Second-to-last commit implementing Capsicum capabilities in the FreeBSD

kernel for FreeBSD 9.0:

Add a new capability mask argument to fget(9) and friends, allowing system
call code to declare what capabilities are required when an integer file
descriptor is converted into an in-kernel struct file *.  With options
CAPABILITIES compiled into the kernel, this enforces capability
protection; without, this change is effectively a no-op.

Some cases require special handling, such as mmap(2), which must preserve
information about the maximum rights at the time of mapping in the memory
map so that they can later be enforced in mprotect(2) -- this is done by
narrowing the rights in the existing max_protection field used for similar
purposes with file permissions.

In namei(9), we assert that the code is not reached from within capability
mode, as we're not yet ready to enforce namespace capabilities there.
This will follow in a later commit.

Update two capability names: CAP_EVENT and CAP_KEVENT become
CAP_POST_KEVENT and CAP_POLL_KEVENT to more accurately indicate what they
represent.

Approved by:	re (bz)
Submitted by:	jonathan
Sponsored by:	Google Inc

6 files changed, 46 insertions, 7 deletions

diff --git a/sys/fs/coda/coda_psdev.c b/sys/fs/coda/coda_psdev.c
index 494f30b..e982a2c 100644
--- a/sys/fs/coda/coda_psdev.c
+++ b/sys/fs/coda/coda_psdev.c
@@ -55,6 +55,7 @@ __FBSDID("$FreeBSD$");
 
 #include <sys/param.h>
 #include <sys/systm.h>
+#include <sys/capability.h>
 #include <sys/conf.h>
 #include <sys/ioccom.h>
 #include <sys/kernel.h>
@@ -371,7 +372,7 @@ vc_write(struct cdev *dev, struct uio *uiop, int flag)
 		struct vnode *vp = NULL;
 
 		if (tmp->oh.result == 0) {
-			error = getvnode(uiop->uio_td->td_proc->p_fd,
+			error = getvnode(uiop->uio_td->td_proc->p_fd, CAP_WRITE,
 			    tmp->fd, &fp);
 			if (!error) {
 				/*
diff --git a/sys/fs/fdescfs/fdesc_vnops.c b/sys/fs/fdescfs/fdesc_vnops.c
index e11b59c..3c4f44d 100644
--- a/sys/fs/fdescfs/fdesc_vnops.c
+++ b/sys/fs/fdescfs/fdesc_vnops.c
@@ -40,6 +40,7 @@
 
 #include <sys/param.h>
 #include <sys/systm.h>
+#include <sys/capability.h>
 #include <sys/conf.h>
 #include <sys/dirent.h>
 #include <sys/filedesc.h>
@@ -305,7 +306,10 @@ fdesc_lookup(ap)
 		fd = fd1;
 	}
 
-	if ((error = fget(td, fd, &fp)) != 0)
+	/*
+	 * No rights to check since 'fp' isn't actually used.
+	 */
+	if ((error = fget(td, fd, 0, &fp)) != 0)
 		goto bad;
 
 	/* Check if we're looking up ourselves. */
@@ -455,7 +459,7 @@ fdesc_setattr(ap)
 	/*
 	 * Allow setattr where there is an underlying vnode.
 	 */
-	error = getvnode(td->td_proc->p_fd, fd, &fp);
+	error = getvnode(td->td_proc->p_fd, fd, CAP_EXTATTR_SET, &fp);
 	if (error) {
 		/*
 		 * getvnode() returns EINVAL if the file descriptor is not
diff --git a/sys/fs/nfsclient/nfs_clport.c b/sys/fs/nfsclient/nfs_clport.c
index 44d3c74..acbfa6c 100644
--- a/sys/fs/nfsclient/nfs_clport.c
+++ b/sys/fs/nfsclient/nfs_clport.c
@@ -36,6 +36,8 @@ __FBSDID("$FreeBSD$");
 
 #include "opt_kdtrace.h"
 
