Hi Dan,
> > No, that doesn't work because the comparison against sizeof() get's
> > promoted to size_t. In other words, negative values still count as
> > large positive values.
>
> You are right. The negative value counts as a large positive number while comparing against sizeof().
> I can add a "const int hdr_size" variable to store the value of sizeof(...) and compare resp_len to
> hdr_size. The "sizeof(...)" has been used multiple times in this function, so I think it's worth
> adding a variable for it.
Well, there is another problem here.
We might have accessed invalid memory while handling switch case in last iteration, because we didn't count tlv_len in "while(...)" condition check.
To fix this, Dan's change makes sense. But we should check the length before accessing the buffer.
@@ -722,6 +722,9 @@ int mwifiex_ret_wmm_get_status(struct mwifiex_private *priv,
tlv_hdr = (struct mwifiex_ie_types_data *) curr;
tlv_len = le16_to_cpu(tlv_hdr->header.len);
+ if (resp_len < tlv_len + sizeof(tlv_hdr->header))
+ break;
+
switch (le16_to_cpu(tlv_hdr->header.type)) {
case TLV_TYPE_WMMQSTATUS:
tlv_wmm_qstatus =
Thanks,
Bing
--
To unsubscribe from this list: send the line "unsubscribe kernel-janitors" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
