Re: [PATCH v4 3/3] platform/x86/amd/pmf: Use ring buffer to store custom BIOS input values

[Shyam Sundar S K <Shyam-sundar.S-k@amd.com>]

On 11/27/2025 17:39, Ilpo Järvinen wrote:
> On Wed, 19 Nov 2025, Shyam Sundar S K wrote:
> 
>> Custom BIOS input values can be updated by multiple sources, such as power
>> mode changes and sensor events, each triggering a custom BIOS input event.
>> When these events occur in rapid succession, new data may overwrite
>> previous values before they are processed, resulting in lost updates.
>>
>> To address this, introduce a fixed-size, power-of-two ring buffer to
>> capture every custom BIOS input event, storing both the pending request
>> and its associated input values. Access to the ring buffer is synchronized
>> using a mutex.
>>
>> The previous use of memset() to clear the pending request structure after
>> each event is removed, as each BIOS input value is now copied into the
>> buffer as a snapshot. Consumers now process entries directly from the ring
>> buffer, making explicit clearing of the pending request structure
>> unnecessary.
>>
>> Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
>> Tested-by: Yijun Shen <Yijun.Shen@Dell.com>
>> Co-developed-by: Patil Rajesh Reddy <Patil.Reddy@amd.com>
>> Signed-off-by: Patil Rajesh Reddy <Patil.Reddy@amd.com>
>> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
>> ---
>> v4:
>>  - Do not store local copy of the ring buffer
>>  - use devm_mutex_init()
>>
>> v3:
>>  - include headers wherever missing
>>  - use dev_warn() instead of dev_WARN_ONCE()
>>  - remove generic struct names
>>  - enhance ringbuffer mechanism to handle common path
>>  - other cosmetic remarks
>>
>> v2:
>>  - Add dev_WARN_ONCE()
>>  - Change variable name rb_mutex to cbi_mutex
>>  - Move tail increment logic above pending request check
>>
>>  drivers/platform/x86/amd/pmf/acpi.c   | 40 +++++++++++++++++++++++++++
>>  drivers/platform/x86/amd/pmf/core.c   |  5 ++++
>>  drivers/platform/x86/amd/pmf/pmf.h    | 21 ++++++++++++++
>>  drivers/platform/x86/amd/pmf/spc.c    | 32 +++++++++++----------
>>  drivers/platform/x86/amd/pmf/tee-if.c |  2 ++
>>  5 files changed, 86 insertions(+), 14 deletions(-)
>>
>> diff --git a/drivers/platform/x86/amd/pmf/acpi.c b/drivers/platform/x86/amd/pmf/acpi.c
>> index 13c4fec2c7ef..3d94b03cf794 100644
>> --- a/drivers/platform/x86/amd/pmf/acpi.c
>> +++ b/drivers/platform/x86/amd/pmf/acpi.c
>> @@ -9,6 +9,9 @@
>>   */
>>  
>>  #include <linux/acpi.h>
>> +#include <linux/array_size.h>
>> +#include <linux/cleanup.h>
>> +#include <linux/dev_printk.h>
>>  #include "pmf.h"
>>  
>>  #define APMF_CQL_NOTIFICATION  2
>> @@ -331,6 +334,39 @@ int apmf_get_sbios_requests(struct amd_pmf_dev *pdev, struct apmf_sbios_req *req
>>  									 req, sizeof(*req));
>>  }
>>  
>> +/* Store custom BIOS inputs data in ring buffer */
>> +static void amd_pmf_custom_bios_inputs_rb(struct amd_pmf_dev *pmf_dev)
>> +{
>> +	struct pmf_cbi_ring_buffer *rb = &pmf_dev->cbi_buf;
>> +	int i;
>> +
>> +	guard(mutex)(&pmf_dev->cbi_mutex);
>> +
>> +	switch (pmf_dev->cpu_id) {
>> +	case AMD_CPU_ID_PS:
>> +		for (i = 0; i < ARRAY_SIZE(custom_bios_inputs_v1); i++)
>> +			rb->data[rb->head].val[i] = pmf_dev->req1.custom_policy[i];
>> +		rb->data[rb->head].preq = pmf_dev->req1.pending_req;
>> +		break;
