From: Li Qiang <liqiang...@360.cn>

If a guest sends an empty string paramater to any 9P operation, the current
code unmarshals it into a V9fsString equal to { .size = 0, .data = NULL }.

This is unfortunate because it can cause NULL pointer dereference to happen
at various locations in the 9pfs code. And we don't want to check str->data
everywhere we pass it to strcmp() or any other function which expects a
dereferenceable pointer.

This patch enforces the allocation of genuine C empty strings instead, so
callers don't have to bother.

Out of all v9fs_iov_vunmarshal() users, only v9fs_xattrwalk() checks if
the returned string is empty. It now uses v9fs_string_size() since
name.data cannot be NULL anymore.

Signed-off-by: Li Qiang <liqiang...@360.cn>
[groug, rewritten title and changelog,
 fix empty string check in v9fs_xattrwalk()]
Signed-off-by: Greg Kurz <gr...@kaod.org>
---
 fsdev/9p-iov-marshal.c | 2 +-
 hw/9pfs/9p.c           | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/fsdev/9p-iov-marshal.c b/fsdev/9p-iov-marshal.c
index 663cad542900..1d16f8df4bd4 100644
--- a/fsdev/9p-iov-marshal.c
+++ b/fsdev/9p-iov-marshal.c
@@ -125,7 +125,7 @@ ssize_t v9fs_iov_vunmarshal(struct iovec *out_sg, int 
out_num, size_t offset,
                 str->data = g_malloc(str->size + 1);
                 copied = v9fs_unpack(str->data, out_sg, out_num, offset,
                                      str->size);
-                if (copied > 0) {
+                if (copied >= 0) {
                     str->data[str->size] = 0;
                 } else {
                     v9fs_string_free(str);
diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
index 119ee584969b..39a7e1d52d2a 100644
--- a/hw/9pfs/9p.c
+++ b/hw/9pfs/9p.c
@@ -3174,7 +3174,7 @@ static void v9fs_xattrwalk(void *opaque)
         goto out;
     }
     v9fs_path_copy(&xattr_fidp->path, &file_fidp->path);
-    if (name.data == NULL) {
+    if (!v9fs_string_size(&name)) {
         /*
          * listxattr request. Get the size first
          */
-- 
2.5.5


Reply via email to