Changes in v3: - rename the hook to "kernfs_init_security" - change the hook interface to simply pass pointers to struct iattr and struct simple_xattrs of both the new node and its parent - add full security xattr support to kernfs (and fixup SELinux behavior to handle it properly) v2: https://lore.kernel.org/selinux/20190109162830.8309-1-omosnace@redhat.com/T/ Changes in v2: - add docstring for the new hook in union security_list_options - initialize *ctx to NULL and *ctxlen to 0 in case the hook is not implemented v1: https://lore.kernel.org/selinux/20190109091028.24485-1-omosnace@redhat.com/T/ TL;DR: This series adds a new security hook that allows to initialize the security context of kernfs properly, taking into account the parent context (and possibly other attributes). Kernfs nodes require special handling here, since they are not bound to specific inodes/superblocks, but instead represent the backing tree structure that is used to build the VFS tree when the kernfs tree is mounted. The kernfs nodes initially do not store any security context and rely on the LSM to assign some default context to inodes created over them. Kernfs inodes, however, allow setting an explicit context via the *setxattr(2) syscalls, in which case the context is stored inside the kernfs node's internal structure. SELinux (and possibly other LSMs) initialize the context of newly created FS objects based on the parent object's context (usually the child inherits the parent's context, unless the policy dictates otherwise). This is done by hooking the creation of the new inode corresponding to the newly created file/directory via security_inode_init_security() (most filesystems always create a fresh inode when a new FS object is created). However, kernfs nodes can be created "behind the scenes" while the filesystem is not mounted anywhere and thus no inodes can exist for them yet. Therefore, to allow maintaining similar behavior for kernfs nodes, a new LSM hook is needed, which will allow initializing the kernfs node's security context based on its own attributes and those of the parent's node. The main motivation for this change is that the userspace users of cgroupfs (which is built on kernfs) expect the usual security context inheritance to work under SELinux (see [1] and [2]). This functionality is required for better confinement of containers under SELinux. Patch 1/5 changes SELinux to fetch security context from extended attributes on kernfs filesystems, falling back to genfs-defined context if that fails. Without this patch the 2/5 would be a regression for SELinux (due to the removal of ...notifysecctx() call. Patch 2/5 implements full security xattr support in kernfs using simple_xattrs; patch 3/5 adds the new LSM hook; patch 4/5 implements the new hook in SELinux; and patch 5/5 modifies kernfs to call the new hook on new node creation. Testing: - passed the reproducer from the commit message of the last patch - passed SELinux testsuite on Fedora 29 (x86_64) when applied on top of current Rawhide kernel (5.0.0-0.rc4.git1.1) [3] - including the new proposed selinux-testsuite subtest [4] (adapted from the reproducer) [1] https://github.com/SELinuxProject/selinux-kernel/issues/39 [2] https://bugzilla.redhat.com/show_bug.cgi?id=1553803 [3] https://copr.fedorainfracloud.org/coprs/omos/kernel-testing/build/851570/ [4] https://github.com/SELinuxProject/selinux-testsuite/pull/48 Ondrej Mosnacek (5): selinux: try security xattr after genfs for kernfs filesystems kernfs: use simple_xattrs for security attributes LSM: add new hook for kernfs node initialization selinux: implement the kernfs_init_security hook kernfs: initialize security of newly created nodes fs/kernfs/dir.c | 42 +++++- fs/kernfs/inode.c | 102 ++++++------- fs/kernfs/kernfs-internal.h | 6 +- include/linux/lsm_hooks.h | 20 +++ include/linux/security.h | 13 ++ security/security.c | 9 ++ security/selinux/hooks.c | 220 +++++++++++++++++++--------- security/selinux/include/security.h | 1 + 8 files changed, 272 insertions(+), 141 deletions(-) -- 2.20.1