>> +	case PCI_DEVICE_ID_AMD_1AH_M20H_ROOT:
>> +	case PCI_DEVICE_ID_AMD_1AH_M60H_ROOT:
>> +		for (i = 0; i < ARRAY_SIZE(custom_bios_inputs); i++)
>> +			rb->data[rb->head].val[i] = pmf_dev->req.custom_policy[i];
>> +		rb->data[rb->head].preq = pmf_dev->req.pending_req;
>> +		break;
>> +	default:
>> +		return;
>> +	}
>> +
>> +	if (CIRC_SPACE(rb->head, rb->tail, CUSTOM_BIOS_INPUT_RING_ENTRIES) == 0) {
>> +		/* Rare case: ensures the newest BIOS input value is kept */
>> +		dev_warn(pmf_dev->dev, "Overwriting BIOS input value, data may be lost\n");
>> +		rb->tail = (rb->tail + 1) & (CUSTOM_BIOS_INPUT_RING_ENTRIES - 1);
>> +	}
>> +
>> +	rb->head = (rb->head + 1) & (CUSTOM_BIOS_INPUT_RING_ENTRIES - 1);
>> +}
>> +
>>  static void amd_pmf_handle_early_preq(struct amd_pmf_dev *pdev)
>>  {
>>  	if (!pdev->cb_flag)
>> @@ -356,6 +392,8 @@ static void apmf_event_handler_v2(acpi_handle handle, u32 event, void *data)
>>  	dev_dbg(pmf_dev->dev, "Pending request (preq): 0x%x\n", pmf_dev->req.pending_req);
>>  
>>  	amd_pmf_handle_early_preq(pmf_dev);
>> +
>> +	amd_pmf_custom_bios_inputs_rb(pmf_dev);
>>  }
>>  
>>  static void apmf_event_handler_v1(acpi_handle handle, u32 event, void *data)
>> @@ -374,6 +412,8 @@ static void apmf_event_handler_v1(acpi_handle handle, u32 event, void *data)
>>  	dev_dbg(pmf_dev->dev, "Pending request (preq1): 0x%x\n", pmf_dev->req1.pending_req);
>>  
>>  	amd_pmf_handle_early_preq(pmf_dev);
>> +
>> +	amd_pmf_custom_bios_inputs_rb(pmf_dev);
>>  }
>>  
>>  static void apmf_event_handler(acpi_handle handle, u32 event, void *data)
>> diff --git a/drivers/platform/x86/amd/pmf/core.c b/drivers/platform/x86/amd/pmf/core.c
>> index 2ec4cb92e34f..71421a5d7afd 100644
>> --- a/drivers/platform/x86/amd/pmf/core.c
>> +++ b/drivers/platform/x86/amd/pmf/core.c
>> @@ -11,6 +11,7 @@
>>  #include <linux/debugfs.h>
>>  #include <linux/iopoll.h>
>>  #include <linux/module.h>
>> +#include <linux/mutex.h>
>>  #include <linux/pci.h>
>>  #include <linux/platform_device.h>
>>  #include <linux/power_supply.h>
>> @@ -477,6 +478,10 @@ static int amd_pmf_probe(struct platform_device *pdev)
>>  	if (err)
>>  		return err;
>>  
>> +	err = devm_mutex_init(dev->dev, &dev->cbi_mutex);
>> +	if (err)
>> +		return err;
>> +
>>  	apmf_acpi_init(dev);
>>  	platform_set_drvdata(pdev, dev);
>>  	amd_pmf_dbgfs_register(dev);
>> diff --git a/drivers/platform/x86/amd/pmf/pmf.h b/drivers/platform/x86/amd/pmf/pmf.h
>> index 2145df4128cd..5a18b3604b6e 100644
>> --- a/drivers/platform/x86/amd/pmf/pmf.h
>> +++ b/drivers/platform/x86/amd/pmf/pmf.h
>> @@ -12,7 +12,9 @@
>>  #define PMF_H
>>  
>>  #include <linux/acpi.h>
>> +#include <linux/circ_buf.h>
>>  #include <linux/input.h>
>> +#include <linux/mutex_types.h>
>>  #include <linux/platform_device.h>
>>  #include <linux/platform_profile.h>
>>  
>> @@ -120,6 +122,7 @@ struct cookie_header {
>>  #define APTS_MAX_STATES		16
>>  #define CUSTOM_BIOS_INPUT_BITS	GENMASK(16, 7)
>>  #define BIOS_INPUTS_MAX		10
>> +#define CUSTOM_BIOS_INPUT_RING_ENTRIES	64	/* Must be power of two for CIRC_* macros */
>>  
>>  typedef void (*apmf_event_handler_t)(acpi_handle handle, u32 event, void *data);
>>  
>> @@ -359,6 +362,22 @@ struct pmf_bios_inputs_prev {
>>  	u32 custom_bios_inputs[BIOS_INPUTS_MAX];
>>  };
>>  
>> +/**
