[PATCH] ufs: core: Initialize a variable mode for PA_PWRMODE

[PATCH] ufs: core: Initialize a variable mode for PA_PWRMODE

[Wonkon Kim]

If ufshcd_dme_get() fails uic cmd error,
a variable mode has a garbage value.
It may return unintended result for pwr mode restore.

Initialize it as 0 and will return true when ufshcd_dme_get() fails,
because PA power mode 0 is not defined.

Signed-off-by: Wonkon Kim <wkon.kim@samsung.com>
---
 drivers/ufs/core/ufshcd.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 9a43102b2b21..a4438a3cb73a 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -6581,7 +6581,7 @@ static inline void ufshcd_recover_pm_error(struct ufs_hba *hba)
 static bool ufshcd_is_pwr_mode_restore_needed(struct ufs_hba *hba)
 {
 	struct ufs_pa_layer_attr *pwr_info = &hba->pwr_info;
-	u32 mode;
+	u32 mode = 0;
 
 	ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &mode);
 
-- 
2.34.1

Re: [PATCH] ufs: core: Initialize a variable mode for PA_PWRMODE

[Peter Wang (王信友)]

On Thu, 2025-10-02 at 16:00 +0900, Wonkon Kim wrote:
> 
> If ufshcd_dme_get() fails uic cmd error,
> a variable mode has a garbage value.
> It may return unintended result for pwr mode restore.
> 
> Initialize it as 0 and will return true when ufshcd_dme_get() fails,
> because PA power mode 0 is not defined.
> 
> Signed-off-by: Wonkon Kim <wkon.kim@samsung.com>
> ---
>  drivers/ufs/core/ufshcd.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> index 9a43102b2b21..a4438a3cb73a 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -6581,7 +6581,7 @@ static inline void
> ufshcd_recover_pm_error(struct ufs_hba *hba)
>  static bool ufshcd_is_pwr_mode_restore_needed(struct ufs_hba *hba)
>  {
>         struct ufs_pa_layer_attr *pwr_info = &hba->pwr_info;
> -       u32 mode;
> +       u32 mode = 0;
> 
>         ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &mode);
> 
> --
> 2.34.1
> 

Reviewed-by: Peter Wang <peter.wang@mediatek.com>

Re: [PATCH] ufs: core: Initialize a variable mode for PA_PWRMODE

[Bart Van Assche]

On 10/2/25 12:00 AM, Wonkon Kim wrote:
>   static bool ufshcd_is_pwr_mode_restore_needed(struct ufs_hba *hba)
>   {
>   	struct ufs_pa_layer_attr *pwr_info = &hba->pwr_info;
> -	u32 mode;
> +	u32 mode = 0;
>   
>   	ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &mode);

Wouldn't it be better to check the ufshcd_dme_get() return value rather
than to zero-initialize 'mode'? I think that would make 
ufshcd_is_pwr_mode_restore_needed() easier to read compared to applying
the above patch.

Thanks,

Bart.

Re: [PATCH] ufs: core: Initialize a variable mode for PA_PWRMODE

[Bart Van Assche]

On 10/2/25 12:00 AM, Wonkon Kim wrote:
>   static bool ufshcd_is_pwr_mode_restore_needed(struct ufs_hba *hba)
>   {
>   	struct ufs_pa_layer_attr *pwr_info = &hba->pwr_info;
> -	u32 mode;
> +	u32 mode = 0;
>   
>   	ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &mode);

Since there is more code that passes a pointer to an uninitialized
variable to ufshcd_dme_get(), the untested patch below may be a better
solution:

diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 127b691402f9..5226fbca29ec 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -4277,8 +4277,8 @@ int ufshcd_dme_get_attr(struct ufs_hba *hba, u32 
attr_sel,
  			get, UIC_GET_ATTR_ID(attr_sel),
  			UFS_UIC_COMMAND_RETRIES - retries);

-	if (mib_val && !ret)
-		*mib_val = uic_cmd.argument3;
+	if (mib_val)
+		*mib_val = ret == 0 ? uic_cmd.argument3 : 0;

  	if (peer && (hba->quirks & UFSHCD_QUIRK_DME_PEER_ACCESS_AUTO_MODE)
  	    && pwr_mode_change)

RE: [PATCH] ufs: core: Initialize a variable mode for PA_PWRMODE

[Wonkon Kim]

