[PATCH 2/4] drm/amd/powerplay: add register thermal interrupt in hwmgr_hw_init

[JinHuiEric.Huang@xxxxxxx (Eric Huang)]

Signed-off-by: Eric Huang <JinHuiEric.Huang at amd.com>
---
 drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c | 75 ++++++++++++++++++++++++++++-
 drivers/gpu/drm/amd/powerplay/inc/hwmgr.h   |  6 +++
 2 files changed, 80 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c
index d1960c1..ae60f74 100644
--- a/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c
+++ b/drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c
@@ -26,8 +26,8 @@
 #include <linux/types.h>
 #include <linux/kernel.h>
 #include <linux/slab.h>
+#include <linux/pci.h>
 #include <drm/amdgpu_drm.h>
-#include "cgs_common.h"
 #include "power_state.h"
 #include "hwmgr.h"
 #include "pppcielanes.h"
@@ -50,6 +50,75 @@ uint8_t convert_to_vid(uint16_t vddc)
 	return (uint8_t) ((6200 - (vddc * VOLTAGE_SCALE)) / 25);
 }
 
+static int phm_get_pci_bus_devfn(struct pp_hwmgr *hwmgr,
+		struct cgs_system_info *sys_info)
+{
+	sys_info->size = sizeof(struct cgs_system_info);
+	sys_info->info_id = CGS_SYSTEM_INFO_PCIE_BUS_DEVFN;
+
+	return cgs_query_system_info(hwmgr->device, sys_info);
+}
+
+static int phm_thermal_l2h_irq(void *private_data,
+		 unsigned src_id, const uint32_t *iv_entry)
+{
+	struct pp_hwmgr *hwmgr = (struct pp_hwmgr *)private_data;
+	struct cgs_system_info sys_info = {0};
+	int result;
+
+	result = phm_get_pci_bus_devfn(hwmgr, &sys_info);
+	if (result)
+		return -EINVAL;
+
+	pr_warn("GPU over temperature range detected on PCIe %lld:%lld.%lld!\n",
+			PCI_BUS_NUM(sys_info.value),
+			PCI_SLOT(sys_info.value),
+			PCI_FUNC(sys_info.value));
+	return 0;
+}
+
+static int phm_thermal_h2l_irq(void *private_data,
+		 unsigned src_id, const uint32_t *iv_entry)
+{
+	struct pp_hwmgr *hwmgr = (struct pp_hwmgr *)private_data;
+	struct cgs_system_info sys_info = {0};
+	int result;
+
+	result = phm_get_pci_bus_devfn(hwmgr, &sys_info);
+	if (result)
+		return -EINVAL;
+
+	pr_warn("GPU under temperature range detected on PCIe %lld:%lld.%lld!\n",
+			PCI_BUS_NUM(sys_info.value),
+			PCI_SLOT(sys_info.value),
+			PCI_FUNC(sys_info.value));
+	return 0;
+}
+
+static int phm_ctf_irq(void *private_data,
+		 unsigned src_id, const uint32_t *iv_entry)
+{
+	struct pp_hwmgr *hwmgr = (struct pp_hwmgr *)private_data;
+	struct cgs_system_info sys_info = {0};
+	int result;
+
+	result = phm_get_pci_bus_devfn(hwmgr, &sys_info);
+	if (result)
+		return -EINVAL;
+
+	pr_warn("GPU Critical Temperature Fault detected on PCIe %lld:%lld.%lld!\n",
+			PCI_BUS_NUM(sys_info.value),
+			PCI_SLOT(sys_info.value),
+			PCI_FUNC(sys_info.value));
+	return 0;
+}
+
+static const struct cgs_irq_src_funcs thermal_irq_src[3] = {
+	{NULL, phm_thermal_l2h_irq},
+	{NULL, phm_thermal_h2l_irq},
+	{NULL, phm_ctf_irq}
+};
+
 int hwmgr_early_init(struct pp_instance *handle)
 {
 	struct pp_hwmgr *hwmgr;
@@ -170,6 +239,10 @@ int hwmgr_hw_init(struct pp_instance *handle)
 	if (ret)
 		goto err2;
 
+	ret = phm_register_thermal_interrupt(hwmgr, &thermal_irq_src);
+	if (ret)
+		goto err2;
+
 	return 0;
 err2:
 	if (hwmgr->hwmgr_func->backend_fini)
diff --git a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h
index 3bbe7d5..7f6d1cb 100644
--- a/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h
+++ b/drivers/gpu/drm/amd/powerplay/inc/hwmgr.h
@@ -32,6 +32,7 @@
 #include "ppatomctrl.h"
 #include "hwmgr_ppt.h"
 #include "power_state.h"
+#include "cgs_linux.h"
 
 struct pp_instance;
 struct pp_hwmgr;
@@ -796,6 +797,11 @@ struct pp_hwmgr {
 	bool en_umd_pstate;
 };
 
+struct cgs_irq_src_funcs {
+	cgs_irq_source_set_func_t set;
+	cgs_irq_handler_func_t handler;
+};
+
 extern int hwmgr_early_init(struct pp_instance *handle);
 extern int hwmgr_hw_init(struct pp_instance *handle);
 extern int hwmgr_hw_fini(struct pp_instance *handle);
-- 
2.7.4