Skip to contentSkip to conversation

ceph: set sec_context xattr on symlink creation

PATCH In review Jeff Layton 2 participants 5 messages Jul 28, 2020 – Aug 4, 2020 Ceph client

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ceph: set sec_context xattr on symlink creation

Symlink inodes should have the security context set in their xattrs on creation. We already set the context on creation, but we don't attach the pagelist. Make it do so.

---
 fs/ceph/dir.c | 4 ++++
 1 file changed, 4 insertions(+)
fs/ceph/dir.c +4 -0
index 39f5311404b0..060bdcc5ce32 100644
@@ -930,6 +930,10 @@ static int ceph_symlink(struct inode *dir, struct dentry *dentry,
 	req->r_num_caps = 2;
 	req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
 	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+	if (as_ctx.pagelist) {
+		req->r_pagelist = as_ctx.pagelist;
+		as_ctx.pagelist = NULL;
+	}
 	err = ceph_mdsc_do_request(mdsc, dir, req);
 	if (!err && !req->r_reply_info.head->is_dentry)
 		err = ceph_handle_notrace_create(dir, dentry);
Signature · 1 line
2.26.2
maintainer
#1 lore raw
Quoted 15 lines · Jeff Layton

Symlink inodes should have the security context set in their xattrs on creation. We already set the context on creation, but we don't attach the pagelist. Make it do so.

---
 fs/ceph/dir.c | 4 ++++
 1 file changed, 4 insertions(+)
fs/ceph/dir.c +0 -0
index 39f5311404b0..060bdcc5ce32 100644

fs/ceph/dir.c static int ceph_symlink(struct inode *dir, struct dentry *dentry, · line 939

Quoted 10 lines
fs/ceph/dir.c +4 -0
@@ -930,6 +930,10 @@ static int ceph_symlink(struct inode *dir, struct dentry *dentry,
        req->r_num_caps = 2;
        req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
        req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+       if (as_ctx.pagelist) {
+               req->r_pagelist = as_ctx.pagelist;
+               as_ctx.pagelist = NULL;
+       }
        err = ceph_mdsc_do_request(mdsc, dir, req);
        if (!err && !req->r_reply_info.head->is_dentry)
                err = ceph_handle_notrace_create(dir, dentry);

What is the side effect? Should this go to stable?

Thanks,

Ilya

#2 lore raw
Quoted 19 lines · Ilya Dryomov
Quoted 18 lines · Jeff Layton

Symlink inodes should have the security context set in their xattrs on creation. We already set the context on creation, but we don't attach the pagelist. Make it do so.

---
 fs/ceph/dir.c | 4 ++++
 1 file changed, 4 insertions(+)
fs/ceph/dir.c +1 -0
index 39f5311404b0..060bdcc5ce32 100644
@@ -930,6 +930,10 @@ static int ceph_symlink(struct inode *dir, struct dentry *dentry,
        req->r_num_caps = 2;
        req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
        req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+       if (as_ctx.pagelist) {
Quoted 9 lines

fs/ceph/dir.c static int ceph_symlink(struct inode *dir, struct dentry *dentry, · line 939

Quoted 6 lines
fs/ceph/dir.c +3 -0
+               req->r_pagelist = as_ctx.pagelist;
+               as_ctx.pagelist = NULL;
+       }
        err = ceph_mdsc_do_request(mdsc, dir, req);
        if (!err && !req->r_reply_info.head->is_dentry)
                err = ceph_handle_notrace_create(dir, dentry);

What is the side effect? Should this go to stable?

The effect is that symlink inodes don't get an SELinux context set on them at creation, so they end up unlabeled instead of inheriting the proper context. As to the severity, it really depends on what ends up being unlabeled.

It's probably harmless enough to put this into stable, but I only noticed it by inspection, so I'm not sure it meets the "it must fix a real bug that bothers people" criterion.