>> + * struct pmf_bios_input_entry - Snapshot of custom BIOS input event
>> + * @val: Array of custom BIOS input values
>> + * @preq: Pending request value associated with this event
>> + */
>> +struct pmf_bios_input_entry {
>> +	u32 val[BIOS_INPUTS_MAX];
>> +	u32 preq;
>> +};
>> +
>> +struct pmf_cbi_ring_buffer {
>> +	struct pmf_bios_input_entry data[CUSTOM_BIOS_INPUT_RING_ENTRIES];
>> +	int head;
>> +	int tail;
>> +};
>> +
>>  struct amd_pmf_dev {
>>  	void __iomem *regbase;
>>  	void __iomem *smu_virt_addr;
>> @@ -407,6 +426,8 @@ struct amd_pmf_dev {
>>  	struct apmf_sbios_req_v1 req1;
>>  	struct pmf_bios_inputs_prev cb_prev; /* To preserve custom BIOS inputs */
>>  	bool cb_flag;			     /* To handle first custom BIOS input */
>> +	struct pmf_cbi_ring_buffer cbi_buf;
>> +	struct mutex cbi_mutex;		     /* Protects ring buffer access */
>>  };
>>  
>>  struct apmf_sps_prop_granular_v2 {
>> diff --git a/drivers/platform/x86/amd/pmf/spc.c b/drivers/platform/x86/amd/pmf/spc.c
>> index 85192c7536b8..34fff41b86fe 100644
>> --- a/drivers/platform/x86/amd/pmf/spc.c
>> +++ b/drivers/platform/x86/amd/pmf/spc.c
>> @@ -11,6 +11,7 @@
>>  
>>  #include <acpi/button.h>
>>  #include <linux/amd-pmf-io.h>
>> +#include <linux/cleanup.h>
>>  #include <linux/power_supply.h>
>>  #include <linux/units.h>
>>  #include "pmf.h"
>> @@ -132,30 +133,37 @@ static void amd_pmf_set_ta_custom_bios_input(struct ta_pmf_enact_table *in, int
>>  	}
>>  }
>>  
>> -static void amd_pmf_update_bios_inputs(struct amd_pmf_dev *pdev, u32 pending_req,
>> +static void amd_pmf_update_bios_inputs(struct amd_pmf_dev *pdev, struct pmf_bios_input_entry *data,
>>  				       const struct amd_pmf_pb_bitmap *inputs,
>> -				       const u32 *custom_policy, struct ta_pmf_enact_table *in)
>> +				       struct ta_pmf_enact_table *in)
>>  {
>>  	unsigned int i;
>>  
>>  	for (i = 0; i < ARRAY_SIZE(custom_bios_inputs); i++) {
>> -		if (!(pending_req & inputs[i].bit_mask))
>> +		if (!(data->preq & inputs[i].bit_mask))
>>  			continue;
>> -		amd_pmf_set_ta_custom_bios_input(in, i, custom_policy[i]);
>> -		pdev->cb_prev.custom_bios_inputs[i] = custom_policy[i];
>> -		dev_dbg(pdev->dev, "Custom BIOS Input[%d]: %u\n", i, custom_policy[i]);
>> +		amd_pmf_set_ta_custom_bios_input(in, i, data->val[i]);
>> +		pdev->cb_prev.custom_bios_inputs[i] = data->val[i];
>> +		dev_dbg(pdev->dev, "Custom BIOS Input[%d]: %u\n", i, data->val[i]);
>>  	}
>>  }
>>  
>>  static void amd_pmf_get_custom_bios_inputs(struct amd_pmf_dev *pdev,
>>  					   struct ta_pmf_enact_table *in)
>>  {
>> +	struct pmf_cbi_ring_buffer *rb = &pdev->cbi_buf;
>>  	unsigned int i;
>>  
>> +	guard(mutex)(&pdev->cbi_mutex);
>> +
>>  	for (i = 0; i < ARRAY_SIZE(custom_bios_inputs); i++)
>>  		amd_pmf_set_ta_custom_bios_input(in, i, pdev->cb_prev.custom_bios_inputs[i]);
>>  
>> -	if (!(pdev->req.pending_req || pdev->req1.pending_req))
>> +	if (CIRC_CNT(rb->head, rb->tail, CUSTOM_BIOS_INPUT_RING_ENTRIES) == 0)
>> +		return;	/* return if ring buffer is empty */
>> +
>> +	/* If no active custom BIOS input pending request, do not consume further work */
>> +	if (!rb->data[rb->tail].preq)
>>  		return;
> 
> I'm left usure if "do not consume further work" comment really means that 
> the entry is supposed to not get removed from the ring, which stalls the 
> ring forever?

