From: Miklos Szeredi <mszeredi@xxxxxxx> If CLONE_NEWNS and CLONE_NEWNS_USERMNT are given to clone(2) or unshare(2), then allow user mounts within the new namespace. This is not flexible enough, because user mounts can't be enabled for the initial namespace. The remaining clone bits also getting dangerously few... Alternatives are: - prctl() flag - setting through the containers filesystem Signed-off-by: Miklos Szeredi <mszeredi@xxxxxxx> --- Index: linux/fs/namespace.c =================================================================== --- linux.orig/fs/namespace.c 2007-04-12 13:46:19.000000000 +0200 +++ linux/fs/namespace.c 2007-04-12 13:54:36.000000000 +0200 @@ -1617,6 +1617,8 @@ struct mnt_namespace *copy_mnt_ns(int fl return ns; new_ns = dup_mnt_ns(ns, new_fs); + if (new_ns && (flags & CLONE_NEWNS_USERMNT)) + new_ns->flags |= MNT_NS_PERMIT_USERMOUNTS; put_mnt_ns(ns); return new_ns; Index: linux/include/linux/sched.h =================================================================== --- linux.orig/include/linux/sched.h 2007-04-12 13:26:48.000000000 +0200 +++ linux/include/linux/sched.h 2007-04-12 13:54:36.000000000 +0200 @@ -26,6 +26,7 @@ #define CLONE_STOPPED 0x02000000 /* Start in stopped state */ #define CLONE_NEWUTS 0x04000000 /* New utsname group? */ #define CLONE_NEWIPC 0x08000000 /* New ipcs */ +#define CLONE_NEWNS_USERMNT 0x10000000 /* Allow user mounts in ns? */ /* * Scheduling policies Index: linux/kernel/fork.c =================================================================== --- linux.orig/kernel/fork.c 2007-04-11 18:27:46.000000000 +0200 +++ linux/kernel/fork.c 2007-04-12 13:59:10.000000000 +0200 @@ -1586,7 +1586,7 @@ asmlinkage long sys_unshare(unsigned lon err = -EINVAL; if (unshare_flags & ~(CLONE_THREAD|CLONE_FS|CLONE_NEWNS|CLONE_SIGHAND| CLONE_VM|CLONE_FILES|CLONE_SYSVSEM| - CLONE_NEWUTS|CLONE_NEWIPC)) + CLONE_NEWUTS|CLONE_NEWIPC|CLONE_NEWNS_USERMNT)) goto bad_unshare_out; if ((err = unshare_thread(unshare_flags))) -- _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/containers
