tar: optional support for restoring selinux context

function                                             old     new   delta
get_header_tar                                      1690    1976    +286
data_extract_all                                     821     881     +60
.rodata                                           151446  151503     +57
get_header_cpio                                     1044    1077     +33
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 4/0 up/down: 436/0)             Total: 436 bytes

Signed-off-by: J. Tang <tang@jtang.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

4 files changed, 108 insertions, 3 deletions

diff --git a/archival/Config.in b/archival/Config.in
index c99896b47..deacc2822 100644
--- a/archival/Config.in
+++ b/archival/Config.in
@@ -289,6 +289,14 @@ config FEATURE_TAR_NOPRESERVE_TIME
           With this option busybox supports GNU tar -m
           (do not preserve time) option.
 
+config FEATURE_TAR_SELINUX
+        bool "Support for extracting SELinux labels"
+        default n
+        depends on TAR && SELINUX
+        help
+          With this option busybox supports restoring SELinux labels
+          when extracting files from tar archives.
+
 config UNCOMPRESS
         bool "uncompress"
         default n
diff --git a/archival/libunarchive/data_extract_all.c b/archival/libunarchive/data_extract_all.c
index 58b05335b..cc4894229 100644
--- a/archival/libunarchive/data_extract_all.c
+++ b/archival/libunarchive/data_extract_all.c
@@ -12,6 +12,17 @@ void FAST_FUNC data_extract_all(archive_handle_t *archive_handle)
         int dst_fd;
         int res;
 
+#if ENABLE_FEATURE_TAR_SELINUX
+        char *sctx = archive_handle->tar__next_file_sctx;
+        if (!sctx)
+                sctx = archive_handle->tar__global_sctx;
+        if (sctx) { /* setfscreatecon is 4 syscalls, avoid if possible */
+                setfscreatecon(sctx);
+                free(archive_handle->tar__next_file_sctx);
+                archive_handle->tar__next_file_sctx = NULL;
+        }
+#endif
+
         if (archive_handle->ah_flags & ARCHIVE_CREATE_LEADING_DIRS) {
                 char *slash = strrchr(file_header->name, '/');
                 if (slash) {
@@ -45,7 +56,7 @@ void FAST_FUNC data_extract_all(archive_handle_t *archive_handle)
                                         "same age file exists", file_header->name);
                         }
                         data_skip(archive_handle);
-                        return;
+                        goto ret;
                 }
                 else if ((unlink(file_header->name) == -1) && (errno != EISDIR)) {
                         bb_perror_msg_and_die("can't remove old file %s",
@@ -158,4 +169,12 @@ void FAST_FUNC data_extract_all(archive_handle_t *archive_handle)
                         utimes(file_header->name, t);
                 }
         }
+
+ ret: ;
+#if ENABLE_FEATURE_TAR_SELINUX
+        if (sctx) {
+                /* reset the context after creating an entry */
+                setfscreatecon(NULL);
+        }
+#endif
 }
diff --git a/archival/libunarchive/get_header_tar.c b/archival/libunarchive/get_header_tar.c
index d5b86ff5c..cf0b9ab02 100644
--- a/archival/libunarchive/get_header_tar.c
+++ b/archival/libunarchive/get_header_tar.c
@@ -103,6 +103,63 @@ static unsigned long long getOctal(char *str, int len)
 }
 #define GET_OCTAL(a) getOctal((a), sizeof(a))
 
