brcmsmac: avoid function pointer casts

Message ID 20240213100548.457854-1-arnd@kernel.org
State New
Headers
Series brcmsmac: avoid function pointer casts |

Commit Message

Arnd Bergmann Feb. 13, 2024, 10:05 a.m. UTC
  From: Arnd Bergmann <arnd@arndb.de>

An old cleanup went a little too far and causes a warning with clang-16
and higher as it breaks control flow integrity (KCFI) rules:

drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c:64:34: error: cast from 'void (*)(struct brcms_phy *)' to 'void (*)(void *)' converts to incompatible function type [-Werror,-Wcast-function-type-strict]
   64 |                         brcms_init_timer(physhim->wl, (void (*)(void *))fn,
      |                                                       ^~~~~~~~~~~~~~~~~~~~

Change this one instance back to passing a void pointer so it can be
used with the timer callback interface.

Fixes: d89a4c80601d ("staging: brcm80211: removed void * from softmac phy")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 .../net/wireless/broadcom/brcm80211/brcmsmac/phy/phy_cmn.c   | 3 ++-
 drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c  | 5 ++---
 drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.h  | 2 +-
 3 files changed, 5 insertions(+), 5 deletions(-)
  

Comments

Kalle Valo Feb. 14, 2024, 8:45 a.m. UTC | #1
Arnd Bergmann <arnd@kernel.org> wrote:

> From: Arnd Bergmann <arnd@arndb.de>
> 
> An old cleanup went a little too far and causes a warning with clang-16
> and higher as it breaks control flow integrity (KCFI) rules:
> 
> drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c:64:34: error: cast from 'void (*)(struct brcms_phy *)' to 'void (*)(void *)' converts to incompatible function type [-Werror,-Wcast-function-type-strict]
>    64 |                         brcms_init_timer(physhim->wl, (void (*)(void *))fn,
>       |                                                       ^~~~~~~~~~~~~~~~~~~~
> 
> Change this one instance back to passing a void pointer so it can be
> used with the timer callback interface.
> 
> Fixes: d89a4c80601d ("staging: brcm80211: removed void * from softmac phy")
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>

I guess this should go to wireless tree?
  
Arend van Spriel Feb. 14, 2024, 9:23 a.m. UTC | #2
On 2/14/2024 9:45 AM, Kalle Valo wrote:
> Arnd Bergmann <arnd@kernel.org> wrote:
> 
>> From: Arnd Bergmann <arnd@arndb.de>
>>
>> An old cleanup went a little too far and causes a warning with clang-16
>> and higher as it breaks control flow integrity (KCFI) rules:
>>
>> drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c:64:34: error: cast from 'void (*)(struct brcms_phy *)' to 'void (*)(void *)' converts to incompatible function type [-Werror,-Wcast-function-type-strict]
>>     64 |                         brcms_init_timer(physhim->wl, (void (*)(void *))fn,
>>        |                                                       ^~~~~~~~~~~~~~~~~~~~
>>
>> Change this one instance back to passing a void pointer so it can be
>> used with the timer callback interface.
>>
>> Fixes: d89a4c80601d ("staging: brcm80211: removed void * from softmac phy")
>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> 
> I guess this should go to wireless tree?

This has been like this forever looking at the "staging" part in the 
Fixes tag. Is it really so urgent now? On the other hand I have no real 
problem with moving this to the wireless tree. Just wondering out loud.

