Add Linux namespace delegation support

This allows ZFS datasets to be delegated to a user/mount namespace
Within that namespace, only the delegated datasets are visible
Works very similarly to Zones/Jailes on other ZFS OSes

As a user:
```
 $ unshare -Um
 $ zfs list
no datasets available
 $ echo $$
1234
```

As root:
```
 # zfs list
NAME                            ZONED  MOUNTPOINT
containers                      off    /containers
containers/host                 off    /containers/host
containers/host/child           off    /containers/host/child
containers/host/child/gchild    off    /containers/host/child/gchild
containers/unpriv               on     /unpriv
containers/unpriv/child         on     /unpriv/child
containers/unpriv/child/gchild  on     /unpriv/child/gchild

 # zfs zone /proc/1234/ns/user containers/unpriv
```

Back to the user namespace:
```
 $ zfs list
NAME                             USED  AVAIL     REFER  MOUNTPOINT
containers                       129M  47.8G       24K  /containers
containers/unpriv                128M  47.8G       24K  /unpriv
containers/unpriv/child          128M  47.8G      128M  /unpriv/child
```

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Will Andrews <will.andrews@klarasystems.com>
Signed-off-by: Allan Jude <allan@klarasystems.com>
Signed-off-by: Mateusz Piotrowski <mateusz.piotrowski@klarasystems.com>
Co-authored-by: Allan Jude <allan@klarasystems.com>
Co-authored-by: Mateusz Piotrowski <mateusz.piotrowski@klarasystems.com>
Sponsored-by: Buddy <https://buddy.works>
Closes openzfs#12263

cmd/zfs/zfs_main.c

@@ -127,6 +127,11 @@ static int zfs_do_jail(int argc, char **argv);
 static int zfs_do_unjail(int argc, char **argv);
 #endif
 
+#ifdef __linux__
+static int zfs_do_zone(int argc, char **argv);
+static int zfs_do_unzone(int argc, char **argv);
+#endif
+
 /*
  * Enable a reasonable set of defaults for libumem debugging on DEBUG builds.
  */
@@ -184,6 +189,8 @@ typedef enum {
 	HELP_JAIL,
 	HELP_UNJAIL,
 	HELP_WAIT,
+	HELP_ZONE,
+	HELP_UNZONE,
 } zfs_help_t;
 
 typedef struct zfs_command {
@@ -254,6 +261,11 @@ static zfs_command_t command_table[] = {
 	{ "jail",	zfs_do_jail,		HELP_JAIL		},
 	{ "unjail",	zfs_do_unjail,		HELP_UNJAIL		},
 #endif
+
+#ifdef __linux__
+	{ "zone",	zfs_do_zone,		HELP_ZONE		},
+	{ "unzone",	zfs_do_unzone,		HELP_UNZONE		},
+#endif
 };
 
 #define	NCOMMAND	(sizeof (command_table) / sizeof (command_table[0]))
@@ -415,6 +427,10 @@ get_usage(zfs_help_t idx)
 		return (gettext("\tunjail <jailid|jailname> <filesystem>\n"));
 	case HELP_WAIT:
 		return (gettext("\twait [-t <activity>] <filesystem>\n"));
+	case HELP_ZONE:
+		return (gettext("\tzone <nsfile> <filesystem>\n"));
+	case HELP_UNZONE:
+		return (gettext("\tunzone <nsfile> <filesystem>\n"));
 	default:
 		__builtin_unreachable();
 	}
@@ -8692,6 +8708,50 @@ main(int argc, char **argv)
 	return (ret);
 }
 