Since kernfs supports the security xattr handlers, we can simply use these to determine the inode's context, dropping the need to update it from kernfs explicitly using a security_inode_notifysecctx() call. We achieve this by setting a new sbsec flag SE_SBGENFS_XATTR to all mounts that are known to use kernfs under the hood and then fetching the xattrs after determining the fallback genfs sid in inode_doinit_with_dentry() when this flag is set. This will allow implementing full security xattr support in kernfs and removing the ...notifysecctx() call in a subsequent patch. Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com> --- security/selinux/hooks.c | 159 +++++++++++++++------------- security/selinux/include/security.h | 1 + 2 files changed, 88 insertions(+), 72 deletions(-) diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c index 81e012c66d95..758a99d1086e 100644 --- a/security/selinux/hooks.c +++ b/security/selinux/hooks.c @@ -793,11 +793,13 @@ static int selinux_set_mnt_opts(struct super_block *sb, if (!strcmp(sb->s_type->name, "debugfs") || !strcmp(sb->s_type->name, "tracefs") || - !strcmp(sb->s_type->name, "sysfs") || - !strcmp(sb->s_type->name, "pstore") || + !strcmp(sb->s_type->name, "pstore")) + sbsec->flags |= SE_SBGENFS; + + if (!strcmp(sb->s_type->name, "sysfs") || !strcmp(sb->s_type->name, "cgroup") || !strcmp(sb->s_type->name, "cgroup2")) - sbsec->flags |= SE_SBGENFS; + sbsec->flags |= SE_SBGENFS | SE_SBGENFS_XATTR; if (!sbsec->behavior) { /* @@ -1392,6 +1394,71 @@ static int selinux_genfs_get_sid(struct dentry *dentry, return rc; } +static int inode_doinit_use_xattr(struct inode *inode, struct dentry *dentry, + u32 def_sid, u32 *sid) +{ +#define INITCONTEXTLEN 255 + char *context = NULL; + unsigned int len = 0; + int rc; + + *sid = def_sid; + + if (!(inode->i_opflags & IOP_XATTR)) + return 0; + + len = INITCONTEXTLEN; + context = kmalloc(len + 1, GFP_NOFS); + if (!context) + return -ENOMEM; + + context[len] = '\0'; + rc = __vfs_getxattr(dentry, inode, XATTR_NAME_SELINUX, context, len); + if (rc == -ERANGE) { + kfree(context); + + /* Need a larger buffer. Query for the right size. */ + rc = __vfs_getxattr(dentry, inode, XATTR_NAME_SELINUX, NULL, 0); + if (rc < 0) + return rc; + + len = rc; + context = kmalloc(len + 1, GFP_NOFS); + if (!context) + return -ENOMEM; + + context[len] = '\0'; + rc = __vfs_getxattr(dentry, inode, XATTR_NAME_SELINUX, + context, len); + } + if (rc < 0) { + kfree(context); + if (rc != -ENODATA) { + pr_warn("SELinux: %s: getxattr returned %d for dev=%s ino=%ld\n", + __func__, -rc, inode->i_sb->s_id, inode->i_ino); + return rc; + } + return 0; + } + + rc = security_context_to_sid_default(&selinux_state, context, rc, sid, + def_sid, GFP_NOFS); + if (rc) { + char *dev = inode->i_sb->s_id; + unsigned long ino = inode->i_ino; + + if (rc == -EINVAL) { + pr_notice_ratelimited("SELinux: inode=%lu on dev=%s was found to have an invalid context=%s. This indicates you may need to relabel the inode or the filesystem in question.\n", + ino, dev, context); + } else { + pr_warn("SELinux: %s: context_to_sid(%s) returned %d for dev=%s ino=%ld\n", + __func__, context, -rc, dev, ino); + } + } + kfree(context); + return 0; +} + /* The inode's security attributes must be initialized before first use. */ static int inode_doinit_with_dentry(struct inode *inode, struct dentry *opt_dentry) { @@ -1401,8 +1468,6 @@ static int inode_doinit_with_dentry(struct inode *inode, struct dentry *opt_dent u16 sclass; struct dentry *dentry; #define INITCONTEXTLEN 255 - char *context = NULL; - unsigned len = 0; int rc = 0; if (isec->initialized == LABEL_INITIALIZED) @@ -1470,72 +1535,11 @@ static int inode_doinit_with_dentry(struct inode *inode, struct dentry *opt_dent goto out; } - len = INITCONTEXTLEN; - context = kmalloc(len+1, GFP_NOFS); - if (!context) { - rc = -ENOMEM; - dput(dentry); - goto out; - } - context[len] = '\0'; - rc = __vfs_getxattr(dentry, inode, XATTR_NAME_SELINUX, context, len); - if (rc == -ERANGE) { - kfree(context); - - /* Need a larger buffer. Query for the right size. */ - rc = __vfs_getxattr(dentry, inode, XATTR_NAME_SELINUX, NULL, 0); - if (rc < 0) { - dput(dentry); - goto out; - } - len = rc; - context = kmalloc(len+1, GFP_NOFS); - if (!context) { - rc = -ENOMEM; - dput(dentry); - goto out; - } - context[len] = '\0'; - rc = __vfs_getxattr(dentry, inode, XATTR_NAME_SELINUX, context, len); - } + rc = inode_doinit_use_xattr(inode, dentry, sbsec->def_sid, + &sid); dput(dentry); - if (rc < 0) { - if (rc != -ENODATA) { - pr_warn("SELinux: %s: getxattr returned " - "%d for dev=%s ino=%ld\n", __func__, - -rc, inode->i_sb->s_id, inode->i_ino); - kfree(context); - goto out; - } - /* Map ENODATA to the default file SID */ - sid = sbsec->def_sid; - rc = 0; - } else { - rc = security_context_to_sid_default(&selinux_state, - context, rc, &sid, - sbsec->def_sid, - GFP_NOFS); - if (rc) { - char *dev = inode->i_sb->s_id; - unsigned long ino = inode->i_ino; - - if (rc == -EINVAL) { - if (printk_ratelimit()) - pr_notice("SELinux: inode=%lu on dev=%s was found to have an invalid " - "context=%s. This indicates you may need to relabel the inode or the " - "filesystem in question.