+#include <sys/capability.h>
+
 /*
  * generally, I don't like #includes inside .h files, but it seems to
  * be the easiest way to handle the port.
@@ -1231,7 +1233,13 @@ nfssvc_nfscl(struct thread *td, struct nfssvc_args *uap)
 		error = copyin(uap->argp, (caddr_t)&nfscbdarg, sizeof(nfscbdarg));
 		if (error)
 			return (error);
-		if ((error = fget(td, nfscbdarg.sock, &fp)) != 0) {
+		/*
+		 * Since we don't know what rights might be required,
+		 * pretend that we need them all. It is better to be too
+		 * careful than too reckless.
+		 */
+		if ((error = fget(td, nfscbdarg.sock, CAP_SOCK_ALL, &fp))
+		    != 0) {
 			return (error);
 		}
 		if (fp->f_type != DTYPE_SOCKET) {
diff --git a/sys/fs/nfsserver/nfs_nfsdport.c b/sys/fs/nfsserver/nfs_nfsdport.c
index 56c563a..da32b0e 100644
--- a/sys/fs/nfsserver/nfs_nfsdport.c
+++ b/sys/fs/nfsserver/nfs_nfsdport.c
@@ -34,6 +34,8 @@
 #include <sys/cdefs.h>
 __FBSDID("$FreeBSD$");
 
+#include <sys/capability.h>
+
 /*
  * Functions that perform the vfs operations required by the routines in
  * nfsd_serv.c. It is hoped that this change will make the server more
@@ -3027,8 +3029,14 @@ nfssvc_nfsd(struct thread *td, struct nfssvc_args *uap)
 		error = copyin(uap->argp, (caddr_t)&sockarg, sizeof (sockarg));
 		if (error)
 			goto out;
-		if ((error = fget(td, sockarg.sock, &fp)) != 0)
+		/*
+		 * Since we don't know what rights might be required,
+		 * pretend that we need them all. It is better to be too
+		 * careful than too reckless.
+		 */
+		if ((error = fget(td, sockarg.sock, CAP_SOCK_ALL, &fp)) != 0)
 			goto out;
+			return (error);
 		if (fp->f_type != DTYPE_SOCKET) {
 			fdrop(fp, td);
 			error = EPERM;
diff --git a/sys/fs/portalfs/portal_vfsops.c b/sys/fs/portalfs/portal_vfsops.c
index a8c16c4..04e7a3d 100644
--- a/sys/fs/portalfs/portal_vfsops.c
+++ b/sys/fs/portalfs/portal_vfsops.c
@@ -40,6 +40,7 @@
 
 #include <sys/param.h>
 #include <sys/systm.h>
+#include <sys/capability.h>
 #include <sys/domain.h>
 #include <sys/filedesc.h>
 #include <sys/kernel.h>
@@ -112,7 +113,12 @@ portal_mount(struct mount *mp)
 	if (error)
 		return (error);
 
-	if ((error = fget(td, v, &fp)) != 0)
+	/*
+	 * Capsicum is not incompatible with portalfs, but we don't really
+	 * know what rights are required. In the spirit of "better safe than
+	 * sorry", pretend that all rights are required for now.
+	 */
+	if ((error = fget(td, v, CAP_MASK_VALID, &fp)) != 0)
 		return (error);
         if (fp->f_type != DTYPE_SOCKET) {
 		fdrop(fp, td);
diff --git a/sys/fs/portalfs/portal_vnops.c b/sys/fs/portalfs/portal_vnops.c
index 6fcc1ce..7cd5267 100644
--- a/sys/fs/portalfs/portal_vnops.c
+++ b/sys/fs/portalfs/portal_vnops.c
@@ -38,7 +38,10 @@
  * Portal Filesystem
  */
 
+#include "opt_capsicum.h"
+
 #include <sys/param.h>
+#include <sys/capability.h>
 #include <sys/fcntl.h>
 #include <sys/file.h>
 #include <sys/kernel.h>
@@ -232,6 +235,15 @@ portal_open(ap)
 	struct file *fp;
 	struct portal_cred pcred;
 
+#ifdef CAPABILITY_MODE
+	/*
+	 * This may require access to a global namespace (e.g. an IP address);
+	 * disallow it entirely, as we do open(2).
+	 */
+	if (IN_CAPABILITY_MODE(td))
+		return (ECAPMODE);
+#endif
+
 	/*
 	 * Nothing to do when opening the root node.
 	 */
@@ -414,7 +426,7 @@ portal_open(ap)
 	 * Check that the mode the file is being opened for is a subset
 	 * of the mode of the existing descriptor.
 	 */
-	if ((error = fget(td, fd, &fp)) != 0)
+	if ((error = fget(td, fd, 0, &fp)) != 0)
 		goto bad;
 	if (((ap->a_mode & (FREAD|FWRITE)) | fp->f_flag) != fp->f_flag) {
 		fdrop(fp, td);