+/*
+ * zfs zone nsfile filesystem
+ *
+ * Add or delete the given dataset to/from the namespace.
+ */
+#ifdef __linux__
+static int
+zfs_do_zone_impl(int argc, char **argv, boolean_t attach)
+{
+	zfs_handle_t *zhp;
+	int ret;
+
+	if (argc < 3) {
+		(void) fprintf(stderr, gettext("missing argument(s)\n"));
+		usage(B_FALSE);
+	}
+	if (argc > 3) {
+		(void) fprintf(stderr, gettext("too many arguments\n"));
+		usage(B_FALSE);
+	}
+
+	zhp = zfs_open(g_zfs, argv[2], ZFS_TYPE_FILESYSTEM);
+	if (zhp == NULL)
+		return (1);
+
+	ret = (zfs_userns(zhp, argv[1], attach) != 0);
+
+	zfs_close(zhp);
+	return (ret);
+}
+
+static int
+zfs_do_zone(int argc, char **argv)
+{
+	return (zfs_do_zone_impl(argc, argv, B_TRUE));
+}
+
+static int
+zfs_do_unzone(int argc, char **argv)
+{
+	return (zfs_do_zone_impl(argc, argv, B_FALSE));
+}
+#endif
+
 #ifdef __FreeBSD__
 #include <sys/jail.h>
 #include <jail.h>

config/kernel-user-ns-inum.m4

@@ -0,0 +1,23 @@
+dnl #
+dnl # 3.18 API change
+dnl # struct user_namespace inum moved from .proc_inum to .ns.inum.
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_USER_NS_COMMON_INUM], [
+	ZFS_LINUX_TEST_SRC([user_ns_common_inum], [
+		#include <linux/user_namespace.h>
+	], [
+		struct user_namespace uns;
+		uns.ns.inum = 0;
+	])
+])
+
+AC_DEFUN([ZFS_AC_KERNEL_USER_NS_COMMON_INUM], [
+	AC_MSG_CHECKING([whether user_namespace->ns.inum exists])
+	ZFS_LINUX_TEST_RESULT([user_ns_common_inum], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_USER_NS_COMMON_INUM, 1,
+		    [user_namespace->ns.inum exists])
+	],[
+		AC_MSG_RESULT(no)
+	])
+])

config/kernel.m4

@@ -145,6 +145,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_SRC], [
 	ZFS_AC_KERNEL_SRC_KTHREAD
 	ZFS_AC_KERNEL_SRC_ZERO_PAGE
 	ZFS_AC_KERNEL_SRC___COPY_FROM_USER_INATOMIC
+	ZFS_AC_KERNEL_SRC_USER_NS_COMMON_INUM
 
 	AC_MSG_CHECKING([for available kernel interfaces])
 	ZFS_LINUX_TEST_COMPILE_ALL([kabi])
@@ -263,6 +264,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_RESULT], [
 	ZFS_AC_KERNEL_KTHREAD
 	ZFS_AC_KERNEL_ZERO_PAGE
 	ZFS_AC_KERNEL___COPY_FROM_USER_INATOMIC
+	ZFS_AC_KERNEL_USER_NS_COMMON_INUM
 ])
 
 dnl #

contrib/pyzfs/libzfs_core/_constants.py

@@ -100,6 +100,7 @@ def enum(*sequential, **named):
         'ZFS_ERR_REBUILD_IN_PROGRESS',
         'ZFS_ERR_BADPROP',
         'ZFS_ERR_VDEV_NOTSUP',
+        'ZFS_ERR_NOT_USER_NAMESPACE',
     ],
     {}
 )

include/libzfs.h

@@ -150,6 +150,7 @@ typedef enum zfs_error {
 	EZFS_EXPORT_IN_PROGRESS,	/* currently exporting the pool */
 	EZFS_REBUILDING,	/* resilvering (sequential reconstrution) */
 	EZFS_VDEV_NOTSUP,	/* ops not supported for this type of vdev */
+	EZFS_NOT_USER_NAMESPACE,	/* a file is not a user namespace */
 	EZFS_UNKNOWN
 } zfs_error_t;
 
@@ -979,6 +980,15 @@ _LIBZFS_H int zpool_nextboot(libzfs_handle_t *, uint64_t, uint64_t,
 
 #endif /* __FreeBSD__ */
 
+#ifdef __linux__
+
+/*
+ * Add or delete the given filesystem to/from the given user namespace.
+ */
+_LIBZFS_H int zfs_userns(zfs_handle_t *zhp, const char *nspath, int attach);
+
+#endif
+
 #ifdef	__cplusplus
 }
 #endif

include/os/linux/spl/sys/zone.h

@@ -25,11 +25,34 @@
 #define	_SPL_ZONE_H
 
 #include <sys/byteorder.h>
+#include <sys/cred.h>
 
-#define	GLOBAL_ZONEID			0
+#include <linux/cred.h>
+#include <linux/user_namespace.h>
 