\n", ino, dev, context); - } else { - pr_warn("SELinux: %s: context_to_sid(%s) " - "returned %d for dev=%s ino=%ld\n", - __func__, context, -rc, dev, ino); - } - kfree(context); - /* Leave with the unlabeled SID */ - rc = 0; - break; - } - } - kfree(context); + if (rc) + goto out; break; case SECURITY_FS_USE_TASK: sid = task_sid; @@ -1586,9 +1590,20 @@ static int inode_doinit_with_dentry(struct inode *inode, struct dentry *opt_dent goto out; rc = selinux_genfs_get_sid(dentry, sclass, sbsec->flags, &sid); - dput(dentry); - if (rc) + if (rc) { + dput(dentry); goto out; + } + + if (sbsec->flags & SE_SBGENFS_XATTR) { + rc = inode_doinit_use_xattr(inode, dentry, + sid, &sid); + if (rc) { + dput(dentry); + goto out; + } + } + dput(dentry); } break; } diff --git a/security/selinux/include/security.h b/security/selinux/include/security.h index f68fb25b5702..6e5928f951da 100644 --- a/security/selinux/include/security.h +++ b/security/selinux/include/security.h @@ -58,6 +58,7 @@ #define SE_SBINITIALIZED 0x0100 #define SE_SBPROC 0x0200 #define SE_SBGENFS 0x0400 +#define SE_SBGENFS_XATTR 0x0800 #define CONTEXT_STR "context=" #define FSCONTEXT_STR "fscontext=" -- 2.20.1
Replace the special handling of security xattrs with simple_xattrs, as is already done for the trusted xattrs. This simplifies the code and allows LSMs to use more than just a single xattr to do their business. Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com> --- fs/kernfs/dir.c | 7 ++- fs/kernfs/inode.c | 100 +++++++++++++++--------------------- fs/kernfs/kernfs-internal.h | 5 +- 3 files changed, 46 insertions(+), 66 deletions(-) diff --git a/fs/kernfs/dir.c b/fs/kernfs/dir.c index 4ca0b5c18192..ad7e3356bcc5 100644 --- a/fs/kernfs/dir.c +++ b/fs/kernfs/dir.c @@ -532,11 +532,10 @@ void kernfs_put(struct kernfs_node *kn) kfree_const(kn->name); if (kn->iattr) { - if (kn->iattr->ia_secdata) - security_release_secctx(kn->iattr->ia_secdata, - kn->iattr->ia_secdata_len); - simple_xattrs_free(&kn->iattr->xattrs); + simple_xattrs_free(&kn->iattr->xattrs_trusted); + simple_xattrs_free(&kn->iattr->xattrs_security); } + kfree(kn->iattr); spin_lock(&kernfs_idr_lock); idr_remove(&root->ino_idr, kn->id.ino); diff --git a/fs/kernfs/inode.c b/fs/kernfs/inode.c index 80cebcd94c90..f0e2cb4379c0 100644 --- a/fs/kernfs/inode.c +++ b/fs/kernfs/inode.c @@ -56,7 +56,8 @@ static struct kernfs_iattrs *kernfs_iattrs(struct kernfs_node *kn) iattrs->ia_mtime = iattrs->ia_atime; iattrs->ia_ctime = iattrs->ia_atime; - simple_xattrs_init(&kn->iattr->xattrs); + simple_xattrs_init(&kn->iattr->xattrs_trusted); + simple_xattrs_init(&kn->iattr->xattrs_security); out_unlock: ret = kn->iattr; mutex_unlock(&iattr_mutex); @@ -135,33 +136,31 @@ out: return error; } -static int kernfs_node_setsecdata(struct kernfs_iattrs *attrs, void **secdata, - u32 *secdata_len) -{ - void *old_secdata; - size_t old_secdata_len; - - old_secdata = attrs->ia_secdata; - old_secdata_len = attrs->ia_secdata_len; - - attrs->ia_secdata = *secdata; - attrs->ia_secdata_len = *secdata_len; - - *secdata = old_secdata; - *secdata_len = old_secdata_len; - return 0; -} - ssize_t kernfs_iop_listxattr(struct dentry *dentry, char *buf, size_t size) { struct kernfs_node *kn = kernfs_dentry_node(dentry); + struct inode *inode = d_inode(dentry); struct kernfs_iattrs *attrs; + ssize_t ret, length = 0; attrs = kernfs_iattrs(kn); if (!attrs) return -ENOMEM; - return simple_xattr_list(d_inode(dentry), &attrs->xattrs, buf, size); + ret = simple_xattr_list(inode, &attrs->xattrs_trusted, buf, size); + if (ret < 0) + return ret; + length += ret; + + buf += ret; + size -= ret; + + ret = simple_xattr_list(inode, &attrs->xattrs_security, buf, size); + if (ret < 0) + return ret; + length += ret; + + return length; } static inline void set_default_inode_attr(struct inode *inode, umode_t mode) @@ -186,15 +185,12 @@ static void kernfs_refresh_inode(struct kernfs_node *kn, struct inode *inode) struct kernfs_iattrs *attrs = kn->iattr; inode->i_mode = kn->mode; - if (attrs) { + if (attrs) /* * kernfs_node has non-default attributes get them from * persistent copy in kernfs_node. */ set_inode_attr(inode, &attrs->ia_iattr); - security_inode_notifysecctx(inode, attrs->ia_secdata, - attrs->ia_secdata_len); - } if (kernfs_type(kn) == KERNFS_DIR) set_nlink(inode, kn->dir.subdirs + 2); @@ -305,19 +301,29 @@ int kernfs_iop_permission(struct inode *inode, int mask) return generic_permission(inode, mask); } +static const struct xattr_handler