Signature · 1 line
Jeff Layton <[email protected]>
#3 lore raw
Quoted 30 lines · Jeff Layton
Quoted 19 lines · Ilya Dryomov
Quoted 18 lines · Jeff Layton

Symlink inodes should have the security context set in their xattrs on creation. We already set the context on creation, but we don't attach the pagelist. Make it do so.

---
 fs/ceph/dir.c | 4 ++++
 1 file changed, 4 insertions(+)
fs/ceph/dir.c +1 -0
index 39f5311404b0..060bdcc5ce32 100644
@@ -930,6 +930,10 @@ static int ceph_symlink(struct inode *dir, struct dentry *dentry,
        req->r_num_caps = 2;
        req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
        req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+       if (as_ctx.pagelist) {
Quoted 9 lines

fs/ceph/dir.c static int ceph_symlink(struct inode *dir, struct dentry *dentry, · line 939

Quoted 6 lines
fs/ceph/dir.c +3 -0
+               req->r_pagelist = as_ctx.pagelist;
+               as_ctx.pagelist = NULL;
+       }
        err = ceph_mdsc_do_request(mdsc, dir, req);
        if (!err && !req->r_reply_info.head->is_dentry)
                err = ceph_handle_notrace_create(dir, dentry);

What is the side effect? Should this go to stable?

Quoted 8 lines

The effect is that symlink inodes don't get an SELinux context set on them at creation, so they end up unlabeled instead of inheriting the proper context. As to the severity, it really depends on what ends up being unlabeled.

It's probably harmless enough to put this into stable, but I only noticed it by inspection, so I'm not sure it meets the "it must fix a real bug that bothers people" criterion.

After thinking about it some more, let's do go ahead and mark this for stable. While no one has complained about it, it's a subtle bug that could be problematic once people start populating cephfs trees with unlabeled symlinks. Better that we fix it early before SELinux support becomes even more widespread.

Ilya, can you add the Cc: stable tag before you send a PR to Linus?

Thanks,

Signature · 1 line
Jeff Layton <[email protected]>
maintainer
#4 lore raw
Quoted 41 lines · Jeff Layton
Quoted 30 lines · Jeff Layton
Quoted 19 lines · Ilya Dryomov
Quoted 18 lines · Jeff Layton

Symlink inodes should have the security context set in their xattrs on creation. We already set the context on creation, but we don't attach the pagelist. Make it do so.

---
 fs/ceph/dir.c | 4 ++++
 1 file changed, 4 insertions(+)
fs/ceph/dir.c +1 -0
index 39f5311404b0..060bdcc5ce32 100644
@@ -930,6 +930,10 @@ static int ceph_symlink(struct inode *dir, struct dentry *dentry,
        req->r_num_caps = 2;
        req->r_dentry_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_AUTH_EXCL;
        req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
+       if (as_ctx.pagelist) {
Quoted 9 lines

fs/ceph/dir.c static int ceph_symlink(struct inode *dir, struct dentry *dentry, · line 939

Quoted 6 lines
fs/ceph/dir.c +3 -0
+               req->r_pagelist = as_ctx.pagelist;
+               as_ctx.pagelist = NULL;
+       }
        err = ceph_mdsc_do_request(mdsc, dir, req);
        if (!err && !req->r_reply_info.head->is_dentry)
                err = ceph_handle_notrace_create(dir, dentry);

What is the side effect? Should this go to stable?

Quoted 8 lines

The effect is that symlink inodes don't get an SELinux context set on them at creation, so they end up unlabeled instead of inheriting the proper context. As to the severity, it really depends on what ends up being unlabeled.

It's probably harmless enough to put this into stable, but I only noticed it by inspection, so I'm not sure it meets the "it must fix a real bug that bothers people" criterion.

Quoted 7 lines

After thinking about it some more, let's do go ahead and mark this for stable. While no one has complained about it, it's a subtle bug that could be problematic once people start populating cephfs trees with unlabeled symlinks. Better that we fix it early before SELinux support becomes even more widespread.

Ilya, can you add the Cc: stable tag before you send a PR to Linus?

Sure, will do.

Thanks,

Ilya