Lennart Poettering observes that if the newly increased MAX_HANDLE_SZ is exported to user space, and then used in an application running on an old kernel, name_to_handle_at() will report -EINVAL, which is unhelpful and inconsistent with the documentation. So: 1/ add a comment making it clear that the new value must not be exposed to user-space. 2/ remove the completely unnecessary restriction on the size of buffer provided by the application. Reported-by: Lennart Poettering Signed-off-by: NeilBrown --- This is a followup to the previous fhandle-for-Amazon-EFS patch which adds further refinements. Thanks, NeilBrown fs/fhandle.c | 2 +- include/linux/exportfs.h | 4 ++++ 2 files changed, 5 insertions(+), 1 deletion(-) diff --git a/fs/fhandle.c b/fs/fhandle.c index 0ace128f5d23..30bf1c20e143 100644 --- a/fs/fhandle.c +++ b/fs/fhandle.c @@ -35,7 +35,7 @@ static long do_sys_name_to_handle(struct path *path, return -EFAULT; if (f_handle.handle_bytes > MAX_HANDLE_SZ) - return -EINVAL; + f_handle.handle_bytes = MAX_HANDLE_SZ; handle = kmalloc(sizeof(struct file_handle) + f_handle.handle_bytes, GFP_KERNEL); diff --git a/include/linux/exportfs.h b/include/linux/exportfs.h index e7ab1b071c5e..097704e80c89 100644 --- a/include/linux/exportfs.h +++ b/include/linux/exportfs.h @@ -14,6 +14,10 @@ struct vfsmount; /* Must be larger than NFSv4 file handle, but small * enough for an on-stack allocation. overlayfs doesn't * want this too close to 255. + * NOTE: This value MUST NOT be exported to user-space. + * Applications must only ever see MAX_HANDLE_SZ == 128. + * If they try a larger number on older kernels, they + * will get -EINVAL which will be confusing. */ #define MAX_HANDLE_SZ 200 -- 2.14.0.rc0.dirty