kernfs_trusted_xattr_handler; +static const struct xattr_handler kernfs_security_xattr_handler; + static int kernfs_xattr_get(const struct xattr_handler *handler, struct dentry *unused, struct inode *inode, const char *suffix, void *value, size_t size) { - const char *name = xattr_full_name(handler, suffix); struct kernfs_node *kn = inode->i_private; struct kernfs_iattrs *attrs; + struct simple_xattrs *xattrs; attrs = kernfs_iattrs(kn); if (!attrs) return -ENOMEM; - return simple_xattr_get(&attrs->xattrs, name, value, size); + if (handler == &kernfs_trusted_xattr_handler) + xattrs = &attrs->xattrs_trusted; + else if (handler == &kernfs_security_xattr_handler) + xattrs = &attrs->xattrs_security; + else + return -EINVAL; + + return simple_xattr_get(xattrs, suffix, value, size); } static int kernfs_xattr_set(const struct xattr_handler *handler, @@ -325,15 +331,22 @@ static int kernfs_xattr_set(const struct xattr_handler *handler, const char *suffix, const void *value, size_t size, int flags) { - const char *name = xattr_full_name(handler, suffix); struct kernfs_node *kn = inode->i_private; struct kernfs_iattrs *attrs; + struct simple_xattrs *xattrs; attrs = kernfs_iattrs(kn); if (!attrs) return -ENOMEM; - return simple_xattr_set(&attrs->xattrs, name, value, size, flags); + if (handler == &kernfs_trusted_xattr_handler) + xattrs = &attrs->xattrs_trusted; + else if (handler == &kernfs_security_xattr_handler) + xattrs = &attrs->xattrs_security; + else + return -EINVAL; + + return simple_xattr_set(xattrs, suffix, value, size, flags); } static const struct xattr_handler kernfs_trusted_xattr_handler = { @@ -342,41 +355,10 @@ static const struct xattr_handler kernfs_trusted_xattr_handler = { .set = kernfs_xattr_set, }; -static int kernfs_security_xattr_set(const struct xattr_handler *handler, - struct dentry *unused, struct inode *inode, - const char *suffix, const void *value, - size_t size, int flags) -{ - struct kernfs_node *kn = inode->i_private; - struct kernfs_iattrs *attrs; - void *secdata; - u32 secdata_len = 0; - int error; - - attrs = kernfs_iattrs(kn); - if (!attrs) - return -ENOMEM; - - error = security_inode_setsecurity(inode, suffix, value, size, flags); - if (error) - return error; - error = security_inode_getsecctx(inode, &secdata, &secdata_len); - if (error) - return error; - - mutex_lock(&kernfs_mutex); - error = kernfs_node_setsecdata(attrs, &secdata, &secdata_len); - mutex_unlock(&kernfs_mutex); - - if (secdata) - security_release_secctx(secdata, secdata_len); - return error; -} - static const struct xattr_handler kernfs_security_xattr_handler = { .prefix = XATTR_SECURITY_PREFIX, .get = kernfs_xattr_get, - .set = kernfs_security_xattr_set, + .set = kernfs_xattr_set, }; const struct xattr_handler *kernfs_xattr_handlers[] = { diff --git a/fs/kernfs/kernfs-internal.h b/fs/kernfs/kernfs-internal.h index 3d83b114bb08..93bf1dcd0306 100644 --- a/fs/kernfs/kernfs-internal.h +++ b/fs/kernfs/kernfs-internal.h @@ -20,10 +20,9 @@ struct kernfs_iattrs { struct iattr ia_iattr; - void *ia_secdata; - u32 ia_secdata_len; - struct simple_xattrs xattrs; + struct simple_xattrs xattrs_trusted; + struct simple_xattrs xattrs_security; }; /* +1 to avoid triggering overflow warning when negating it */ -- 2.20.1
This patch introduces a new security hook that is intended for initializing the security data for newly created kernfs nodes, which provide a way of storing a non-default security context, but need to operate independently from mounts (and therefore may not have an associated inode at the moment of creation). The main motivation is to allow kernfs nodes to inherit the context of the parent under SELinux, similar to the behavior of security_inode_init_security(). Other LSMs may implement their own logic for handling the creation of new nodes. The interface of the new hook provides the following to the LSM: * a qstr containing the name of the new node * initial inode attributes (struct iattr + simple_xattrs) of the new node * inode attributes of the parent node (directory) Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com> --- include/linux/lsm_hooks.h | 20 ++++++++++++++++++++ include/linux/security.h | 13 +++++++++++++ security/security.c | 9 +++++++++ 3 files changed, 42 insertions(+) diff --git a/include/linux/lsm_hooks.h b/include/linux/lsm_hooks.h index 9a0bdf91e646..7a7ea511cc61 100644 --- a/include/linux/lsm_hooks.h +++ b/include/linux/lsm_hooks.h @@ -429,6 +429,19 @@ * to abort the copy up. Note that the caller is responsible for reading * and writing the xattrs as this hook is merely a filter. * + * Security