Re: [PATCH v2 11/11] platform/x86:intel/pmc: Get LPM information for Lunar Lake

From: Xi Pardee
Date: Mon Sep 09 2024 - 20:50:02 EST



On 8/29/2024 5:13 AM, Ilpo Järvinen wrote:
On Wed, 28 Aug 2024, Xi Pardee wrote:

Add support to find and read the requirements from the telemetry
entries for Lunar Lake platforms.

Signed-off-by: Xi Pardee <xi.pardee@xxxxxxxxxxxxxxx>
---
drivers/platform/x86/intel/pmc/lnl.c | 15 +++++++++++++++
1 file changed, 15 insertions(+)

diff --git a/drivers/platform/x86/intel/pmc/lnl.c b/drivers/platform/x86/intel/pmc/lnl.c
index 109b08d43fc8..f5fee9e105e2 100644
--- a/drivers/platform/x86/intel/pmc/lnl.c
+++ b/drivers/platform/x86/intel/pmc/lnl.c
@@ -13,8 +13,13 @@
#include "core.h"
+#define SOCM_LPM_REQ_GUID 0x15099748
+
+static const u8 LNL_LPM_REG_INDEX[] = {0, 4, 5, 6, 8, 9, 10, 11, 12, 13, 14, 15, 16, 20};
+
static struct pmc_info lnl_pmc_info_list[] = {
{
+ .guid = SOCM_LPM_REQ_GUID,
.devid = PMC_DEVID_LNL_SOCM,
.map = &lnl_socm_reg_map,
},
@@ -536,6 +541,7 @@ const struct pmc_reg_map lnl_socm_reg_map = {
.lpm_live_status_offset = MTL_LPM_LIVE_STATUS_OFFSET,
.s0ix_blocker_maps = lnl_blk_maps,
.s0ix_blocker_offset = LNL_S0IX_BLOCKER_OFFSET,
+ .lpm_reg_index = LNL_LPM_REG_INDEX,
};
#define LNL_NPU_PCI_DEV 0x643e
@@ -561,6 +567,8 @@ static int lnl_resume(struct pmc_dev *pmcdev)
int lnl_core_init(struct pmc_dev *pmcdev)
{
+ bool ssram_init = true;
+ int func = 2;
int ret;
struct pmc *pmc = pmcdev->pmcs[PMC_IDX_SOC];
@@ -578,6 +586,7 @@ int lnl_core_init(struct pmc_dev *pmcdev)
/* If regbase not assigned, set map and discover using legacy method */
if (ret) {
+ ssram_init = false;
pmc->map = &lnl_socm_reg_map;
ret = get_primary_reg_base(pmc);
if (ret)
@@ -586,5 +595,11 @@ int lnl_core_init(struct pmc_dev *pmcdev)
pmc_core_get_low_power_modes(pmcdev);
+ if (ssram_init) {
+ ret = pmc_core_ssram_get_lpm_reqs(pmcdev, func);
+ if (ret)
+ return ret;
There's quite much duplication related to this legacy/ssram init in
the per arch core init functions. And some inconsistencies too which
seem incidental such as mtl.c using return directly here.
I will try to extract the duplicate code to a common function and remove the
inconsistencies between different architecture in next version.

Xi