virtiofsd: add seccomp whitelist

Only allow system calls that are needed by virtiofsd.  All other system
calls cause SIGSYS to be directed at the thread and the process will
coredump.

Restricting system calls reduces the kernel attack surface and limits
what the process can do when compromised.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
with additional entries by:
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
Signed-off-by: Misono Tomohiro <misono.tomohiro@jp.fujitsu.com>
Signed-off-by: piaojun <piaojun@huawei.com>
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Eric Ren <renzhen@linux.alibaba.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>

diff --git a/Makefile b/Makefile
index 2f81271ee7845a82e7bd228812e35a59c58e4113..ad400f36b211c06adc95dcf308890bb650baec90 100644 (file)
--- a/Makefile
+++ b/Makefile
@@ -327,7 +327,7 @@ HELPERS-y += vhost-user-gpu$(EXESUF)
 vhost-user-json-y += contrib/vhost-user-gpu/50-qemu-gpu.json
 endif
 
-ifdef CONFIG_LINUX
+ifeq ($(CONFIG_LINUX)$(CONFIG_SECCOMP),yy)
 HELPERS-y += virtiofsd$(EXESUF)
 vhost-user-json-y += tools/virtiofsd/50-qemu-virtiofsd.json
 endif
@@ -674,7 +674,8 @@ rdmacm-mux$(EXESUF): LIBS += "-libumad"
 rdmacm-mux$(EXESUF): $(rdmacm-mux-obj-y) $(COMMON_LDADDS)
        $(call LINK, $^)
 
-ifdef CONFIG_LINUX # relies on Linux-specific syscalls
+# relies on Linux-specific syscalls
+ifeq ($(CONFIG_LINUX)$(CONFIG_SECCOMP),yy)
 virtiofsd$(EXESUF): $(virtiofsd-obj-y) libvhost-user.a $(COMMON_LDADDS)
        $(call LINK, $^)
 endif

diff --git a/tools/virtiofsd/Makefile.objs b/tools/virtiofsd/Makefile.objs
index 45a807500d932383c9c51bbaec63ff8f13878330..076f667e46b36da47980d31445689a7d7f128b41 100644 (file)
--- a/tools/virtiofsd/Makefile.objs
+++ b/tools/virtiofsd/Makefile.objs
@@ -5,5 +5,8 @@ virtiofsd-obj-y = buffer.o \
                   fuse_signals.o \
                   fuse_virtio.o \
                   helper.o \
-                  passthrough_ll.o
+                  passthrough_ll.o \
+                  seccomp.o
 
+seccomp.o-cflags := $(SECCOMP_CFLAGS)
+seccomp.o-libs := $(SECCOMP_LIBS)

diff --git a/tools/virtiofsd/passthrough_ll.c b/tools/virtiofsd/passthrough_ll.c
index 0947d14e5bc8d84e079b9190b29b2eeeb4a792cd..bd8925bd8365b5a9839bfce6c2a0eb9e4c9082cc 100644 (file)
--- a/tools/virtiofsd/passthrough_ll.c
+++ b/tools/virtiofsd/passthrough_ll.c
@@ -59,6 +59,7 @@
 #include <unistd.h>
 
 #include "passthrough_helpers.h"
+#include "seccomp.h"
 
 struct lo_map_elem {
     union {
@@ -2091,6 +2092,7 @@ static void setup_sandbox(struct lo_data *lo, struct fuse_session *se)
 {
     setup_namespaces(lo, se);
     setup_mounts(lo->source);
+    setup_seccomp();
 }
 
 int main(int argc, char *argv[])