hooks for kernfs node operations + * + * @kernfs_init_security + * Initialize the security context of a newlycreated kernfs node based + * on its own and its parent's attributes. The security context should + * be stored in the simple_xattrs structure as an extended attribute. + * + * @qstr contains the last path component of the new node. + * @iattr contains the inode attributes of the new node. + * @secattr is the list of security xattrs of the new node. + * @dir_iattr contains the inode attributes of the parent node. + * @dir_secattr is the list of security xattrs of the parent node. + * * Security hooks for file operations * * @file_permission: @@ -1558,6 +1571,12 @@ union security_list_options { int (*inode_copy_up)(struct dentry *src, struct cred **new); int (*inode_copy_up_xattr)(const char *name); + int (*kernfs_init_security)(const struct qstr *qstr, + struct iattr *iattr, + struct simple_xattrs *secattr, + struct iattr *dir_iattr, + struct simple_xattrs *dir_secattr); + int (*file_permission)(struct file *file, int mask); int (*file_alloc_security)(struct file *file); void (*file_free_security)(struct file *file); @@ -1858,6 +1877,7 @@ struct security_hook_heads { struct hlist_head inode_getsecid; struct hlist_head inode_copy_up; struct hlist_head inode_copy_up_xattr; + struct hlist_head kernfs_init_security; struct hlist_head file_permission; struct hlist_head file_alloc_security; struct hlist_head file_free_security; diff --git a/include/linux/security.h b/include/linux/security.h index dbfb5a66babb..92234c19727f 100644 --- a/include/linux/security.h +++ b/include/linux/security.h @@ -51,6 +51,7 @@ struct fown_struct; struct file_operations; struct msg_msg; struct xattr; +struct simple_xattrs; struct xfrm_sec_ctx; struct mm_struct; @@ -291,6 +292,10 @@ int security_inode_listsecurity(struct inode *inode, char *buffer, size_t buffer void security_inode_getsecid(struct inode *inode, u32 *secid); int security_inode_copy_up(struct dentry *src, struct cred **new); int security_inode_copy_up_xattr(const char *name); +int security_kernfs_init_security(const struct qstr *qstr, struct iattr *iattr, + struct simple_xattrs *secattr, + struct iattr *dir_iattr, + struct simple_xattrs *dir_secattr); int security_file_permission(struct file *file, int mask); int security_file_alloc(struct file *file); void security_file_free(struct file *file); @@ -783,6 +788,14 @@ static inline int security_inode_copy_up(struct dentry *src, struct cred **new) return 0; } +static inline int security_kernfs_init_security( + const struct qstr *qstr, struct iattr *iattr, + struct simple_xattrs *secattr, struct iattr *dir_iattr, + struct simple_xattrs *dir_secattr) +{ + return 0; +} + static inline int security_inode_copy_up_xattr(const char *name) { return -EOPNOTSUPP; diff --git a/security/security.c b/security/security.c index f1b8d2587639..3a79e0b53833 100644 --- a/security/security.c +++ b/security/security.c @@ -892,6 +892,15 @@ int security_inode_copy_up_xattr(const char *name) } EXPORT_SYMBOL(security_inode_copy_up_xattr); +int security_kernfs_init_security(const struct qstr *qstr, struct iattr *iattr, + struct simple_xattrs *secattr, + struct iattr *dir_iattr, + struct simple_xattrs *dir_secattr) +{ + return call_int_hook(kernfs_init_security, 0, qstr, iattr, secattr, + dir_iattr, dir_secattr); +} + int security_file_permission(struct file *file, int mask) { int ret; -- 2.20.1
The hook applies the same logic as selinux_determine_inode_label(), with the exception of the super_block handling, which will be enforced on the actual inodes later by other hooks. Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com> --- security/selinux/hooks.c | 61 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 61 insertions(+) diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c index 758a99d1086e..1f2239c191bd 100644 --- a/security/selinux/hooks.c +++ b/security/selinux/hooks.c @@ -3354,6 +3354,65 @@ static int selinux_inode_copy_up_xattr(const char *name) return -EOPNOTSUPP; } +/* kernfs node operations */ + +int selinux_kernfs_init_security(const struct qstr *qstr, struct iattr *iattr, + struct simple_xattrs *secattr, + struct iattr *dir_iattr, + struct simple_xattrs *dir_secattr) +{ + const struct task_security_struct *tsec = current_security(); + u32 parent_sid, newsid, clen; + int rc; + char *context; + + rc = simple_xattr_get(dir_secattr, XATTR_SELINUX_SUFFIX, NULL, 0); + if (rc == -ENODATA) + return 0; + else if (rc < 0) + return rc; + + clen = (u32)rc; + context = kmalloc(clen, GFP_KERNEL); + if (!context) + return -ENOMEM; + + rc = simple_xattr_get(dir_secattr, XATTR_SELINUX_SUFFIX, context, clen); + if (rc < 0) { + kfree(context); + return rc; + } + + rc = security_context_to_sid(&selinux_state, context, clen, &parent_sid, + GFP_KERNEL); + kfree(context); + if (rc) + return rc; + + if (tsec->create_sid) { + newsid = tsec->create_sid; + } else { + u16 secclass = inode_mode_to_security_class(iattr->ia_mode); + + rc = security_transition_sid(&selinux_state, tsec->sid, + parent_sid, secclass, qstr, + &newsid); + if (rc) + return rc; + } + + rc = security_sid_to_context_force(&selinux_state, newsid, + &context, &clen); + if (rc) + return rc; + + rc = simple_xattr_set(secattr, XATTR_SELINUX_SUFFIX, context, clen, + XATTR_CREATE); + kfree(context); + return rc; +} + + /* file security operations */ static int selinux_revalidate_file_permission(struct file *file, int mask) @@ -6800,6 +6859,8 @@ static struct security_hook_list selinux_hooks[] __lsm_ro_after_init = { LSM_HOOK_INIT(inode_copy_up, selinux_inode_copy_up), LSM_HOOK_INIT(inode_copy_up_xattr, selinux_inode_copy_up_xattr), + LSM_HOOK_INIT(kernfs_init_security, selinux_kernfs_init_security), + LSM_HOOK_INIT(file_permission, selinux_file_permission), LSM_HOOK_INIT(file_alloc_security, selinux_file_alloc_security), LSM_HOOK_INIT(file_free_security, selinux_file_free_security), -- 2.20.1
Use the new security_kernfs_init_security() hook to allow LSMs to possibly assign a non-default security context to a newly created kernfs node based on the attributes of the new node and also its parent node. This fixes an issue with cgroupfs under SELinux, where newly created cgroup subdirectories/files would not inherit its parent's context if it had been set explicitly to a non-default value (other than the genfs context specified by the policy). This can be reproduced as follows (on Fedora/RHEL): # mkdir /sys/fs/cgroup/unified/test # # Need permissive to change the label under Fedora policy: # setenforce 0 # chcon -t container_file_t /sys/fs/cgroup/unified/test # ls -lZ /sys/fs/cgroup/unified total 0 -r--r--r--. 1 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:06 cgroup.controllers -rw-r--r--. 1 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:06 cgroup.max.depth -rw-r--r--. 1 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:06 cgroup.max.descendants -rw-r--r--. 1 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:06 cgroup.procs -r--r--r--. 1 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:06 cgroup.stat -rw-r--r--. 1 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:06 cgroup.subtree_control -rw-r--r--. 1 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:06 cgroup.threads drwxr-xr-x. 2 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:06 init.scope drwxr-xr-x. 26 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:21 system.slice drwxr-xr-x. 3 root root system_u:object_r:container_file_t:s0 0 Jan 29 03:15 test drwxr-xr-x. 3 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:06 user.slice # mkdir /sys/fs/cgroup/unified/test/subdir Actual result: # ls -ldZ /sys/fs/cgroup/unified/test/subdir drwxr-xr-x. 2 root root system_u:object_r:cgroup_t:s0 0 Jan 29 03:15 /sys/fs/cgroup/unified/test/subdir Expected result: # ls -ldZ /sys/fs/cgroup/unified/test/subdir drwxr-xr-x. 2 root root unconfined_u:object_r:container_file_t:s0 0 Jan 29 03:15 /sys/fs/cgroup/unified/test/subdir Link: https://github.com/SELinuxProject/selinux-kernel/issues/39 Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com> --- fs/kernfs/dir.c | 35 +++++++++++++++++++++++++++++++++-- fs/kernfs/inode.c | 2 +- fs/kernfs/kernfs-internal.h | 1 + 3 files changed, 35 insertions(+), 3 deletions(-) diff --git a/fs/kernfs/dir.c b/fs/kernfs/dir.c index ad7e3356bcc5..797199a748d7 100644 --- a/fs/kernfs/dir.c +++ b/fs/kernfs/dir.c @@ -15,6 +15,7 @@ #include <linux/slab.h> #include <linux/security.h> #include <linux/hash.h> +#include <linux/stringhash.h> #include "kernfs-internal.h" @@ -616,7 +617,31 @@ struct kernfs_node *kernfs_node_from_dentry(struct dentry *dentry) return NULL; } +static int kernfs_node_init_security(struct kernfs_node *parent, + struct kernfs_node *kn) +{ + struct kernfs_iattrs *attrs, *pattrs; + struct qstr q; + + attrs = kernfs_iattrs(kn); + if (!attrs) + return -ENOMEM; + + pattrs = kernfs_iattrs(parent); + if (!pattrs) + return -ENOMEM; + + q.name = kn->name; + q.hash_len = hashlen_string(parent, kn->name); + + return security_kernfs_init_security(&q, &attrs->ia_iattr, + &attrs->xattrs_security, + &pattrs->ia_iattr, + &pattrs->xattrs_security); +} + static