGjsW@suse.de \
--to=lhenriques@suse.de \
--cc=ceph-devel@vger.kernel.org \
--cc=jlayton@kernel.org \
--cc=linux-fscrypt@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).