diff --git a/tools/virtiofsd/seccomp.c b/tools/virtiofsd/seccomp.c
new file mode 100644 (file)
index 0000000..691fb63
--- /dev/null
+++ b/tools/virtiofsd/seccomp.c
@@ -0,0 +1,151 @@
+/*
+ * Seccomp sandboxing for virtiofsd
+ *
+ * Copyright (C) 2019 Red Hat, Inc.
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+
+#include "qemu/osdep.h"
+#include "seccomp.h"
+#include "fuse_i.h"
+#include "fuse_log.h"
+#include <errno.h>
+#include <glib.h>
+#include <seccomp.h>
+#include <stdlib.h>
+
+/* Bodge for libseccomp 2.4.2 which broke ppoll */
+#if !defined(__SNR_ppoll) && defined(__SNR_brk)
+#ifdef __NR_ppoll
+#define __SNR_ppoll __NR_ppoll
+#else
+#define __SNR_ppoll __PNR_ppoll
+#endif
+#endif
+
+static const int syscall_whitelist[] = {
+    /* TODO ireg sem*() syscalls */
+    SCMP_SYS(brk),
+    SCMP_SYS(capget), /* For CAP_FSETID */
+    SCMP_SYS(capset),
+    SCMP_SYS(clock_gettime),
+    SCMP_SYS(clone),
+#ifdef __NR_clone3
+    SCMP_SYS(clone3),
+#endif
+    SCMP_SYS(close),
+    SCMP_SYS(copy_file_range),
+    SCMP_SYS(dup),
+    SCMP_SYS(eventfd2),
+    SCMP_SYS(exit),
+    SCMP_SYS(exit_group),
+    SCMP_SYS(fallocate),
+    SCMP_SYS(fchmodat),
+    SCMP_SYS(fchownat),
+    SCMP_SYS(fcntl),
+    SCMP_SYS(fdatasync),
+    SCMP_SYS(fgetxattr),
+    SCMP_SYS(flistxattr),
+    SCMP_SYS(flock),
+    SCMP_SYS(fremovexattr),
+    SCMP_SYS(fsetxattr),
+    SCMP_SYS(fstat),
+    SCMP_SYS(fstatfs),
+    SCMP_SYS(fsync),
+    SCMP_SYS(ftruncate),
+    SCMP_SYS(futex),
+    SCMP_SYS(getdents),
+    SCMP_SYS(getdents64),
+    SCMP_SYS(getegid),
+    SCMP_SYS(geteuid),
+    SCMP_SYS(getpid),
+    SCMP_SYS(gettid),
+    SCMP_SYS(gettimeofday),
+    SCMP_SYS(linkat),
+    SCMP_SYS(lseek),
+    SCMP_SYS(madvise),
+    SCMP_SYS(mkdirat),
+    SCMP_SYS(mknodat),
+    SCMP_SYS(mmap),
+    SCMP_SYS(mprotect),
+    SCMP_SYS(mremap),
+    SCMP_SYS(munmap),
+    SCMP_SYS(newfstatat),
+    SCMP_SYS(open),
+    SCMP_SYS(openat),
+    SCMP_SYS(ppoll),
+    SCMP_SYS(prctl), /* TODO restrict to just PR_SET_NAME? */
+    SCMP_SYS(preadv),
+    SCMP_SYS(pread64),
+    SCMP_SYS(pwritev),
+    SCMP_SYS(pwrite64),
+    SCMP_SYS(read),
+    SCMP_SYS(readlinkat),
+    SCMP_SYS(recvmsg),
+    SCMP_SYS(renameat),
+    SCMP_SYS(renameat2),
+    SCMP_SYS(rt_sigaction),
+    SCMP_SYS(rt_sigprocmask),
+    SCMP_SYS(rt_sigreturn),
+    SCMP_SYS(sendmsg),
+    SCMP_SYS(setresgid),
+    SCMP_SYS(setresuid),
+#ifdef __NR_setresgid32
+    SCMP_SYS(setresgid32),
+#endif
+#ifdef __NR_setresuid32
+    SCMP_SYS(setresuid32),
+#endif
+    SCMP_SYS(set_robust_list),
+    SCMP_SYS(symlinkat),
+    SCMP_SYS(time), /* Rarely needed, except on static builds */
+    SCMP_SYS(tgkill),
+    SCMP_SYS(unlinkat),
+    SCMP_SYS(utimensat),
+    SCMP_SYS(write),
+    SCMP_SYS(writev),
+};
+
+void setup_seccomp(void)
+{
+    scmp_filter_ctx ctx;
+    size_t i;
+
+#ifdef SCMP_ACT_KILL_PROCESS
+    ctx = seccomp_init(SCMP_ACT_KILL_PROCESS);
+    /* Handle a newer libseccomp but an older kernel */
+    if (!ctx && errno == EOPNOTSUPP) {
+        ctx = seccomp_init(SCMP_ACT_TRAP);
+    }
+#else
+    ctx = seccomp_init(SCMP_ACT_TRAP);
+#endif
+    if (!ctx) {
+        fuse_log(FUSE_LOG_ERR, "seccomp_init() failed\n");
+        exit(1);
+    }
+
+    for (i = 0; i < G_N_ELEMENTS(syscall_whitelist); i++) {
+        if (seccomp_rule_add(ctx, SCMP_ACT_ALLOW,
+                             syscall_whitelist[i], 0) != 0) {
+            fuse_log(FUSE_LOG_ERR, "seccomp_rule_add syscall %d",
+                     syscall_whitelist[i]);
+            exit(1);
+        }
+    }
+
+    /* libvhost-user calls this for post-copy migration, we don't need it */
+    if (seccomp_rule_add(ctx, SCMP_ACT_ERRNO(ENOSYS),
+                         SCMP_SYS(userfaultfd), 0) != 0) {
+        fuse_log(FUSE_LOG_ERR, "seccomp_rule_add userfaultfd failed\n");
+        exit(1);
+    }
+
+    if (seccomp_load(ctx) < 0) {
+        fuse_log(FUSE_LOG_ERR, "seccomp_load() failed\n");
+        exit(1);
+    }
+
+    seccomp_release(ctx);
+}

diff --git a/tools/virtiofsd/seccomp.h b/tools/virtiofsd/seccomp.h
new file mode 100644 (file)
index 0000000..86bce72
--- /dev/null
+++ b/tools/virtiofsd/seccomp.h
@@ -0,0 +1,14 @@
+/*
+ * Seccomp sandboxing for virtiofsd
+ *
+ * Copyright (C) 2019 Red Hat, Inc.
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+
+#ifndef VIRTIOFSD_SECCOMP_H
+#define VIRTIOFSD_SECCOMP_H
+
+void setup_seccomp(void);
+
+#endif /* VIRTIOFSD_SECCOMP_H */