Regards,
Arend
  
Arnd Bergmann Feb. 14, 2024, 3:07 p.m. UTC | #3
On Wed, Feb 14, 2024, at 10:23, Arend van Spriel wrote:
> On 2/14/2024 9:45 AM, Kalle Valo wrote:
>> Arnd Bergmann <arnd@kernel.org> wrote:
>> 
>>> From: Arnd Bergmann <arnd@arndb.de>
>>>
>>> An old cleanup went a little too far and causes a warning with clang-16
>>> and higher as it breaks control flow integrity (KCFI) rules:
>>>
>>> drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c:64:34: error: cast from 'void (*)(struct brcms_phy *)' to 'void (*)(void *)' converts to incompatible function type [-Werror,-Wcast-function-type-strict]
>>>     64 |                         brcms_init_timer(physhim->wl, (void (*)(void *))fn,
>>>        |                                                       ^~~~~~~~~~~~~~~~~~~~
>>>
>>> Change this one instance back to passing a void pointer so it can be
>>> used with the timer callback interface.
>>>
>>> Fixes: d89a4c80601d ("staging: brcm80211: removed void * from softmac phy")
>>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>> 
>> I guess this should go to wireless tree?
>
> This has been like this forever looking at the "staging" part in the 
> Fixes tag. Is it really so urgent now? On the other hand I have no real 
> problem with moving this to the wireless tree. Just wondering out loud.

It's probably fine either way. Some maintainers like to backport
the warning fixes to stable kernels, others don't. Since the
warning is currently only enabled at W=1 level, it's probably fine
to fix it for linux-next only, but if we want the fix backported,
it should also go into 6.8.

    Arnd
  
Kalle Valo Feb. 14, 2024, 5:55 p.m. UTC | #4
"Arnd Bergmann" <arnd@arndb.de> writes:

> On Wed, Feb 14, 2024, at 10:23, Arend van Spriel wrote:
>
>> On 2/14/2024 9:45 AM, Kalle Valo wrote:
>>> Arnd Bergmann <arnd@kernel.org> wrote:
>>> 
>>>> From: Arnd Bergmann <arnd@arndb.de>
>>>>
>>>> An old cleanup went a little too far and causes a warning with clang-16
>>>> and higher as it breaks control flow integrity (KCFI) rules:
>>>>
>>>> drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c:64:34: error: cast from 'void (*)(struct brcms_phy *)' to 'void (*)(void *)' converts to incompatible function type [-Werror,-Wcast-function-type-strict]
>>>>     64 |                         brcms_init_timer(physhim->wl, (void (*)(void *))fn,
>>>>        |                                                       ^~~~~~~~~~~~~~~~~~~~
>>>>
>>>> Change this one instance back to passing a void pointer so it can be
>>>> used with the timer callback interface.
>>>>
>>>> Fixes: d89a4c80601d ("staging: brcm80211: removed void * from softmac phy")
>>>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>>> 
>>> I guess this should go to wireless tree?
>>
>> This has been like this forever looking at the "staging" part in the 
>> Fixes tag. Is it really so urgent now? On the other hand I have no real 
>> problem with moving this to the wireless tree. Just wondering out loud.
>
> It's probably fine either way. Some maintainers like to backport
> the warning fixes to stable kernels, others don't. Since the
> warning is currently only enabled at W=1 level, it's probably fine
> to fix it for linux-next only, but if we want the fix backported,
> it should also go into 6.8.

Thanks, let's take it wireless-next then.
  
Kalle Valo Feb. 15, 2024, 10:33 a.m. UTC | #5
Arnd Bergmann <arnd@kernel.org> wrote:

> From: Arnd Bergmann <arnd@arndb.de>
> 
> An old cleanup went a little too far and causes a warning with clang-16
> and higher as it breaks control flow integrity (KCFI) rules:
> 
> drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c:64:34: error: cast from 'void (*)(struct brcms_phy *)' to 'void (*)(void *)' converts to incompatible function type [-Werror,-Wcast-function-type-strict]
>    64 |                         brcms_init_timer(physhim->wl, (void (*)(void *))fn,
>       |                                                       ^~~~~~~~~~~~~~~~~~~~
> 
> Change this one instance back to passing a void pointer so it can be
> used with the timer callback interface.
> 
> Fixes: d89a4c80601d ("staging: brcm80211: removed void * from softmac phy")
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>

I can add 'wifi:' during commit. Arend, ack?
  
