[PATCH v2] soc: qcom: llcc: Update the SCT table for IPQ5424

[PATCH v2] soc: qcom: llcc: Update the SCT table for IPQ5424

[Kathiravan Thirumoorthy]

Update the IPQ5424 LLCC slice configuration table to give the PPE
and Wi-Fi datapaths dedicated slices.

Assign slice IDs 1 and 5 to the PPE RX descriptor and RX fill use cases,
slice IDs 6 and 7 to the Wi-Fi radios connected over PCIe, and slice ID
2 to the CPU subsystem. Adjust the corresponding capacities, priorities,
and way masks for these consumers.

Having the dedicated slice configuration for each of these use cases helps
to improve the networking throughput by more than 1 Gbps.

Rename the usecase IDs for slice IDs 1 and 2 because LLCC_CPUSS and
LLCC_VIDSC0 do not describe the actual users of those slices.

Assisted-by: LLM
Signed-off-by: Kathiravan Thirumoorthy <kathiravan.thirumoorthy@oss.qualcomm.com>
---
Changes in v2:
- Updated the commit message decribing why the change is needed.
- Link to v1: https://patch.msgid.link/20260807-ipq5424_llcc_updates-v1-1-1f56362ff687@oss.qualcomm.com

To: Bjorn Andersson <andersson@kernel.org>
To: Konrad Dybcio <konradybcio@kernel.org>
Cc: linux-arm-msm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
---
 drivers/soc/qcom/llcc-qcom.c       | 56 ++++++++++++++++++++++++++++++++------
 include/linux/soc/qcom/llcc-qcom.h |  4 +++
 2 files changed, 52 insertions(+), 8 deletions(-)

diff --git a/drivers/soc/qcom/llcc-qcom.c b/drivers/soc/qcom/llcc-qcom.c
index 733999867bbf..d60431ace50f 100644
--- a/drivers/soc/qcom/llcc-qcom.c
+++ b/drivers/soc/qcom/llcc-qcom.c
@@ -626,11 +626,11 @@ static const struct llcc_slice_config glymur_data[] = {
 
 static const struct llcc_slice_config ipq5424_data[] =  {
 	{
-		.usecase_id = LLCC_CPUSS,
+		.usecase_id = LLCC_PPE_RXDESC,
 		.slice_id = 1,
-		.max_cap = 768,
-		.priority = 1,
-		.bonus_ways = 0xFFFF,
+		.max_cap = 128,
+		.priority = 3,
+		.bonus_ways = 0x0FFF,
 		.retain_on_pc = true,
 		.activate_on_init = true,
 		.write_scid_cacheable_en = true,
@@ -642,12 +642,52 @@ static const struct llcc_slice_config ipq5424_data[] =  {
 		.vict_prio = true,
 	},
 	{
-		.usecase_id = LLCC_VIDSC0,
+		.usecase_id = LLCC_CPUSS,
 		.slice_id = 2,
-		.max_cap = 256,
-		.priority = 2,
+		.max_cap = 768,
+		.priority = 1,
 		.fixed_size = true,
-		.bonus_ways = 0xF000,
+		.bonus_ways = 0x0FFF,
+		.retain_on_pc = true,
+		.activate_on_init = true,
+		.write_scid_cacheable_en = true,
+		.stale_en = true,
+		.stale_cap_en = true,
+	},
+	{
+		.usecase_id = LLCC_PPE_RXFILL,
+		.slice_id = 5,
+		.max_cap = 128,
+		.priority = 3,
+		.bonus_ways = 0x0FFF,
+		.retain_on_pc = true,
+		.activate_on_init = true,
+		.write_scid_cacheable_en = true,
+		.stale_en = true,
+		.stale_cap_en = true,
+		.alloc_oneway_en = true,
+		.ovcap_en = true,
+		.ovcap_prio = true,
+		.vict_prio = true,
+	},
+	{
+		.usecase_id = LLCC_WLAN_5G,
+		.slice_id = 6,
+		.max_cap = 128,
+		.priority = 3,
+		.bonus_ways = 0xC000,
+		.retain_on_pc = true,
+		.activate_on_init = true,
+		.write_scid_cacheable_en = true,
+		.stale_en = true,
+		.stale_cap_en = true,
+	},
+	{
+		.usecase_id = LLCC_WLAN_6G,
+		.slice_id = 7,
+		.max_cap = 128,
+		.priority = 3,
+		.bonus_ways = 0x3000,
 		.retain_on_pc = true,
 		.activate_on_init = true,
 		.write_scid_cacheable_en = true,
diff --git a/include/linux/soc/qcom/llcc-qcom.h b/include/linux/soc/qcom/llcc-qcom.h
index f3ed63e475ab..1ccc429401e8 100644
--- a/include/linux/soc/qcom/llcc-qcom.h
+++ b/include/linux/soc/qcom/llcc-qcom.h
@@ -86,6 +86,10 @@
 #define LLCC_CPUSS_HEU	 94
 #define LLCC_PCIE_TCU	 97
 #define LLCC_MDM_PNG_FIXED	 100
+#define LLCC_PPE_RXDESC		 101
+#define LLCC_PPE_RXFILL		 102
+#define LLCC_WLAN_5G		 103
+#define LLCC_WLAN_6G		 104
 
 /**
  * struct llcc_slice_desc - Cache slice descriptor

---
base-commit: ea2bff00da89d7767d677bb68470130ba96f4928
change-id: 20260806-ipq5424_llcc_updates-f9851d6db886

