[lxc-devel] [PATCH RFC] seccomp: introduce v2 policy

[Serge Hallyn]

v2 allows specifying system calls by name, and specifying
architecture.  A policy might look like:

2
whitelist
# native architecture is x86_64;  we could put
# [x86-64] here but don't need to
open
read
write
close
mount
# let's also allow some 32-bit syscalls
[x86]
open
read

Also use SCMP_ACT_KILL rather than SCMP_ACT_ERRNO(31)  - which
confusingly returns 'EMLINK' on x86_64.

Signed-off-by: Serge Hallyn <serge.hallyn at ubuntu.com>
---
 src/lxc/seccomp.c | 164 +++++++++++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 144 insertions(+), 20 deletions(-)

diff --git a/src/lxc/seccomp.c b/src/lxc/seccomp.c
index ea23b3a..c8c232a 100644
--- a/src/lxc/seccomp.c
+++ b/src/lxc/seccomp.c
@@ -34,6 +34,143 @@
 
 lxc_log_define(lxc_seccomp, lxc);
 
+static int parse_config_v1(FILE *f, struct lxc_conf *conf)
+{
+	char line[1024];
+	int ret;
+
+	while (fgets(line, 1024, f)) {
+		int nr;
+		ret = sscanf(line, "%d", &nr);
+		if (ret != 1)
+			return -1;
+		ret = seccomp_rule_add(
+#if HAVE_SCMP_FILTER_CTX
+			conf->seccomp_ctx,
+#endif
+			SCMP_ACT_ALLOW, nr, 0);
+		if (ret < 0) {
+			ERROR("failed loading allow rule for %d", nr);
+			return ret;
+		}
+	}
+	return 0;
+}
+
+static void remove_trailing_newlines(char *l)
+{
+	char *p = l;
+
+	while (*p)
+		p++;
+	while (--p >= l && *p == '\n')
+		*p = '\0';
+}
+
+/*
+ * v2 consists of
+ * [x86]
+ * open
+ * read
+ * write
+ * close
+ * # a comment
+ * [x86_64]
+ * open
+ * read
+ * write
+ * close
+ */
+static int parse_config_v2(FILE *f, struct lxc_conf *conf)
+{
+#if HAVE_SCMP_FILTER_CTX
+	char line[1024];
+	int ret;
+	uint32_t arch = SCMP_ARCH_NATIVE;
+	scmp_filter_ctx *ctx = NULL;
+
+	while (fgets(line, 1024, f)) {
+		int nr;
+
+		if (line[0] == '#')
+			continue;
+		if (strlen(line) == 0)
+			continue;
+		remove_trailing_newlines(line);
+		INFO("processing: .%s.", line);
+		if (line[0] == '[') {
+			if (strcmp(line, "[x86]") == 0 ||
+					strcmp(line, "[X86]") == 0)
+				arch = SCMP_ARCH_X86;
+			else if (strcmp(line, "[X86_64]") == 0 ||
+					strcmp(line, "[x86_64]") == 0)
+				arch = SCMP_ARCH_X86_64;
+			else if (strcmp(line, "[arm]") == 0 ||
+					strcmp(line, "[ARM]") == 0)
+				arch = SCMP_ARCH_ARM;
+			else
+				goto bad_arch;
+			if (ctx) {
+				ERROR("Only two arch sections per policy supported");
+				goto bad_arch;
+			}
+			if ((ctx = seccomp_init(SCMP_ACT_KILL)) == NULL) {
+				ERROR("Error initializing seccomp context");
+				return -1;
+			}
+			if (seccomp_attr_set(ctx, SCMP_FLTATR_CTL_NNP, 0)) {
+				ERROR("failed to turn off n-new-privs");
+				seccomp_release(ctx);
+				return -1;
+			}
+			ret = seccomp_arch_add(ctx, arch);
+			if (ret == -EEXIST) {
+				seccomp_release(ctx);
+				ctx = NULL;
+				continue;
+			}
+			if (ret != 0) {
+				ERROR("Error %d adding arch: %s", ret, line);
+				goto bad_arch;
+			}
+			if (seccomp_arch_remove(ctx, SCMP_ARCH_NATIVE) != 0) {
+				ERROR("Error removing native arch from %s", line);
+				goto bad_arch;
+			}
+			continue;
+		}
+
+		nr = seccomp_syscall_resolve_name_arch(arch, line);
+		if (nr < 0) {
+			ERROR("Failed to resolve syscall: %s", line);
+			goto bad_rule;
+		}
+		ret = seccomp_rule_add(ctx ? ctx : conf->seccomp_ctx,
+			SCMP_ACT_ALLOW, nr, 0);
+		if (ret < 0) {
+			ERROR("failed (%d) loading allow rule for %d", ret, nr);
+			goto bad_rule;
+		}
+	}
+	if (ctx) {
+		if (seccomp_merge(conf->seccomp_ctx, ctx) != 0) {
+			seccomp_release(ctx);
+			ERROR("Error merging seccomp contexts");
+			return -1;
+		}
+	}
+	return 0;
+bad_arch:
+	ERROR("Unsupported arch: %s", line);
+bad_rule:
+	if (ctx)
+		seccomp_release(ctx);
+	return -1;
+#else
+	return -1;
+#endif
+}
+
 /*
  * The first line of the config file has a policy language version
  * the second line has some directives
@@ -48,7 +185,7 @@ static int parse_config(FILE *f, struct lxc_conf *conf)
 	int ret, version;
 
 	ret = fscanf(f, "%d\n", &version);
-	if (ret != 1 || version != 1) {
+	if (ret != 1 || (version != 1 && version != 2)) {
 		ERROR("invalid version");
 		return -1;
 	}
@@ -64,23 +201,10 @@ static int parse_config(FILE *f, struct lxc_conf *conf)
 		ERROR("debug not yet implemented");
 		return -1;
 	}
-	/* now read in the whitelist entries one per line */
-	while (fgets(line, 1024, f)) {
-		int nr;
-		ret = sscanf(line, "%d", &nr);
-		if (ret != 1)
-			return -1;
-		ret = seccomp_rule_add(
-#if HAVE_SCMP_FILTER_CTX
-			conf->seccomp_ctx,
-#endif
-			SCMP_ACT_ALLOW, nr, 0);
-		if (ret < 0) {
-			ERROR("failed loading allow rule for %d", nr);
-			return ret;
-		}
-	}
-	return 0;
+
+	if (version == 1)
+		return parse_config_v1(f, conf);
+	return parse_config_v2(f, conf);
 }
 
 int lxc_read_seccomp_config(struct lxc_conf *conf)
@@ -93,10 +217,10 @@ int lxc_read_seccomp_config(struct lxc_conf *conf)
 
 #if HAVE_SCMP_FILTER_CTX
 	/* XXX for debug, pass in SCMP_ACT_TRAP */
-	conf->seccomp_ctx = seccomp_init(SCMP_ACT_ERRNO(31));
+	conf->seccomp_ctx = seccomp_init(SCMP_ACT_KILL);
 	ret = !conf->seccomp_ctx;
 #else
-	ret = seccomp_init(SCMP_ACT_ERRNO(31)) < 0;
+	ret = seccomp_init(SCMP_ACT_KILL) < 0;
 #endif
 	if (ret) {
 		ERROR("failed initializing seccomp");
-- 
1.9.rc1