Skip to contentSkip to conversation

platform/x86/amd: pmc: Require at least 2.5 seconds between HW sleep cycles

PATCH In review Mario Limonciello 3 participants 3 messages Apr 14, 2025 – Apr 17, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 platform/x86/amd: pmc: Require at least 2.5 seconds between HW sleep cycles

From: Mario Limonciello <[email protected]>

When an APU exits HW sleep with no active wake sources the Linux kernel will rapidly assert that the APU can enter back into HW sleep. This happens in a few ms. Contrasting this to Windows, Windows can take 10s of seconds to enter back into the resiliency phase for Modern Standby.

For some situations this can be problematic because it can cause leakage from VDDCR_SOC to VDD_MISC and force VDD_MISC outside of the electrical design guide specifications. On some designs this will trip the over voltage protection feature (OVP) of the voltage regulator module, but it could cause APU damage as well.

To prevent this risk, add an explicit sleep call so that future attempts to enter into HW sleep will have enough time to settle. This will occur while the screen is dark and only on cases that the APU should enter HW sleep again, so it shouldn't be noticeable to any user.

---
 drivers/platform/x86/amd/pmc/pmc.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
drivers/platform/x86/amd/pmc/pmc.c +3 -4
index d789d6cab7948..0329fafe14ebc 100644
@@ -644,10 +644,9 @@ static void amd_pmc_s2idle_check(void)
 	struct smu_metrics table;
 	int rc;
 
-	/* CZN: Ensure that future s0i3 entry attempts at least 10ms passed */
-	if (pdev->cpu_id == AMD_CPU_ID_CZN && !get_metrics_table(pdev, &table) &&
-	    table.s0i3_last_entry_status)
-		usleep_range(10000, 20000);
+	/* Avoid triggering OVP */
+	if (!get_metrics_table(pdev, &table) && table.s0i3_last_entry_status)
+		msleep(2500);
 
 	/* Dump the IdleMask before we add to the STB */
 	amd_pmc_idlemask_read(pdev, pdev->dev, NULL);
Signature · 1 line
2.43.0
maintainer Acked-by on v1 1/1
#1 lore raw
Quoted 13 lines · Mario Limonciello

From: Mario Limonciello <[email protected]>

When an APU exits HW sleep with no active wake sources the Linux kernel will rapidly assert that the APU can enter back into HW sleep. This happens in a few ms. Contrasting this to Windows, Windows can take 10s of seconds to enter back into the resiliency phase for Modern Standby.

For some situations this can be problematic because it can cause leakage from VDDCR_SOC to VDD_MISC and force VDD_MISC outside of the electrical design guide specifications. On some designs this will trip the over voltage protection feature (OVP) of the voltage regulator module, but it could cause APU damage as well.

Quoted 7 lines

To prevent this risk, add an explicit sleep call so that future attempts to enter into HW sleep will have enough time to settle. This will occur while the screen is dark and only on cases that the APU should enter HW sleep again, so it shouldn't be noticeable to any user.

Thank you.

Quoted 22 lines
---
 drivers/platform/x86/amd/pmc/pmc.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
drivers/platform/x86/amd/pmc/pmc.c +3 -4
index d789d6cab7948..0329fafe14ebc 100644
@@ -644,10 +644,9 @@ static void amd_pmc_s2idle_check(void)
 	struct smu_metrics table;
 	int rc;
 
-	/* CZN: Ensure that future s0i3 entry attempts at least 10ms passed */
-	if (pdev->cpu_id == AMD_CPU_ID_CZN && !get_metrics_table(pdev, &table) &&
-	    table.s0i3_last_entry_status)
-		usleep_range(10000, 20000);
+	/* Avoid triggering OVP */
+	if (!get_metrics_table(pdev, &table) && table.s0i3_last_entry_status)
+		msleep(2500);
 
 	/* Dump the IdleMask before we add to the STB */
 	amd_pmc_idlemask_read(pdev, pdev->dev, NULL);
maintainer
#2 lore raw
In reply to Mario Limonciello
Quoted 5 lines · Mario Limonciello

When an APU exits HW sleep with no active wake sources the Linux kernel will rapidly assert that the APU can enter back into HW sleep. This happens in a few ms. Contrasting this to Windows, Windows can take 10s of seconds to enter back into the resiliency phase for Modern Standby.

Quoted 7 lines

For some situations this can be problematic because it can cause leakage from VDDCR_SOC to VDD_MISC and force VDD_MISC outside of the electrical design guide specifications. On some designs this will trip the over voltage protection feature (OVP) of the voltage regulator module, but it could cause APU damage as well.

[...]

Thank you for your contribution, it has been applied to my local review-ilpo-fixes branch. Note it will show up in the public platform-drivers-x86/review-ilpo-fixes branch only once I've pushed my local branch there, which might take a while.

The list of commits applied:
[1/1] platform/x86/amd: pmc: Require at least 2.5 seconds between HW sleep cycles

      commit: 9f5595d5f03fd4dc640607a71e89a1daa68fd19d

--
i.