On 20. Aug 2024, at 16:52, Tom Talpey <tom@xxxxxxxxxx> wrote:
On 8/20/2024 10:11 AM, Namjae Jeon wrote:
On Mon, Aug 19, 2024 at 1:22 AM Thorsten Blum <thorsten.blum@xxxxxxxxxx> wrote:
We can use <= instead of +1.
Replace the deprecated one-element arrays with flexible-array members
in the structs copychunk_ioctl_req and smb2_ea_info_req.
There are no binary differences after this conversion.
Link: https://github.com/KSPP/linux/issues/79
Signed-off-by: Thorsten Blum <thorsten.blum@xxxxxxxxxx>
---
fs/smb/server/smb2pdu.c | 4 ++--
fs/smb/server/smb2pdu.h | 4 ++--
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/fs/smb/server/smb2pdu.c b/fs/smb/server/smb2pdu.c
index 2df1354288e6..83667cb78fa6 100644
--- a/fs/smb/server/smb2pdu.c
+++ b/fs/smb/server/smb2pdu.c
@@ -4580,7 +4580,7 @@ static int smb2_get_ea(struct ksmbd_work *work, struct ksmbd_file *fp,
/* single EA entry is requested with given user.* name */
if (req->InputBufferLength) {
if (le32_to_cpu(req->InputBufferLength) <
- sizeof(struct smb2_ea_info_req))
+ sizeof(struct smb2_ea_info_req) + 1)
This is better, but maybe this test was actually not right in
the first place.
I think a strict "<" is correct here, because the ea name
field is a counted array of length EaNameLength. So, it's
a layering issue to fail with EINVAL this early in the
processing. All that should be checked up front is
that a complete smb2_ea_info_req header is present.
Just to clarify before I submit a v2: Is a strict "<" and without "+1"
correct?
return -EINVAL;Ditto.
ea_req = (struct smb2_ea_info_req *)((char *)req +
@@ -8083,7 +8083,7 @@ int smb2_ioctl(struct ksmbd_work *work)
goto out;
}
- if (in_buf_len < sizeof(struct copychunk_ioctl_req)) {
+ if (in_buf_len < sizeof(struct copychunk_ioctl_req) + 1) {
And ditto.
Same here, strict "<" and without "+ 1"? Or just a refactor to "<="
without changing the condition?
Thanks,
Thorsten