> On 10/2/25 12:00 AM, Wonkon Kim wrote:
> >   static bool ufshcd_is_pwr_mode_restore_needed(struct ufs_hba *hba)
> >   {
> >   	struct ufs_pa_layer_attr *pwr_info = &hba->pwr_info;
> > -	u32 mode;
> > +	u32 mode = 0;
> >
> >   	ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &mode);
> 
> Since there is more code that passes a pointer to an uninitialized
> variable to ufshcd_dme_get(), the untested patch below may be a better
> solution:
> 
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c index
> 127b691402f9..5226fbca29ec 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -4277,8 +4277,8 @@ int ufshcd_dme_get_attr(struct ufs_hba *hba, u32
> attr_sel,
>   			get, UIC_GET_ATTR_ID(attr_sel),
>   			UFS_UIC_COMMAND_RETRIES - retries);
> 
> -	if (mib_val && !ret)
> -		*mib_val = uic_cmd.argument3;
> +	if (mib_val)
> +		*mib_val = ret == 0 ? uic_cmd.argument3 : 0;
> 
>   	if (peer && (hba->quirks & UFSHCD_QUIRK_DME_PEER_ACCESS_AUTO_MODE)
>   	    && pwr_mode_change)
> 
> 

Sorry for late reply.
Ok, I agree with you and it would be more general.
There is a code to initialize a variable.
Do you think it also needs to fix?

diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index a4438a3cb73a..d593ff7ea63d 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -4990,7 +4990,7 @@ EXPORT_SYMBOL_GPL(ufshcd_hba_enable);
 
 static int ufshcd_disable_tx_lcc(struct ufs_hba *hba, bool peer)
 {
-       int tx_lanes = 0, i, err = 0;
+       int tx_lanes, i, err = 0;
 
        if (!peer)
		ufshcd_dme_get(hba, UIC_ARG_MIB(PA_CONNECTEDTXDATALANES),
			       &tx_lanes);
	else
		ufshcd_dme_peer_get(hba, UIC_ARG_MIB(PA_CONNECTEDTXDATALANES),
				    &tx_lanes);

Thanks,
Wonkon Kim.

RE: [PATCH] ufs: core: Initialize a variable mode for PA_PWRMODE

[Wonkon Kim]

> On 10/2/25 12:00 AM, Wonkon Kim wrote:
> >   static bool ufshcd_is_pwr_mode_restore_needed(struct ufs_hba *hba)
> >   {
> >   	struct ufs_pa_layer_attr *pwr_info = &hba->pwr_info;
> > -	u32 mode;
> > +	u32 mode = 0;
> >
> >   	ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &mode);
> 
> Since there is more code that passes a pointer to an uninitialized
> variable to ufshcd_dme_get(), the untested patch below may be a better
> solution:
> 
> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c index
> 127b691402f9..5226fbca29ec 100644
> --- a/drivers/ufs/core/ufshcd.c
> +++ b/drivers/ufs/core/ufshcd.c
> @@ -4277,8 +4277,8 @@ int ufshcd_dme_get_attr(struct ufs_hba *hba, u32
> attr_sel,
>   			get, UIC_GET_ATTR_ID(attr_sel),
>   			UFS_UIC_COMMAND_RETRIES - retries);
> 
> -	if (mib_val && !ret)
> -		*mib_val = uic_cmd.argument3;
> +	if (mib_val)
> +		*mib_val = ret == 0 ? uic_cmd.argument3 : 0;
> 
>   	if (peer && (hba->quirks & UFSHCD_QUIRK_DME_PEER_ACCESS_AUTO_MODE)
>   	    && pwr_mode_change)
> 
> 

There are some attributes to use 0 as valid value.
e.g. PA_MAXRXHSGEAR is set to 0 for NO_HS=0
If it has 0 for valid value, most of value 0 are regarded as FALSE, unsupported or minimum.
And these cases seems to check ret for command success/fail in code.
However, is it ok to set 0 for ufshcd_send_uic_cmd() fail?

If it can't, it needs to initialize mode.
Value 0 for PA_PWRMODE is invalid.


Thanks,
Wonkon Kim.