struct kernfs_node *__kernfs_new_node(struct kernfs_root *root, + struct kernfs_node *parent, const char *name, umode_t mode, kuid_t uid, kgid_t gid, unsigned flags) @@ -673,6 +698,12 @@ static struct kernfs_node *__kernfs_new_node(struct kernfs_root *root, goto err_out3; } + if (parent) { + ret = kernfs_node_init_security(parent, kn); + if (ret) + goto err_out3; + } + return kn; err_out3: @@ -691,7 +722,7 @@ struct kernfs_node *kernfs_new_node(struct kernfs_node *parent, { struct kernfs_node *kn; - kn = __kernfs_new_node(kernfs_root(parent), + kn = __kernfs_new_node(kernfs_root(parent), parent, name, mode, uid, gid, flags); if (kn) { kernfs_get(parent); @@ -961,7 +992,7 @@ struct kernfs_root *kernfs_create_root(struct kernfs_syscall_ops *scops, INIT_LIST_HEAD(&root->supers); root->next_generation = 1; - kn = __kernfs_new_node(root, "", S_IFDIR | S_IRUGO | S_IXUGO, + kn = __kernfs_new_node(root, NULL, "", S_IFDIR | S_IRUGO | S_IXUGO, GLOBAL_ROOT_UID, GLOBAL_ROOT_GID, KERNFS_DIR); if (!kn) { diff --git a/fs/kernfs/inode.c b/fs/kernfs/inode.c index f0e2cb4379c0..645c404b8644 100644 --- a/fs/kernfs/inode.c +++ b/fs/kernfs/inode.c @@ -31,7 +31,7 @@ static const struct inode_operations kernfs_iops = { .listxattr = kernfs_iop_listxattr, }; -static struct kernfs_iattrs *kernfs_iattrs(struct kernfs_node *kn) +struct kernfs_iattrs *kernfs_iattrs(struct kernfs_node *kn) { static DEFINE_MUTEX(iattr_mutex); struct kernfs_iattrs *ret; diff --git a/fs/kernfs/kernfs-internal.h b/fs/kernfs/kernfs-internal.h index 93bf1dcd0306..90215f8e503a 100644 --- a/fs/kernfs/kernfs-internal.h +++ b/fs/kernfs/kernfs-internal.h @@ -90,6 +90,7 @@ int kernfs_iop_getattr(const struct path *path, struct kstat *stat, u32 request_mask, unsigned int query_flags); ssize_t kernfs_iop_listxattr(struct dentry *dentry, char *buf, size_t size); int __kernfs_setattr(struct kernfs_node *kn, const struct iattr *iattr); +struct kernfs_iattrs *kernfs_iattrs(struct kernfs_node *kn); /* * dir.c -- 2.20.1
Hello,
On Wed, Jan 30, 2019 at 12:41:50PM +0100, Ondrej Mosnacek wrote:
> @@ -673,6 +698,12 @@ static struct kernfs_node *__kernfs_new_node(struct kernfs_root *root,
> goto err_out3;
> }
>
> + if (parent) {
> + ret = kernfs_node_init_security(parent, kn);
> + if (ret)
> + goto err_out3;
> + }
So, doing this unconditionally isn't a good idea. kernfs doesn't use
the usual dentry/inode because there are machines with 6, even 7 digit
number of kernfs nodes and some of them even failed to boot due to
memory shortage. Please don't blow it up by default.
Thanks.
--
tejun
Hi Tejun,
On Wed, Jan 30, 2019 at 6:09 PM Tejun Heo <tj@kernel.org> wrote:
>
> Hello,
>
> On Wed, Jan 30, 2019 at 12:41:50PM +0100, Ondrej Mosnacek wrote:
> > @@ -673,6 +698,12 @@ static struct kernfs_node *__kernfs_new_node(struct kernfs_root *root,
> > goto err_out3;
> > }
> >
> > + if (parent) {
> > + ret = kernfs_node_init_security(parent, kn);
> > + if (ret)
> > + goto err_out3;
> > + }
>
> So, doing this unconditionally isn't a good idea. kernfs doesn't use
> the usual dentry/inode because there are machines with 6, even 7 digit
> number of kernfs nodes and some of them even failed to boot due to
> memory shortage. Please don't blow it up by default.
Hm, I see... basically the only thing that gets allocated in
kernfs_node_init_security() by default (at least under SELinux/ no
LSM) is the kernfs_iattrs structures, so I assume you are pointing at
that. I think this can be easily fixed, if we again use the assumption
that whenever the parent node has only default attributes
(parent->iattrs == NULL), then the child node should also have just
default attributes (and so we don't need to call kernfs_iattrs() on it
nor call the security hook). Something along these lines:
[...]
+static int kernfs_node_init_security(struct kernfs_node *parent,
+ struct kernfs_node *kn)
+{
+ struct kernfs_iattrs *attrs, *pattrs;
+ struct qstr q;
+
+ pattrs = parent->iattrs;
+ if (!pattrs)
+ return 0;
+
+ attrs = kernfs_iattrs(kn);
+ if (!attrs)
+ return -ENOMEM;
+
+ q.name = kn->name;
+ q.hash_len = hashlen_string(parent, kn->name);
[...]
Technically this might make some LSMs unhappy, if they want to set
some non-default context even if parent is all default, but this is
already impossible now and in this case I think we have no better
choice than sacrificing a bit of flexibility for memory efficiency,
which is apparently critical here.
Tejun, Casey, would the above modification be fine with you?
--
Ondrej Mosnacek <omosnace at redhat dot com>
Associate Software Engineer, Security Technologies
Red Hat, Inc.
