2017-07-13 07:46:16

by Dan Carpenter

[permalink] [raw]
Subject: [PATCH] rtlwifi: rtl8821ae: Fix HW_VAR_NAV_UPPER operation

The cast here is wrong. We want to cast the pointer but we accidentally
do a no-op cast of the value. We normally want to set us_nav_upper to
WIFI_NAV_UPPER_US (30000) but because of this bug we instead set it to
184 on little endian systems and 0 on big endian ones.

Fixes: 3c05bedb5fef ("Staging: rtl8812ae: Add Realtek 8821 PCI WIFI driver")
Signed-off-by: Dan Carpenter <[email protected]>

diff --git a/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c b/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c
index 2bc6bace069c..230f92a48f3c 100644
--- a/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c
+++ b/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c
@@ -779,7 +779,7 @@ void rtl8821ae_set_hw_reg(struct ieee80211_hw *hw, u8 variable, u8 *val)
_rtl8821ae_resume_tx_beacon(hw);
break; }
case HW_VAR_NAV_UPPER: {
- u32 us_nav_upper = ((u32)*val);
+ u32 us_nav_upper = *(u32 *)val;

if (us_nav_upper > HAL_92C_NAV_UPPER_UNIT * 0xFF) {
RT_TRACE(rtlpriv, COMP_INIT , DBG_WARNING,


2017-07-14 19:09:48

by Larry Finger

[permalink] [raw]
Subject: Re: [PATCH] rtlwifi: rtl8821ae: Fix HW_VAR_NAV_UPPER operation

On 07/13/2017 02:43 AM, Dan Carpenter wrote:
> The cast here is wrong. We want to cast the pointer but we accidentally
> do a no-op cast of the value. We normally want to set us_nav_upper to
> WIFI_NAV_UPPER_US (30000) but because of this bug we instead set it to
> 184 on little endian systems and 0 on big endian ones.
>
> Fixes: 3c05bedb5fef ("Staging: rtl8812ae: Add Realtek 8821 PCI WIFI driver")
> Signed-off-by: Dan Carpenter <[email protected]>

Acked-by: Larry Finger <[email protected]>

Thanks for catching this error. I suspect a typo, but your fix is obviously correct.

Does the "Fixes" notation imply a Cc to stable? If not, then the patch needs a
Cc to stable for 3.14+.

Thanks again,

Larry

2017-07-28 15:43:19

by Kalle Valo

[permalink] [raw]
Subject: Re: rtlwifi: rtl8821ae: Fix HW_VAR_NAV_UPPER operation

Dan Carpenter <[email protected]> wrote:

> The cast here is wrong. We want to cast the pointer but we accidentally
> do a no-op cast of the value. We normally want to set us_nav_upper to
> WIFI_NAV_UPPER_US (30000) but because of this bug we instead set it to
> 184 on little endian systems and 0 on big endian ones.
>
> Fixes: 3c05bedb5fef ("Staging: rtl8812ae: Add Realtek 8821 PCI WIFI driver")
> Signed-off-by: Dan Carpenter <[email protected]>
> Acked-by: Larry Finger <[email protected]>
>
> diff --git a/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c b/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c
> index 2bc6bace069c..230f92a48f3c 100644
> --- a/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c
> +++ b/drivers/net/wireless/realtek/rtlwifi/rtl8821ae/hw.c
> @@ -779,7 +779,7 @@ void rtl8821ae_set_hw_reg(struct ieee80211_hw *hw, u8 variable, u8 *val)
> _rtl8821ae_resume_tx_beacon(hw);
> break; }
> case HW_VAR_NAV_UPPER: {
> - u32 us_nav_upper = ((u32)*val);
> + u32 us_nav_upper = *(u32 *)val;
>
> if (us_nav_upper > HAL_92C_NAV_UPPER_UNIT * 0xFF) {
> RT_TRACE(rtlpriv, COMP_INIT , DBG_WARNING,

Patch applied to wireless-drivers-next.git, thanks.

d28ac7be15c7 rtlwifi: rtl8821ae: Fix HW_VAR_NAV_UPPER operation

--
https://patchwork.kernel.org/patch/9837847/

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches

2017-07-15 07:43:33

by Dan Carpenter

[permalink] [raw]
Subject: Re: [PATCH] rtlwifi: rtl8821ae: Fix HW_VAR_NAV_UPPER operation

On Fri, Jul 14, 2017 at 02:09:46PM -0500, Larry Finger wrote:
> On 07/13/2017 02:43 AM, Dan Carpenter wrote:
> > The cast here is wrong. We want to cast the pointer but we accidentally
> > do a no-op cast of the value. We normally want to set us_nav_upper to
> > WIFI_NAV_UPPER_US (30000) but because of this bug we instead set it to
> > 184 on little endian systems and 0 on big endian ones.
> >
> > Fixes: 3c05bedb5fef ("Staging: rtl8812ae: Add Realtek 8821 PCI WIFI driver")
> > Signed-off-by: Dan Carpenter <[email protected]>
>
> Acked-by: Larry Finger <[email protected]>
>
> Thanks for catching this error. I suspect a typo, but your fix is obviously correct.
>
> Does the "Fixes" notation imply a Cc to stable?

No. Fixes doesn't necessarily imply a Cc to stable.

regards,
dan carpenter