-#define	zone_dataset_visible(x, y)	(1)
-#define	crgetzoneid(x)			(GLOBAL_ZONEID)
-#define	INGLOBALZONE(z)			(1)
+/*
+ * Attach the given dataset to the given user namespace.
+ */
+extern int zone_dataset_attach(cred_t *, const char *, int);
+
+/*
+ * Detach the given dataset from the given user namespace.
+ */
+extern int zone_dataset_detach(cred_t *, const char *, int);
+
+/*
+ * Returns true if the named pool/dataset is visible in the current zone.
+ */
+extern int zone_dataset_visible(const char *dataset, int *write);
+
+int spl_zone_init(void);
+void spl_zone_fini(void);
+
+extern unsigned int crgetzoneid(const cred_t *);
+extern unsigned int global_zoneid(void);
+extern boolean_t inglobalzone(proc_t *);
+
+#define	INGLOBALZONE(x) inglobalzone(x)
+#define	GLOBAL_ZONEID	global_zoneid()
 
 #endif /* SPL_ZONE_H */

include/sys/fs/zfs.h

@@ -1471,7 +1471,9 @@ typedef enum zfs_ioc {
 	ZFS_IOC_EVENTS_SEEK,			/* 0x83 (Linux) */
 	ZFS_IOC_NEXTBOOT,			/* 0x84 (FreeBSD) */
 	ZFS_IOC_JAIL,				/* 0x85 (FreeBSD) */
+	ZFS_IOC_USERNS_ATTACH = ZFS_IOC_JAIL,	/* 0x85 (Linux) */
 	ZFS_IOC_UNJAIL,				/* 0x86 (FreeBSD) */
+	ZFS_IOC_USERNS_DETACH = ZFS_IOC_UNJAIL,	/* 0x86 (Linux) */
 	ZFS_IOC_SET_BOOTENV,			/* 0x87 */
 	ZFS_IOC_GET_BOOTENV,			/* 0x88 */
 	ZFS_IOC_UNREGISTER_FS,			/* 0x89 (Windows) */
@@ -1555,6 +1557,7 @@ typedef enum {
 	ZFS_ERR_REBUILD_IN_PROGRESS,
 	ZFS_ERR_BADPROP,
 	ZFS_ERR_VDEV_NOTSUP,
+	ZFS_ERR_NOT_USER_NAMESPACE,
 } zfs_errno_t;
 
 /*

lib/libspl/include/sys/types.h

@@ -44,7 +44,7 @@
 #include <inttypes.h>
 #endif /* HAVE_INTTYPES */
 
-typedef int		zoneid_t;
+typedef uint_t		zoneid_t;
 typedef int		projid_t;
 
 /*

lib/libspl/include/zone.h

@@ -33,7 +33,17 @@
 extern "C" {
 #endif
 
-#define	GLOBAL_ZONEID		0
+#ifdef __FreeBSD__
+#define	GLOBAL_ZONEID	0
+#else
+/*
+ * Hardcoded in the kernel's root user namespace.  A "better" way to get
+ * this would be by using ioctl_ns(2), but this would need to be performed
+ * recursively on NS_GET_PARENT and then NS_GET_USERNS.  Also, that's only
+ * supported since Linux 4.9.
+ */
+#define	GLOBAL_ZONEID	4026531837U
+#endif
 
 extern zoneid_t		getzoneid(void);
 

lib/libspl/os/linux/zone.c

@@ -23,10 +23,40 @@
  * Use is subject to license terms.
  */
 
+#include <unistd.h>
+#include <stdio.h>
+#include <errno.h>
+#include <stdlib.h>
+#include <limits.h>
+#include <string.h>
+
 #include <zone.h>
 