Hello, On Thu, Jan 31, 2019 at 11:20:57AM +0100, Ondrej Mosnacek wrote: > Hm, I see... basically the only thing that gets allocated in > kernfs_node_init_security() by default (at least under SELinux/ no > LSM) is the kernfs_iattrs structures, so I assume you are pointing at > that. I think this can be easily fixed, if we again use the assumption Yeap. > Technically this might make some LSMs unhappy, if they want to set > some non-default context even if parent is all default, but this is > already impossible now and in this case I think we have no better > choice than sacrificing a bit of flexibility for memory efficiency, > which is apparently critical here. > > Tejun, Casey, would the above modification be fine with you? Generally looks good but maybe it can check the attr to see whether there actually are things which need inheritance? Thanks. -- tejun
On 1/31/2019 2:20 AM, Ondrej Mosnacek wrote: > Hi Tejun, > > On Wed, Jan 30, 2019 at 6:09 PM Tejun Heo <tj@kernel.org> wrote: >> Hello, >> >> On Wed, Jan 30, 2019 at 12:41:50PM +0100, Ondrej Mosnacek wrote: >>> @@ -673,6 +698,12 @@ static struct kernfs_node *__kernfs_new_node(struct kernfs_root *root, >>> goto err_out3; >>> } >>> >>> + if (parent) { >>> + ret = kernfs_node_init_security(parent, kn); >>> + if (ret) >>> + goto err_out3; >>> + } >> So, doing this unconditionally isn't a good idea. kernfs doesn't use >> the usual dentry/inode because there are machines with 6, even 7 digit >> number of kernfs nodes and some of them even failed to boot due to >> memory shortage. Please don't blow it up by default. > Hm, I see... basically the only thing that gets allocated in > kernfs_node_init_security() by default (at least under SELinux/ no > LSM) is the kernfs_iattrs structures, so I assume you are pointing at > that. I think this can be easily fixed, if we again use the assumption > that whenever the parent node has only default attributes > (parent->iattrs == NULL), then the child node should also have just > default attributes (and so we don't need to call kernfs_iattrs() on it > nor call the security hook). Something along these lines: > > [...] > +static int kernfs_node_init_security(struct kernfs_node *parent, > + struct kernfs_node *kn) > +{ > + struct kernfs_iattrs *attrs, *pattrs; > + struct qstr q; > + > + pattrs = parent->iattrs; > + if (!pattrs) > + return 0; > + > + attrs = kernfs_iattrs(kn); > + if (!attrs) > + return -ENOMEM; > + > + q.name = kn->name; > + q.hash_len = hashlen_string(parent, kn->name); > [...] > > Technically this might make some LSMs unhappy, if they want to set > some non-default context even if parent is all default, The only possibility I see as a potential problem is a kernfs mounted with the smackfstransmute=Something option. This sets the security.SMACK64 to "Something" and the security.SMACK64TRANSMUTE to true on the root node. But that doesn't seem like a rational thing to do for a kernfs based filesystem. > but this is > already impossible now and in this case I think we have no better > choice than sacrificing a bit of flexibility for memory efficiency, > which is apparently critical here. > > Tejun, Casey, would the above modification be fine with you? I *think so*, but I can't say 100% that I really understand the entire issue. > > -- > Ondrej Mosnacek <omosnace at redhat dot com> > Associate Software Engineer, Security Technologies > Red Hat, Inc. >
On Thu, Jan 31, 2019 at 5:39 PM Casey Schaufler <casey@schaufler-ca.com> wrote: > On 1/31/2019 2:20 AM, Ondrej Mosnacek wrote: > > Hi Tejun, > > > > On Wed, Jan 30, 2019 at 6:09 PM Tejun Heo <tj@kernel.org> wrote: > >> Hello, > >> > >> On Wed, Jan 30, 2019 at 12:41:50PM +0100, Ondrej Mosnacek wrote: > >>> @@ -673,6 +698,12 @@ static struct kernfs_node *__kernfs_new_node(struct kernfs_root *root, > >>> goto err_out3; > >>> } > >>> > >>> + if (parent) { > >>> + ret = kernfs_node_init_security(parent, kn); > >>> + if (ret) > >>> + goto err_out3; > >>> + } > >> So, doing this unconditionally isn't a good idea. kernfs doesn't use > >> the usual dentry/inode because there are machines with 6, even 7 digit > >> number of kernfs nodes and some of them even failed to boot due to > >> memory shortage. Please don't blow it up by default. > > Hm, I see... basically the only thing that gets allocated in > > kernfs_node_init_security() by default (at least under SELinux/ no > > LSM) is the kernfs_iattrs structures, so I assume you are pointing at > > that. I think this can be easily fixed, if we again use the assumption > > that whenever the parent node has only default attributes > > (parent->iattrs == NULL), then the child node should also have just > > default attributes (and so we don't need to call kernfs_iattrs() on it > > nor call the security hook). Something along these lines: > > > > [...] > > +static int kernfs_node_init_security(struct kernfs_node *parent, > > + struct kernfs_node *kn) > > +{ > > + struct kernfs_iattrs *attrs, *pattrs; > > + struct qstr q; > > + > > + pattrs = parent->iattrs; > > + if (!pattrs) > > + return 0; > > + > > + attrs = kernfs_iattrs(kn); > > + if (!attrs) > > + return -ENOMEM; > > + > > + q.name = kn->name; > > + q.hash_len = hashlen_string(parent, kn->name); > > [...] > > > > Technically this might make some LSMs unhappy, if they want to set > > some non-default context even if parent is all default, > > The only possibility I see as a potential problem is a kernfs > mounted with the smackfstransmute=Something option. This sets > the security.SMACK64 to "Something" and the security.SMACK64TRANSMUTE > to true on the root node. But that doesn't seem like a rational > thing to do for a kernfs based filesystem. Actually... I am now experimenting with a slightly different kernfs_node_init_security() implementation that should allow for calling the hook every time and only allocating kernfs_iattrs when it detects that the hook actually did add some security xattrs. It is somewhat more hacky and complex, but it should provide the best possible compromise. I will post it for review soon. > > > but this is > > already impossible now and in this case I think we have no better > > choice than sacrificing a bit of flexibility for memory efficiency, > > which is apparently critical here. > > > > Tejun, Casey, would the above modification be fine with you? > > I *think so*, but I can't say 100% that I really understand the > entire issue. > > > > > -- > > Ondrej Mosnacek <omosnace at redhat dot com> > > Associate Software Engineer, Security Technologies > > Red Hat, Inc. > > -- Ondrej Mosnacek <omosnace at redhat dot com> Associate Software Engineer, Security Technologies Red Hat, Inc.