[PATCH v3] staging: rtl8723bs: Fix return type for implementation of ndo_start_xmit

[PATCH v3] staging: rtl8723bs: Fix return type for implementation of ndo_start_xmit

[GUO Zihua]

CFI (Control Flow Integrity) is a safety feature allowing the system to
detect and react should a potential control flow hijacking occurs. In
particular, the Forward-Edge CFI protects indirect function calls by
ensuring the prototype of function that is actually called matches the
definition of the function hook.

Since Linux now supports CFI, it will be a good idea to fix mismatched
return type for implementation of hooks. Otherwise this would get
cought out by CFI and cause a panic.

As returns from _rtw_xmit_entry() would always be 0, remove unneeded
variable ret and use enums from netdev_tx_t as return value instead.
Then change return type to netdev_tx_t.

Signed-off-by: GUO Zihua <guozihua@huawei.com>
---

v3:
  Provide detail on CFI.

v2:
  Fix truncated subject.

---
 drivers/staging/rtl8723bs/include/xmit_osdep.h | 2 +-
 drivers/staging/rtl8723bs/os_dep/xmit_linux.c  | 8 +++-----
 2 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/drivers/staging/rtl8723bs/include/xmit_osdep.h b/drivers/staging/rtl8723bs/include/xmit_osdep.h
index e781cd5dfd01..7ac8cf589f90 100644
--- a/drivers/staging/rtl8723bs/include/xmit_osdep.h
+++ b/drivers/staging/rtl8723bs/include/xmit_osdep.h
@@ -26,7 +26,7 @@ struct xmit_frame;
 struct xmit_buf;
 
 extern int _rtw_xmit_entry(struct sk_buff *pkt, struct net_device *pnetdev);
-extern int rtw_xmit_entry(struct sk_buff *pkt, struct net_device *pnetdev);
+extern netdev_tx_t rtw_xmit_entry(struct sk_buff *pkt, struct net_device *pnetdev);
 
 void rtw_os_xmit_schedule(struct adapter *padapter);
 
diff --git a/drivers/staging/rtl8723bs/os_dep/xmit_linux.c b/drivers/staging/rtl8723bs/os_dep/xmit_linux.c
index 530e7a6c67c5..d42fd0878dcd 100644
--- a/drivers/staging/rtl8723bs/os_dep/xmit_linux.c
+++ b/drivers/staging/rtl8723bs/os_dep/xmit_linux.c
@@ -220,12 +220,10 @@ int _rtw_xmit_entry(struct sk_buff *pkt, struct net_device *pnetdev)
 	return 0;
 }
 
-int rtw_xmit_entry(struct sk_buff *pkt, struct net_device *pnetdev)
+netdev_tx_t rtw_xmit_entry(struct sk_buff *pkt, struct net_device *pnetdev)
 {
-	int ret = 0;
-
 	if (pkt)
-		ret = _rtw_xmit_entry(pkt, pnetdev);
+		_rtw_xmit_entry(pkt, pnetdev);
 
-	return ret;
+	return NETDEV_TX_OK;
 }
-- 
2.17.1

Re: [PATCH v3] staging: rtl8723bs: Fix return type for implementation of ndo_start_xmit

[Greg KH]

On Mon, Sep 05, 2022 at 09:03:10PM +0800, GUO Zihua wrote:
> CFI (Control Flow Integrity) is a safety feature allowing the system to
> detect and react should a potential control flow hijacking occurs. In
> particular, the Forward-Edge CFI protects indirect function calls by
> ensuring the prototype of function that is actually called matches the
> definition of the function hook.
> 
> Since Linux now supports CFI, it will be a good idea to fix mismatched
> return type for implementation of hooks. Otherwise this would get
> cought out by CFI and cause a panic.
> 
> As returns from _rtw_xmit_entry() would always be 0, remove unneeded
> variable ret and use enums from netdev_tx_t as return value instead.
> Then change return type to netdev_tx_t.

If _rtw_xmit_entry() can never fail, then it needs to not have a return
value at all, and that needs to also be fixed.  Can you make this a
patch series that does the return type change for _rtw_xmit_entry()
first, and then this change on top of that one.

thanks,

greg k-h

Re: [PATCH v3] staging: rtl8723bs: Fix return type for implementation of ndo_start_xmit

[Guozihua (Scott)]

On 2022/9/9 15:46, Greg KH wrote:
> On Mon, Sep 05, 2022 at 09:03:10PM +0800, GUO Zihua wrote:
>> CFI (Control Flow Integrity) is a safety feature allowing the system to
>> detect and react should a potential control flow hijacking occurs. In
>> particular, the Forward-Edge CFI protects indirect function calls by
>> ensuring the prototype of function that is actually called matches the
>> definition of the function hook.
>>
>> Since Linux now supports CFI, it will be a good idea to fix mismatched
>> return type for implementation of hooks. Otherwise this would get
>> cought out by CFI and cause a panic.
>>
>> As returns from _rtw_xmit_entry() would always be 0, remove unneeded
>> variable ret and use enums from netdev_tx_t as return value instead.
>> Then change return type to netdev_tx_t.
> 
> If _rtw_xmit_entry() can never fail, then it needs to not have a return
> value at all, and that needs to also be fixed.  Can you make this a
> patch series that does the return type change for _rtw_xmit_entry()
> first, and then this change on top of that one.
> 
> thanks,
> 
> greg k-h
> .

Sure Greg

-- 
Best
GUO Zihua