+#if ENABLE_FEATURE_TAR_SELINUX
+/* Scan a PAX header for SELinux contexts, via "RHT.security.selinux" keyword.
+ * This is what Red Hat's patched version of tar uses.
+ */
+# define SELINUX_CONTEXT_KEYWORD "RHT.security.selinux"
+static char *get_selinux_sctx_from_pax_hdr(archive_handle_t *archive_handle, unsigned sz)
+{
+        char *buf, *p;
+        char *result;
+
+        p = buf = xmalloc(sz + 1);
+        /* prevent bb_strtou from running off the buffer */
+        buf[sz] = '\0';
+        xread(archive_handle->src_fd, buf, sz);
+        archive_handle->offset += sz;
+
+        result = NULL;
+        while (sz != 0) {
+                char *end, *value;
+                unsigned len;
+
+                /* Every record has this format: "LEN NAME=VALUE\n" */
+                len = bb_strtou(p, &end, 10);
+                /* expect errno to be EINVAL, because the character
+                 * following the digits should be a space
+                 */
+                p += len;
+                sz -= len;
+                if ((int)sz < 0
+                 || len == 0
+                 || errno != EINVAL
+                 || *end != ' '
+                ) {
+                        bb_error_msg("malformed extended header, skipped");
+                        // More verbose version:
+                        //bb_error_msg("malformed extended header at %"OFF_FMT"d, skipped",
+                        //              archive_handle->offset - (sz + len));
+                        break;
+                }
+                /* overwrite the terminating newline with NUL
+                 * (we do not bother to check that it *was* a newline)
+                 */
+                p[-1] = '\0';
+                /* Is it selinux security context? */
+                value = end + 1;
+                if (strncmp(value, SELINUX_CONTEXT_KEYWORD"=", sizeof(SELINUX_CONTEXT_KEYWORD"=") - 1) == 0) {
+                        value += sizeof(SELINUX_CONTEXT_KEYWORD"=") - 1;
+                        result = xstrdup(value);
+                        break;
+                }
+        }
+
+        free(buf);
+        return result;
+}
+#endif
+
 void BUG_tar_header_size(void);
 char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
 {
@@ -150,7 +207,7 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
         if (sizeof(tar) != 512)
                 BUG_tar_header_size();
 
-#if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
+#if ENABLE_FEATURE_TAR_GNU_EXTENSIONS || ENABLE_FEATURE_TAR_SELINUX
  again:
 #endif
         /* Align header */
@@ -392,8 +449,13 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
         case 'S':       /* Sparse file */
         case 'V':       /* Volume header */
 #endif
+#if !ENABLE_FEATURE_TAR_SELINUX
         case 'g':       /* pax global header */
-        case 'x': {     /* pax extended header */
+        case 'x':       /* pax extended header */
+#else
+ skip_ext_hdr:
+#endif
+        {
                 off_t sz;
                 bb_error_msg("warning: skipping header '%c'", tar.typeflag);
                 sz = (file_header->size + 511) & ~(off_t)511;
@@ -404,6 +466,18 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
                 /* return get_header_tar(archive_handle); */
                 goto again_after_align;
         }
+#if ENABLE_FEATURE_TAR_SELINUX
+        case 'g':       /* pax global header */
+        case 'x': {     /* pax extended header */
+                char **pp;
+                if ((uoff_t)file_header->size > 0xfffff) /* paranoia */
+                        goto skip_ext_hdr;
+                pp = (tar.typeflag == 'g') ? &archive_handle->tar__global_sctx : &archive_handle->tar__next_file_sctx;
+                free(*pp);
+                *pp = get_selinux_sctx_from_pax_hdr(archive_handle, file_header->size);
+                goto again;
+        }
+#endif
         default:
                 bb_error_msg_and_die("unknown typeflag: 0x%x", tar.typeflag);
         }
diff --git a/include/unarchive.h b/include/unarchive.h
index 92ebd6565..a834816ba 100644
--- a/include/unarchive.h
+++ b/include/unarchive.h
@@ -59,6 +59,10 @@ typedef struct archive_handle_t {
         char* tar__longname;
         char* tar__linkname;
 # endif
+# if ENABLE_FEATURE_TAR_SELINUX
+        char* tar__global_sctx;
+        char* tar__next_file_sctx;
+# endif
 #endif
 #if ENABLE_CPIO || ENABLE_RPM2CPIO || ENABLE_RPM
         uoff_t cpio__blocks;