Re: [PATCH] ufs: core: Initialize a variable mode for PA_PWRMODE

[Bart Van Assche]

On 10/13/25 1:20 AM, Wonkon Kim wrote:
>> On 10/2/25 12:00 AM, Wonkon Kim wrote:
>>>    static bool ufshcd_is_pwr_mode_restore_needed(struct ufs_hba *hba)
>>>    {
>>>    	struct ufs_pa_layer_attr *pwr_info = &hba->pwr_info;
>>> -	u32 mode;
>>> +	u32 mode = 0;
>>>
>>>    	ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &mode);
>>
>> Since there is more code that passes a pointer to an uninitialized
>> variable to ufshcd_dme_get(), the untested patch below may be a better
>> solution:
>>
>> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c index
>> 127b691402f9..5226fbca29ec 100644
>> --- a/drivers/ufs/core/ufshcd.c
>> +++ b/drivers/ufs/core/ufshcd.c
>> @@ -4277,8 +4277,8 @@ int ufshcd_dme_get_attr(struct ufs_hba *hba, u32
>> attr_sel,
>>    			get, UIC_GET_ATTR_ID(attr_sel),
>>    			UFS_UIC_COMMAND_RETRIES - retries);
>>
>> -	if (mib_val && !ret)
>> -		*mib_val = uic_cmd.argument3;
>> +	if (mib_val)
>> +		*mib_val = ret == 0 ? uic_cmd.argument3 : 0;
>>
>>    	if (peer && (hba->quirks & UFSHCD_QUIRK_DME_PEER_ACCESS_AUTO_MODE)
>>    	    && pwr_mode_change)
>>
>>
> 
> There are some attributes to use 0 as valid value.
> e.g. PA_MAXRXHSGEAR is set to 0 for NO_HS=0
> If it has 0 for valid value, most of value 0 are regarded as FALSE, unsupported or minimum.
> And these cases seems to check ret for command success/fail in code.
> However, is it ok to set 0 for ufshcd_send_uic_cmd() fail?
> 
> If it can't, it needs to initialize mode.
> Value 0 for PA_PWRMODE is invalid.

Hi Wonkon,

Modifying ufshcd_dme_get_attr() doesn't exclude checking the return
value of ufshcd_dme_get_attr(). I propose to modify
ufshcd_dme_get_attr() such that it always initializes *mib_val and also
to check the ufshcd_dme_get_attr() return value wherever appropriate.

Thanks,

Bart.

RE: [PATCH] ufs: core: Initialize a variable mode for PA_PWRMODE

[Wonkon Kim]

> On 10/13/25 1:20 AM, Wonkon Kim wrote:
> >> On 10/2/25 12:00 AM, Wonkon Kim wrote:
> >>>    static bool ufshcd_is_pwr_mode_restore_needed(struct ufs_hba *hba)
> >>>    {
> >>>    	struct ufs_pa_layer_attr *pwr_info = &hba->pwr_info;
> >>> -	u32 mode;
> >>> +	u32 mode = 0;
> >>>
> >>>    	ufshcd_dme_get(hba, UIC_ARG_MIB(PA_PWRMODE), &mode);
> >>
> >> Since there is more code that passes a pointer to an uninitialized
> >> variable to ufshcd_dme_get(), the untested patch below may be a
> >> better
> >> solution:
> >>
> >> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> >> index 127b691402f9..5226fbca29ec 100644
> >> --- a/drivers/ufs/core/ufshcd.c
> >> +++ b/drivers/ufs/core/ufshcd.c
> >> @@ -4277,8 +4277,8 @@ int ufshcd_dme_get_attr(struct ufs_hba *hba,
> >> u32 attr_sel,
> >>    			get, UIC_GET_ATTR_ID(attr_sel),
> >>    			UFS_UIC_COMMAND_RETRIES - retries);
> >>
> >> -	if (mib_val && !ret)
> >> -		*mib_val = uic_cmd.argument3;
> >> +	if (mib_val)
> >> +		*mib_val = ret == 0 ? uic_cmd.argument3 : 0;
> >>
> >>    	if (peer && (hba->quirks & UFSHCD_QUIRK_DME_PEER_ACCESS_AUTO_MODE)
> >>    	    && pwr_mode_change)
> >>
> >>
> >
> > There are some attributes to use 0 as valid value.
> > e.g. PA_MAXRXHSGEAR is set to 0 for NO_HS=0 If it has 0 for valid
> > value, most of value 0 are regarded as FALSE, unsupported or minimum.
> > And these cases seems to check ret for command success/fail in code.
> > However, is it ok to set 0 for ufshcd_send_uic_cmd() fail?
> >
> > If it can't, it needs to initialize mode.
> > Value 0 for PA_PWRMODE is invalid.
> 
> Hi Wonkon,
> 
> Modifying ufshcd_dme_get_attr() doesn't exclude checking the return value
> of ufshcd_dme_get_attr(). I propose to modify
> ufshcd_dme_get_attr() such that it always initializes *mib_val and also to
> check the ufshcd_dme_get_attr() return value wherever appropriate.
> 
> Thanks,
> 
> Bart.

Hi Bart,
I got it. I'll update it.


Thanks,
Wonkon Kim.