Skip to contentSkip to conversation

hwmon: (k10temp): Use cpu_feature_enabled() for detecting zen

PATCH Applied Mario Limonciello 4 participants 6 messages Aug 20, 2024 – Aug 27, 2024

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 hwmon: (k10temp): Use cpu_feature_enabled() for detecting zen

From: Mario Limonciello <[email protected]>

This removes some boilerplate from the code and will allow adding future CPUs by just device IDs.

---
 drivers/hwmon/k10temp.c | 26 ++++++++++----------------
 1 file changed, 10 insertions(+), 16 deletions(-)
drivers/hwmon/k10temp.c +10 -16
index 543526bac0425..85a7632f3b50a 100644
@@ -438,16 +438,21 @@ static int k10temp_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 		data->disp_negative = true;
 	}
 
-	if (boot_cpu_data.x86 == 0x15 &&
+	data->is_zen = cpu_feature_enabled(X86_FEATURE_ZEN);
+	if (data->is_zen) {
+		data->temp_adjust_mask = ZEN_CUR_TEMP_RANGE_SEL_MASK;
+		data->read_tempreg = read_tempreg_nb_zen;
+	} else if (boot_cpu_data.x86 == 0x15 &&
 	    ((boot_cpu_data.x86_model & 0xf0) == 0x60 ||
 	     (boot_cpu_data.x86_model & 0xf0) == 0x70)) {
 		data->read_htcreg = read_htcreg_nb_f15;
 		data->read_tempreg = read_tempreg_nb_f15;
-	} else if (boot_cpu_data.x86 == 0x17 || boot_cpu_data.x86 == 0x18) {
-		data->temp_adjust_mask = ZEN_CUR_TEMP_RANGE_SEL_MASK;
-		data->read_tempreg = read_tempreg_nb_zen;
-		data->is_zen = true;
+	} else {
+		data->read_htcreg = read_htcreg_pci;
+		data->read_tempreg = read_tempreg_pci;
+	}
 
+	if (boot_cpu_data.x86 == 0x17 || boot_cpu_data.x86 == 0x18) {
 		switch (boot_cpu_data.x86_model) {
 		case 0x1:	/* Zen */
 		case 0x8:	/* Zen+ */
@@ -469,10 +474,6 @@ static int k10temp_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 			break;
 		}
 	} else if (boot_cpu_data.x86 == 0x19) {
-		data->temp_adjust_mask = ZEN_CUR_TEMP_RANGE_SEL_MASK;
-		data->read_tempreg = read_tempreg_nb_zen;
-		data->is_zen = true;
-
 		switch (boot_cpu_data.x86_model) {
 		case 0x0 ... 0x1:	/* Zen3 SP3/TR */
 		case 0x8:		/* Zen3 TR Chagall */
@@ -496,13 +497,6 @@ static int k10temp_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 			k10temp_get_ccd_support(data, 12);
 			break;
 		}
-	} else if (boot_cpu_data.x86 == 0x1a) {
-		data->temp_adjust_mask = ZEN_CUR_TEMP_RANGE_SEL_MASK;
-		data->read_tempreg = read_tempreg_nb_zen;
-		data->is_zen = true;
-	} else {
-		data->read_htcreg = read_htcreg_pci;
-		data->read_tempreg = read_tempreg_pci;
 	}
 
 	for (i = 0; i < ARRAY_SIZE(tctl_offset_table); i++) {
Signature · 1 line
2.43.0
#1 lore raw
Quoted 5 lines · Mario Limonciello

From: Mario Limonciello <[email protected]>

This removes some boilerplate from the code and will allow adding future CPUs by just device IDs.

I've been thinking that maybe we stop using PCI IDs entirely.

The PCIe devices that we match on are internal to the SoC. And they're not optional. They're basically processor components that are exposed through PCI config space for software convenience.

Thoughts?

Thanks,
Yazen

#2 lore raw
Quoted 9 lines · Yazen Ghannam
Quoted 5 lines · Mario Limonciello

From: Mario Limonciello <[email protected]>

This removes some boilerplate from the code and will allow adding future CPUs by just device IDs.

I've been thinking that maybe we stop using PCI IDs entirely.

Quoted 8 lines

The PCIe devices that we match on are internal to the SoC. And they're not optional. They're basically processor components that are exposed through PCI config space for software convenience.

Thoughts?

Thanks,
Yazen

Yeah I think that's a good idea. The one thing I want to make sure remains though is that k10temp automatically loads from a CPU modalias.

This is "tangential" to this patch except for the commit message reference to the PCI IDs.

maintainer
#3 lore raw
Quoted 16 lines · Mario Limonciello
Quoted 15 lines · Yazen Ghannam
Quoted 5 lines · Mario Limonciello

From: Mario Limonciello <[email protected]>

This removes some boilerplate from the code and will allow adding future CPUs by just device IDs.

I've been thinking that maybe we stop using PCI IDs entirely.

The PCIe devices that we match on are internal to the SoC. And they're not optional. They're basically processor components that are exposed through PCI config space for software convenience.

Thoughts?

Quoted 9 lines
Quoted 2 lines

Thanks,
Yazen

Yeah I think that's a good idea. The one thing I want to make sure remains though is that k10temp automatically loads from a CPU modalias.

This is "tangential" to this patch except for the commit message reference to the PCI IDs.

Based on the feedback I will not accept this patch but wait for a solution which is acceptable for everyone involved from AMD.

Guenter

#4 lore raw
Quoted 23 lines · Guenter Roeck
Quoted 22 lines · Mario Limonciello
Quoted 9 lines · Yazen Ghannam
Quoted 5 lines · Mario Limonciello

From: Mario Limonciello <[email protected]>

This removes some boilerplate from the code and will allow adding future CPUs by just device IDs.

I've been thinking that maybe we stop using PCI IDs entirely.

Quoted 8 lines

The PCIe devices that we match on are internal to the SoC. And they're not optional. They're basically processor components that are exposed through PCI config space for software convenience.

Thoughts?

Thanks,
Yazen

Yeah I think that's a good idea. The one thing I want to make sure remains though is that k10temp automatically loads from a CPU modalias.

Quoted 7 lines
Quoted 3 lines

This is "tangential" to this patch except for the commit message reference to the PCI IDs.

Based on the feedback I will not accept this patch but wait for a solution which is acceptable for everyone involved from AMD.

Hi Guenter,

Sorry, I didn't mean to imply that this is unacceptable. I think this is okay. I just wanted to highlight our general trend to move away from continuing to add PCI IDs. We'll still need to do so in the short term though.

Mario and I are working on some possible solutions. It'll be slow and iterative, since we don't want to break existing functionality.

Thanks,
Yazen