Best regards,
--  
Kathiravan Thirumoorthy <kathiravan.thirumoorthy@oss.qualcomm.com>

Re: [PATCH v2] soc: qcom: llcc: Update the SCT table for IPQ5424

[Dmitry Baryshkov]

On Tue, Aug 11, 2026 at 06:04:53PM +0530, Kathiravan Thirumoorthy wrote:
> Update the IPQ5424 LLCC slice configuration table to give the PPE
> and Wi-Fi datapaths dedicated slices.
> 
> Assign slice IDs 1 and 5 to the PPE RX descriptor and RX fill use cases,
> slice IDs 6 and 7 to the Wi-Fi radios connected over PCIe, and slice ID
> 2 to the CPU subsystem. Adjust the corresponding capacities, priorities,
> and way masks for these consumers.
> 
> Having the dedicated slice configuration for each of these use cases helps
> to improve the networking throughput by more than 1 Gbps.
> 
> Rename the usecase IDs for slice IDs 1 and 2 because LLCC_CPUSS and
> LLCC_VIDSC0 do not describe the actual users of those slices.
> 
> Assisted-by: LLM
> Signed-off-by: Kathiravan Thirumoorthy <kathiravan.thirumoorthy@oss.qualcomm.com>
> ---
> Changes in v2:
> - Updated the commit message decribing why the change is needed.
> - Link to v1: https://patch.msgid.link/20260807-ipq5424_llcc_updates-v1-1-1f56362ff687@oss.qualcomm.com
> 

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

-- 
With best wishes
Dmitry

Re: [PATCH v2] soc: qcom: llcc: Update the SCT table for IPQ5424

[Konrad Dybcio]

On 8/11/26 2:34 PM, Kathiravan Thirumoorthy wrote:
> Update the IPQ5424 LLCC slice configuration table to give the PPE
> and Wi-Fi datapaths dedicated slices.
> 
> Assign slice IDs 1 and 5 to the PPE RX descriptor and RX fill use cases,
> slice IDs 6 and 7 to the Wi-Fi radios connected over PCIe, and slice ID
> 2 to the CPU subsystem. Adjust the corresponding capacities, priorities,
> and way masks for these consumers.
> 
> Having the dedicated slice configuration for each of these use cases helps
> to improve the networking throughput by more than 1 Gbps.
> 
> Rename the usecase IDs for slice IDs 1 and 2 because LLCC_CPUSS and
> LLCC_VIDSC0 do not describe the actual users of those slices.

This reiterates the diff, but doesn't mention the actual reason for this
change (updated recommendations from the hardware team)

Konrad

Re: [PATCH v2] soc: qcom: llcc: Update the SCT table for IPQ5424

[Kathiravan Thirumoorthy]

On 8/19/2026 5:23 PM, Konrad Dybcio wrote:
> On 8/11/26 2:34 PM, Kathiravan Thirumoorthy wrote:
>> Update the IPQ5424 LLCC slice configuration table to give the PPE
>> and Wi-Fi datapaths dedicated slices.
>>
>> Assign slice IDs 1 and 5 to the PPE RX descriptor and RX fill use cases,
>> slice IDs 6 and 7 to the Wi-Fi radios connected over PCIe, and slice ID
>> 2 to the CPU subsystem. Adjust the corresponding capacities, priorities,
>> and way masks for these consumers.
>>
>> Having the dedicated slice configuration for each of these use cases helps
>> to improve the networking throughput by more than 1 Gbps.
>>
>> Rename the usecase IDs for slice IDs 1 and 2 because LLCC_CPUSS and
>> LLCC_VIDSC0 do not describe the actual users of those slices.
> This reiterates the diff, but doesn't mention the actual reason for this
> change (updated recommendations from the hardware team)

These slice assignments are based on recommendations from our 
performance team, following throughput measurements on IPQ5424 — not 
from a hardware team. Please let me know if I'm missing something here.

>
> Konrad

Re: [PATCH v2] soc: qcom: llcc: Update the SCT table for IPQ5424

[Konrad Dybcio]

On 8/19/26 4:54 PM, Kathiravan Thirumoorthy wrote:
> 
> On 8/19/2026 5:23 PM, Konrad Dybcio wrote:
>> On 8/11/26 2:34 PM, Kathiravan Thirumoorthy wrote:
>>> Update the IPQ5424 LLCC slice configuration table to give the PPE
>>> and Wi-Fi datapaths dedicated slices.
>>>
>>> Assign slice IDs 1 and 5 to the PPE RX descriptor and RX fill use cases,
>>> slice IDs 6 and 7 to the Wi-Fi radios connected over PCIe, and slice ID
>>> 2 to the CPU subsystem. Adjust the corresponding capacities, priorities,
>>> and way masks for these consumers.
>>>
>>> Having the dedicated slice configuration for each of these use cases helps
>>> to improve the networking throughput by more than 1 Gbps.
>>>
>>> Rename the usecase IDs for slice IDs 1 and 2 because LLCC_CPUSS and
>>> LLCC_VIDSC0 do not describe the actual users of those slices.
>> This reiterates the diff, but doesn't mention the actual reason for this
>> change (updated recommendations from the hardware team)
> 
> These slice assignments are based on recommendations from our performance team, following throughput measurements on IPQ5424 — not from a hardware team. Please let me know if I'm missing something here.

They can even be called the navy seals, just mention the reason for
the change!

FWIW the diff itself looks in line with the docs

Konrad