Re: [PATCH] platform/x86: Add support for improved performance mode

[Mark Pearson]

Hi Ilpo,

On Thu, Nov 9, 2023, at 5:10 AM, Ilpo Järvinen wrote:
> On Wed, 8 Nov 2023, Mark Pearson wrote:
>
>> Some new Thinkpads have a new improved performance mode available.
>> Add support to make this mode usable.
>> 
>> To avoid having to create a new profile, just use the improved performance
>> mode in place of the existing performance mode, when available.
>> 
>> Tested on T14 AMD G4 AMD.
>> 
>> Signed-off-by: Mark Pearson <mpearson-lenovo@xxxxxxxxx>
>> ---
>>  drivers/platform/x86/thinkpad_acpi.c | 28 ++++++++++++++++++++++++++++
>>  1 file changed, 28 insertions(+)
>> 
>> diff --git a/drivers/platform/x86/thinkpad_acpi.c b/drivers/platform/x86/thinkpad_acpi.c
>> index ad460417f901..eba701ab340e 100644
>> --- a/drivers/platform/x86/thinkpad_acpi.c
>> +++ b/drivers/platform/x86/thinkpad_acpi.c
>> @@ -10136,6 +10136,7 @@ static struct ibm_struct proxsensor_driver_data = {
>>  
>>  #define DYTC_CMD_SET          1 /* To enable/disable IC function mode */
>>  #define DYTC_CMD_MMC_GET      8 /* To get current MMC function and mode */
>> +#define DYTC_CMD_UP_CAP     0xA /* To get Ultra-performance capability */
>>  #define DYTC_CMD_RESET    0x1ff /* To reset back to default */
>>  
>>  #define DYTC_CMD_FUNC_CAP     3 /* To get DYTC capabilities */
>> @@ -10152,6 +10153,7 @@ static struct ibm_struct proxsensor_driver_data = {
>>  
>>  #define DYTC_FUNCTION_STD     0  /* Function = 0, standard mode */
>>  #define DYTC_FUNCTION_CQL     1  /* Function = 1, lap mode */
>> +#define DYTC_FUNCTION_TMS     9  /* Function = 9, TMS mode */
>>  #define DYTC_FUNCTION_MMC     11 /* Function = 11, MMC mode */
>>  #define DYTC_FUNCTION_PSC     13 /* Function = 13, PSC mode */
>>  #define DYTC_FUNCTION_AMT     15 /* Function = 15, AMT mode */
>> @@ -10163,11 +10165,14 @@ static struct ibm_struct proxsensor_driver_data = {
>>  #define DYTC_MODE_MMC_LOWPOWER 3  /* Low power mode */
>>  #define DYTC_MODE_MMC_BALANCE  0xF  /* Default mode aka balanced */
>>  #define DYTC_MODE_MMC_DEFAULT  0  /* Default mode from MMC_GET, aka balanced */
>> +#define DYTC_NOMODE            0xF  /* When Function does not have a mode */
>>  
>>  #define DYTC_MODE_PSC_LOWPOWER 3  /* Low power mode */
>>  #define DYTC_MODE_PSC_BALANCE  5  /* Default mode aka balanced */
>>  #define DYTC_MODE_PSC_PERFORM  7  /* High power mode aka performance */
>>  
>> +#define DYTC_UP_SUPPORT_BIT    8  /* Bit 8 - 1 = supported, 0 = not */
>
> It would be preferrable to comment what is supported rather than have a 
> comment like above which isn't particularly helpful.

OK - so  just have:
#define DYTC_UP_SUPPORT_BIT    8  /* Ultra-performance (TMS) mode support */

Or...reading ahead in the review this should actually be
#define DYTC_UP_SUPPORT_BIT    BIT(8)  /* Ultra-performance (TMS) mode support */

>
>>  #define DYTC_ERR_MASK       0xF  /* Bits 0-3 in cmd result are the error result */
>>  #define DYTC_ERR_SUCCESS      1  /* CMD completed successful */
>>  
>> @@ -10185,6 +10190,7 @@ static enum platform_profile_option dytc_current_profile;
>>  static atomic_t dytc_ignore_event = ATOMIC_INIT(0);
>>  static DEFINE_MUTEX(dytc_mutex);
>>  static int dytc_capabilities;
>> +static bool dytc_ultraperf_cap; /* ultra performance capable */
>>  static bool dytc_mmc_get_available;
>>  static int profile_force;
>>  
>> @@ -10355,6 +10361,17 @@ static int dytc_profile_set(struct platform_profile_handler *pprof,
>>  	if (err)
>>  		goto unlock;
>>  
>> +	/* Set TMS mode appropriately (enable for performance), if available */
>> +	if (dytc_ultraperf_cap) {
>> +		int cmd;
>> +
>> +		cmd = DYTC_SET_COMMAND(DYTC_FUNCTION_TMS, DYTC_NOMODE,
>> +				       profile == PLATFORM_PROFILE_PERFORMANCE);
>> +		err = dytc_command(cmd, &output);
>> +		if (err)
>> +			return err;
>> +	}
>> +
>>  	if (dytc_capabilities & BIT(DYTC_FC_MMC)) {
>>  		if (profile == PLATFORM_PROFILE_BALANCED) {
>>  			/*
>> @@ -10429,6 +10446,7 @@ static struct platform_profile_handler dytc_profile = {
>>  static int tpacpi_dytc_profile_init(struct ibm_init_struct *iibm)
>>  {
>>  	int err, output;
>> +	int cmd;
>>  
>>  	/* Setup supported modes */
>>  	set_bit(PLATFORM_PROFILE_LOW_POWER, dytc_profile.choices);
>> @@ -10484,6 +10502,16 @@ static int tpacpi_dytc_profile_init(struct ibm_init_struct *iibm)
>>  		dbg_printk(TPACPI_DBG_INIT, "No DYTC support available\n");
>>  		return -ENODEV;
>>  	}
>> +	err = dytc_command(DYTC_CMD_UP_CAP, &output);
>> +	dytc_ultraperf_cap = output & BIT(DYTC_UP_SUPPORT_BIT) ? true : false;
>
> It would be better to put this BIT() into the define itself and remove 
> _BIT from the name because it doesn't really add that much information.
> Since you're assigning to bool, ? true : false construct is not required 
> but implicit cast will handle it for you. So in the end, this line would 
> be:
>
> 	dytc_ultraperf_cap = output & DYTC_UP_SUPPORT;

Agreed. I will make that change.
I'll wait and see if there is any more feedback and then do that with a v2 patch.

>
> Looking into the driver a bit more, there are a few other defines which 
> could also move BIT() from the code into defines. Please tell if you're 
> going to look at those because if not, I might try to make the patches.

Happy to look at doing that as I'm playing around with this driver anyway.

Thanks for the review!
Mark