 zoneid_t
 getzoneid(void)
 {
-	return (GLOBAL_ZONEID);
+	char path[PATH_MAX];
+	char buf[128] = { '\0' };
+	char *cp;
+
+	int c = snprintf(path, sizeof (path), "/proc/self/ns/user");
+	/* This API doesn't have any error checking... */
+	if (c < 0)
+		return (0);
+
+	ssize_t r = readlink(path, buf, sizeof (buf) - 1);
+	if (r < 0)
+		return (0);
+
+	cp = strchr(buf, '[');
+	if (cp == NULL)
+		return (0);
+	cp++;
+
+	unsigned long n = strtoul(cp, NULL, 10);
+	if (n == ULONG_MAX && errno == ERANGE)
+		return (0);
+	zoneid_t z = (zoneid_t)n;
+
+	return (z);
 }

lib/libuutil/libuutil.abi

@@ -1081,7 +1081,7 @@
     </function-decl>
   </abi-instr>
   <abi-instr address-size='64' path='os/linux/zone.c' language='LANG_C99'>
-    <typedef-decl name='zoneid_t' type-id='95e97e5e' id='4da03624'/>
+    <typedef-decl name='zoneid_t' type-id='3502e3ff' id='4da03624'/>
     <function-decl name='getzoneid' mangled-name='getzoneid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getzoneid'>
       <return type-id='4da03624'/>
     </function-decl>

lib/libzfs/libzfs.abi

@@ -433,6 +433,7 @@
     <elf-symbol name='zfs_unmountall' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='zfs_unshare' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='zfs_unshareall' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
+    <elf-symbol name='zfs_userns' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='zfs_userspace' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='zfs_valid_proplist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='zfs_version_kernel' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
@@ -1537,7 +1538,7 @@
     </function-decl>
   </abi-instr>
   <abi-instr address-size='64' path='lib/libspl/os/linux/zone.c' language='LANG_C99'>
-    <typedef-decl name='zoneid_t' type-id='95e97e5e' id='4da03624'/>
+    <typedef-decl name='zoneid_t' type-id='3502e3ff' id='4da03624'/>
     <function-decl name='getzoneid' mangled-name='getzoneid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getzoneid'>
       <return type-id='4da03624'/>
     </function-decl>
@@ -4414,6 +4415,12 @@
     <function-decl name='zfs_version_kernel' mangled-name='zfs_version_kernel' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_version_kernel'>
       <return type-id='26a90f95'/>
     </function-decl>
+    <function-decl name='zfs_userns' mangled-name='zfs_userns' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_userns'>
+      <parameter type-id='9200a744' name='zhp'/>
+      <parameter type-id='80f4b756' name='nspath'/>
+      <parameter type-id='95e97e5e' name='attach'/>
+      <return type-id='95e97e5e'/>
+    </function-decl>
   </abi-instr>
   <abi-instr address-size='64' path='lib/libzutil/os/linux/zutil_device_path_os.c' language='LANG_C99'>
     <function-decl name='zfs_append_partition' mangled-name='zfs_append_partition' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_append_partition'>

lib/libzfs/libzfs_util.c

@@ -298,6 +298,9 @@ libzfs_error_description(libzfs_handle_t *hdl)
 	case EZFS_VDEV_NOTSUP:
 		return (dgettext(TEXT_DOMAIN, "operation not supported "
 		    "on this type of vdev"));
+	case EZFS_NOT_USER_NAMESPACE:
+		return (dgettext(TEXT_DOMAIN, "the provided file "
+		    "was not a user namespace file"));
 	case EZFS_UNKNOWN:
 		return (dgettext(TEXT_DOMAIN, "unknown error"));
 	default:
@@ -484,6 +487,9 @@ zfs_standard_error_fmt(libzfs_handle_t *hdl, int error, const char *fmt, ...)
 	case ZFS_ERR_BADPROP:
 		zfs_verror(hdl, EZFS_BADPROP, fmt, ap);
 		break;
+	case ZFS_ERR_NOT_USER_NAMESPACE:
+		zfs_verror(hdl, EZFS_NOT_USER_NAMESPACE, fmt, ap);
+		break;
 	default:
 		zfs_error_aux(hdl, "%s", strerror(error));
 		zfs_verror(hdl, EZFS_UNKNOWN, fmt, ap);