Arend van Spriel Feb. 15, 2024, 1:48 p.m. UTC | #6
On February 15, 2024 11:33:30 AM Kalle Valo <kvalo@kernel.org> wrote:

> Arnd Bergmann <arnd@kernel.org> wrote:
>
>> From: Arnd Bergmann <arnd@arndb.de>
>>
>> An old cleanup went a little too far and causes a warning with clang-16
>> and higher as it breaks control flow integrity (KCFI) rules:
>>
>> drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c:64:34: error: 
>> cast from 'void (*)(struct brcms_phy *)' to 'void (*)(void *)' converts to 
>> incompatible function type [-Werror,-Wcast-function-type-strict]
>> 64 |                         brcms_init_timer(physhim->wl, (void (*)(void 
>> *))fn,
>> |                                                       ^~~~~~~~~~~~~~~~~~~~
>>
>> Change this one instance back to passing a void pointer so it can be
>> used with the timer callback interface.
>>
>> Fixes: d89a4c80601d ("staging: brcm80211: removed void * from softmac phy")
>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>
> I can add 'wifi:' during commit. Arend, ack?

Here it is:

Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com>

>
> --
> https://patchwork.kernel.org/project/linux-wireless/patch/20240213100548.457854-1-arnd@kernel.org/
>
> https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
  
Kalle Valo Feb. 16, 2024, 3:49 p.m. UTC | #7
Arnd Bergmann <arnd@kernel.org> wrote:

> From: Arnd Bergmann <arnd@arndb.de>
> 
> An old cleanup went a little too far and causes a warning with clang-16
> and higher as it breaks control flow integrity (KCFI) rules:
> 
> drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c:64:34: error: cast from 'void (*)(struct brcms_phy *)' to 'void (*)(void *)' converts to incompatible function type [-Werror,-Wcast-function-type-strict]
>    64 |                         brcms_init_timer(physhim->wl, (void (*)(void *))fn,
>       |                                                       ^~~~~~~~~~~~~~~~~~~~
> 
> Change this one instance back to passing a void pointer so it can be
> used with the timer callback interface.
> 
> Fixes: d89a4c80601d ("staging: brcm80211: removed void * from softmac phy")
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com>

Patch applied to wireless-next.git, thanks.

e1ea6db35fc3 wifi: brcmsmac: avoid function pointer casts
  

Patch

diff --git a/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy/phy_cmn.c b/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy/phy_cmn.c
index 07f83ff5a54a..a27d6f0b8819 100644
--- a/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy/phy_cmn.c
+++ b/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy/phy_cmn.c
@@ -383,8 +383,9 @@  struct shared_phy *wlc_phy_shared_attach(struct shared_phy_params *shp)
 	return sh;
 }
 
-static void wlc_phy_timercb_phycal(struct brcms_phy *pi)
+static void wlc_phy_timercb_phycal(void *ptr)
 {
+	struct brcms_phy *pi = ptr;
 	uint delay = 5;
 
 	if (PHY_PERICAL_MPHASE_PENDING(pi)) {
diff --git a/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c b/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c
index a0de5db0cd64..b72381791536 100644
--- a/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c
+++ b/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.c
@@ -57,12 +57,11 @@  void wlc_phy_shim_detach(struct phy_shim_info *physhim)
 }
 
 struct wlapi_timer *wlapi_init_timer(struct phy_shim_info *physhim,
-				     void (*fn)(struct brcms_phy *pi),
+				     void (*fn)(void *pi),
 				     void *arg, const char *name)
 {
 	return (struct wlapi_timer *)
-			brcms_init_timer(physhim->wl, (void (*)(void *))fn,
-					 arg, name);
+			brcms_init_timer(physhim->wl, fn, arg, name);
 }
 
 void wlapi_free_timer(struct wlapi_timer *t)
diff --git a/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.h b/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.h
index dd8774717ade..27d0934e600e 100644
--- a/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.h
+++ b/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy_shim.h
@@ -131,7 +131,7 @@  void wlc_phy_shim_detach(struct phy_shim_info *physhim);
 
 /* PHY to WL utility functions */
 struct wlapi_timer *wlapi_init_timer(struct phy_shim_info *physhim,
-				     void (*fn)(struct brcms_phy *pi),
+				     void (*fn)(void *pi),
 				     void *arg, const char *name);
 void wlapi_free_timer(struct wlapi_timer *t);
 void wlapi_add_timer(struct wlapi_timer *t, uint ms, int periodic);