No. The amd_pmf_populate_ta_inputs() function runs periodically and
calls amd_pmf_get_custom_bios_inputs(). The preq check prevents
unnecessary processing when the ring buffer is empty or contains no
valid pending requests.

Regarding the ring stall:
Entries are added to the ring buffer only when a valid ACPI event
triggers the event handler. Once an entry with preq set is queued, the
consumer processes it and increments the tail.

The check acts as a guard when the periodic function runs but no new
events are available - either the ring buffer is empty or the current
entry has no active pending request.

> 
> If that's the wanted behavior, does that imply overwrite dev_warn() above 
> can spam the logs from that point on as the ring can fill up without 
> anything consuming entries from it?

Entries with valid preq are always consumed, preventing ring stalls.
The dev_warn() only triggers when the buffer overflows, not from
stalled consumption.

Thanks,
Shyam

> 
> I've taken first two patches into review-ilpo-next.
> 
>>  	if (!pdev->smart_pc_enabled)
>> @@ -165,20 +173,16 @@ static void amd_pmf_get_custom_bios_inputs(struct amd_pmf_dev *pdev,
>>  	case PMF_IF_V1:
>>  		if (!is_apmf_bios_input_notifications_supported(pdev))
>>  			return;
>> -		amd_pmf_update_bios_inputs(pdev, pdev->req1.pending_req, custom_bios_inputs_v1,
>> -					   pdev->req1.custom_policy, in);
>> +		amd_pmf_update_bios_inputs(pdev, &rb->data[rb->tail], custom_bios_inputs_v1, in);
>>  		break;
>>  	case PMF_IF_V2:
>> -		amd_pmf_update_bios_inputs(pdev, pdev->req.pending_req, custom_bios_inputs,
>> -					   pdev->req.custom_policy, in);
>> +		amd_pmf_update_bios_inputs(pdev, &rb->data[rb->tail], custom_bios_inputs, in);
>>  		break;
>>  	default:
>>  		break;
>>  	}
>>  
>> -	/* Clear pending requests after handling */
>> -	memset(&pdev->req, 0, sizeof(pdev->req));
>> -	memset(&pdev->req1, 0, sizeof(pdev->req1));
>> +	rb->tail = (rb->tail + 1) & (CUSTOM_BIOS_INPUT_RING_ENTRIES - 1);
>>  }
>>  
>>  static void amd_pmf_get_c0_residency(u16 *core_res, size_t size, struct ta_pmf_enact_table *in)
>> diff --git a/drivers/platform/x86/amd/pmf/tee-if.c b/drivers/platform/x86/amd/pmf/tee-if.c
>> index 6e8116bef4f6..add742e33e1e 100644
>> --- a/drivers/platform/x86/amd/pmf/tee-if.c
>> +++ b/drivers/platform/x86/amd/pmf/tee-if.c
>> @@ -579,6 +579,8 @@ int amd_pmf_init_smart_pc(struct amd_pmf_dev *dev)
>>  		status = ret == TA_PMF_TYPE_SUCCESS;
>>  		if (status) {
>>  			dev->cb_flag = true;
>> +			dev->cbi_buf.head = 0;
>> +			dev->cbi_buf.tail = 0;
>>  			break;
>>  		}
>>  		amd_pmf_tee_deinit(dev);
>>
>