| Message ID | 20150310172206.23081.95005.stgit@buzz |
|---|---|
| State | Accepted, archived |
| Headers | show |
On Tue, Mar 10, 2015 at 10:22 AM, Konstantin Khlebnikov <khlebnikov@yandex-team.ru> wrote: > Historically XFS project id doesn't have any permission control: file owner > is able to set any project id. Later they was sealed with user-namespace: > XFS allows to change it only from init user-ns. That works fine for isolated > containers or if user doesn't have direct access to the filesystem (NFS/FTP). > > This patch adds sysctl fs.protected_projects which makes changing project id > privileged operation which requires CAP_SYS_RESOURCE in current user-namespace. > Thus there are two levels of protection: project id mapping in user-ns defines > set of permitted projects and capability protects operations within this set. If I understand this right, this doesn't work. If I lack CAP_SYS_RESOURCE but I have two projids mapped, then I can create a new userns, map both projids, and get CAP_SYS_RESOURCE. --Andy -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Mar 10, 2015 at 8:32 PM, Andy Lutomirski <luto@amacapital.net> wrote: > On Tue, Mar 10, 2015 at 10:22 AM, Konstantin Khlebnikov > <khlebnikov@yandex-team.ru> wrote: >> Historically XFS project id doesn't have any permission control: file owner >> is able to set any project id. Later they was sealed with user-namespace: >> XFS allows to change it only from init user-ns. That works fine for isolated >> containers or if user doesn't have direct access to the filesystem (NFS/FTP). >> >> This patch adds sysctl fs.protected_projects which makes changing project id >> privileged operation which requires CAP_SYS_RESOURCE in current user-namespace. >> Thus there are two levels of protection: project id mapping in user-ns defines >> set of permitted projects and capability protects operations within this set. > > If I understand this right, this doesn't work. If I lack > CAP_SYS_RESOURCE but I have two projids mapped, then I can create a > new userns, map both projids, and get CAP_SYS_RESOURCE. Setting project id mapping for nested user-namespace also requires this capability in parent namespace. The same as for setting uid/gid mapping but without special case for mapping current uid/gid because task has no "current" project id. This is mentioned in cover letter but I forget it here. Sorry. > > --Andy > -- > To unsubscribe from this list: send the line "unsubscribe linux-ext4" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Mar 10, 2015 at 11:51 AM, Konstantin Khlebnikov <koct9i@gmail.com> wrote: > On Tue, Mar 10, 2015 at 8:32 PM, Andy Lutomirski <luto@amacapital.net> wrote: >> On Tue, Mar 10, 2015 at 10:22 AM, Konstantin Khlebnikov >> <khlebnikov@yandex-team.ru> wrote: >>> Historically XFS project id doesn't have any permission control: file owner >>> is able to set any project id. Later they was sealed with user-namespace: >>> XFS allows to change it only from init user-ns. That works fine for isolated >>> containers or if user doesn't have direct access to the filesystem (NFS/FTP). >>> >>> This patch adds sysctl fs.protected_projects which makes changing project id >>> privileged operation which requires CAP_SYS_RESOURCE in current user-namespace. >>> Thus there are two levels of protection: project id mapping in user-ns defines >>> set of permitted projects and capability protects operations within this set. >> >> If I understand this right, this doesn't work. If I lack >> CAP_SYS_RESOURCE but I have two projids mapped, then I can create a >> new userns, map both projids, and get CAP_SYS_RESOURCE. > > Setting project id mapping for nested user-namespace also requires > this capability in parent namespace. The same as for setting uid/gid > mapping but without special case for mapping current uid/gid because > task has no "current" project id. > > This is mentioned in cover letter but I forget it here. Sorry. Right, sorry. I'm still used to projid mappings being unprotected. --Andy > >> >> --Andy >> -- >> To unsubscribe from this list: send the line "unsubscribe linux-ext4" in >> the body of a message to majordomo@vger.kernel.org >> More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/Documentation/sysctl/fs.txt b/Documentation/sysctl/fs.txt index 88152f214f48..9f6579b99be6 100644 --- a/Documentation/sysctl/fs.txt +++ b/Documentation/sysctl/fs.txt @@ -34,6 +34,7 @@ Currently, these files are in /proc/sys/fs: - overflowgid - protected_hardlinks - protected_symlinks +- protected_projects - suid_dumpable - super-max - super-nr @@ -199,6 +200,21 @@ This protection is based on the restrictions in Openwall and grsecurity. ============================================================== +protected_projects: + +Project id allows to enforce disk quota for several subtrees or individual +files on the filesystem. Historically changing project id was a unprivileged +operation and file owner is able to set any project id. + +When set to "0", changing project id is unprivileged operation. File owner +can set any project id mapped in current user namespace. + +When set to "1" changing project id requires capability CAP_SYS_RESOURCE +in current user namespace. Also defining project id mapping for nested +user namespace requires CAP_SYS_RESOURCE in the parent user namespace. + +============================================================== + suid_dumpable: This value can be used to query and set the core dump mode for setuid diff --git a/fs/ioctl.c b/fs/ioctl.c index d351576d95c8..2acf5efbc045 100644 --- a/fs/ioctl.c +++ b/fs/ioctl.c @@ -565,6 +565,8 @@ static int ioctl_getproject(struct file *filp, projid_t __user *argp) return put_user(projid, argp); } +int sysctl_protected_projects; + static int ioctl_setproject(struct file *filp, projid_t __user *argp) { struct user_namespace *ns = current_user_ns(); @@ -576,7 +578,9 @@ static int ioctl_setproject(struct file *filp, projid_t __user *argp) if (!sb->s_op->set_project) return -EOPNOTSUPP; - if (ns != &init_user_ns) + if (sysctl_protected_projects ? + !ns_capable(ns, CAP_SYS_RESOURCE) : + (ns != &init_user_ns)) return -EPERM; ret = get_user(projid, argp); if (ret) diff --git a/include/linux/fs.h b/include/linux/fs.h index 42156801739e..d3021feb3f7f 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -64,6 +64,7 @@ extern struct inodes_stat_t inodes_stat; extern int leases_enable, lease_break_time; extern int sysctl_protected_symlinks; extern int sysctl_protected_hardlinks; +extern int sysctl_protected_projects; struct buffer_head; typedef int (get_block_t)(struct inode *inode, sector_t iblock, diff --git a/include/uapi/linux/capability.h b/include/uapi/linux/capability.h index 12c37a197d24..0292885567cc 100644 --- a/include/uapi/linux/capability.h +++ b/include/uapi/linux/capability.h @@ -278,6 +278,7 @@ struct vfs_cap_data { /* Override resource limits. Set resource limits. */ /* Override quota limits. */ /* Override reserved space on ext2 filesystem */ +/* Modify file project id if protected_projects = 1 */ /* Modify data journaling mode on ext3 filesystem (uses journaling resources) */ /* NOTE: ext2 honors fsuid when checking for resource overrides, so diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 88ea2d6e0031..cb6f9fb13de3 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -1649,6 +1649,15 @@ static struct ctl_table fs_table[] = { .extra2 = &one, }, { + .procname = "protected_projects", + .data = &sysctl_protected_projects, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &zero, + .extra2 = &one, + }, + { .procname = "suid_dumpable", .data = &suid_dumpable, .maxlen = sizeof(int), diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c index 4109f8320684..88f66198b251 100644 --- a/kernel/user_namespace.c +++ b/kernel/user_namespace.c @@ -807,8 +807,8 @@ ssize_t proc_projid_map_write(struct file *file, const char __user *buf, if ((seq_ns != ns) && (seq_ns != ns->parent)) return -EPERM; - /* Anyone can set any valid project id no capability needed */ - return map_write(file, buf, size, ppos, -1, + return map_write(file, buf, size, ppos, + sysctl_protected_projects ? CAP_SYS_RESOURCE : -1, &ns->projid_map, &ns->parent->projid_map); }
Historically XFS project id doesn't have any permission control: file owner is able to set any project id. Later they was sealed with user-namespace: XFS allows to change it only from init user-ns. That works fine for isolated containers or if user doesn't have direct access to the filesystem (NFS/FTP). This patch adds sysctl fs.protected_projects which makes changing project id privileged operation which requires CAP_SYS_RESOURCE in current user-namespace. Thus there are two levels of protection: project id mapping in user-ns defines set of permitted projects and capability protects operations within this set. Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> --- Documentation/sysctl/fs.txt | 16 ++++++++++++++++ fs/ioctl.c | 6 +++++- include/linux/fs.h | 1 + include/uapi/linux/capability.h | 1 + kernel/sysctl.c | 9 +++++++++ kernel/user_namespace.c | 4 ++-- 6 files changed, 34 insertions(+), 3 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html