[PATCH 91/95] drm/amdgpu: Expose ualink info under each xcp

[Alex Deucher <alexander.deucher@amd.com>]

From: Lijo Lazar <lijo.lazar@amd.com>

Mirror the read-only ualink info attributes on each secondary compute
partition as a per-partition ualink node, so a partition-scoped consumer
(e.g. a container that only sees its partition's device node) can read
the ualink identity and state. Partition 0 shares the primary device,
which already exposes that node, so it is skipped. An inactive partition
device won't be having any attributes listed under ualink node.

The per-partition attributes are served by thin wrappers that delegate
to the existing device-level info show functions. A reference on the
info kobject is held for the node's lifetime so it cannot be freed
while a partition still uses it.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Assisted-by: Claude (claude-opus-4.7)
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c | 165 +++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.h |   3 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c    |   6 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h    |   7 +
 4 files changed, 181 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c
index 0baa8c07077d5..765105aea09e2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c
@@ -746,6 +746,37 @@ UALINK_ENUM_SHOW(info, addr_mode, vpod.addr_mode);
 UALINK_ENUM_SHOW(info, accel_state, accel_state);
 UALINK_IDARRAY_SHOW(info, local_accels, local_accels, n_local_accels);
 
+static struct amdgpu_device *ualink_xcp_kobj_to_adev(struct kobject *kobj)
+{
+	struct amdgpu_xcp *xcp = container_of(kobj, struct amdgpu_xcp,
+					      ualink.kobj);
+
+	return xcp->xcp_mgr->adev;
+}
+
+#define UALINK_XCP_INFO_SHOW(name)					\
+static ssize_t ualink_xcp_info_##name##_show(struct kobject *kobj,	\
+		struct kobj_attribute *attr, char *buf)			\
+{									\
+	struct amdgpu_device *adev = ualink_xcp_kobj_to_adev(kobj);	\
+									\
+	return ualink_info_##name##_show(&adev->ualink.info->kobj,	\
+					 attr, buf);			\
+}
+
+UALINK_XCP_INFO_SHOW(link_type)
+UALINK_XCP_INFO_SHOW(accel_id)
+UALINK_XCP_INFO_SHOW(bandwidth)
+UALINK_XCP_INFO_SHOW(latency)
+UALINK_XCP_INFO_SHOW(ppod_id)
+UALINK_XCP_INFO_SHOW(ppod_size)
+UALINK_XCP_INFO_SHOW(vpod_id)
+UALINK_XCP_INFO_SHOW(vpod_size)
+UALINK_XCP_INFO_SHOW(vpod_active_accels)
+UALINK_XCP_INFO_SHOW(addr_mode)
+UALINK_XCP_INFO_SHOW(accel_state)
+UALINK_XCP_INFO_SHOW(local_accels)
+
 #define UALINK_INFO_ATTR(name) __ATTR(name, 0444, ualink_info_##name##_show, NULL)
 static struct kobj_attribute ualink_info_link_type = UALINK_INFO_ATTR(link_type);
 static struct kobj_attribute ualink_info_accel_id  = UALINK_INFO_ATTR(accel_id);
@@ -776,6 +807,58 @@ static const struct attribute *ualink_info_attrs[] = {
 	NULL
 };
 
+#define UALINK_XCP_INFO_ATTR(name) \
+	__ATTR(name, 0444, ualink_xcp_info_##name##_show, NULL)
+static struct kobj_attribute ualink_xcp_info_link_type = UALINK_XCP_INFO_ATTR(link_type);
+static struct kobj_attribute ualink_xcp_info_accel_id  = UALINK_XCP_INFO_ATTR(accel_id);
+static struct kobj_attribute ualink_xcp_info_bandwidth = UALINK_XCP_INFO_ATTR(bandwidth);
+static struct kobj_attribute ualink_xcp_info_latency   = UALINK_XCP_INFO_ATTR(latency);
+static struct kobj_attribute ualink_xcp_info_ppod_id   = UALINK_XCP_INFO_ATTR(ppod_id);
+static struct kobj_attribute ualink_xcp_info_ppod_size = UALINK_XCP_INFO_ATTR(ppod_size);
+static struct kobj_attribute ualink_xcp_info_vpod_id   = UALINK_XCP_INFO_ATTR(vpod_id);
+static struct kobj_attribute ualink_xcp_info_vpod_size = UALINK_XCP_INFO_ATTR(vpod_size);
+static struct kobj_attribute ualink_xcp_info_vpod_active_accels = UALINK_XCP_INFO_ATTR(vpod_active_accels);
+static struct kobj_attribute ualink_xcp_info_addr_mode = UALINK_XCP_INFO_ATTR(addr_mode);
+static struct kobj_attribute ualink_xcp_info_accel_state = UALINK_XCP_INFO_ATTR(accel_state);
+static struct kobj_attribute ualink_xcp_info_local_accels = UALINK_XCP_INFO_ATTR(local_accels);
+
+static struct attribute *ualink_xcp_info_attrs[] = {
+	&ualink_xcp_info_link_type.attr,
+	&ualink_xcp_info_accel_id.attr,
+	&ualink_xcp_info_bandwidth.attr,
+	&ualink_xcp_info_latency.attr,
+	&ualink_xcp_info_ppod_id.attr,
+	&ualink_xcp_info_ppod_size.attr,
+	&ualink_xcp_info_vpod_id.attr,
+	&ualink_xcp_info_vpod_size.attr,
+	&ualink_xcp_info_vpod_active_accels.attr,
+	&ualink_xcp_info_addr_mode.attr,
+	&ualink_xcp_info_accel_state.attr,
+	&ualink_xcp_info_local_accels.attr,
+	NULL
+};
+
+static umode_t ualink_xcp_info_is_visible(struct kobject *kobj,
+					  struct attribute *attr, int n)
+{
+	struct amdgpu_xcp *xcp = container_of(kobj, struct amdgpu_xcp,
+					      ualink.kobj);
+
+	if (!xcp->valid)
+		return 0;
+
+	return attr->mode;
+}
+
+static const struct attribute_group ualink_xcp_info_group = {
+	.attrs = ualink_xcp_info_attrs,
+	.is_visible = ualink_xcp_info_is_visible,
+};
+
+static const struct kobj_type ualink_xcp_info_ktype = {
+	.sysfs_ops = &kobj_sysfs_ops
+};
+
 static void ualink_info_release(struct kobject *kobj)
 {
 	kfree(to_ualink_info(kobj));
@@ -1467,6 +1550,9 @@ static int ualink_kobj_add(struct kobject *kobj, struct kobject *parent,
 	return r;
 }
 
+static void amdgpu_ualink_xcp_sysfs_init(struct amdgpu_device *adev);
+static void amdgpu_ualink_xcp_sysfs_fini(struct amdgpu_device *adev);
+
 int amdgpu_ualink_sysfs_init(struct amdgpu_device *adev)
 {
 	struct amdgpu_ualink_info *info = adev->ualink.info;
@@ -1496,7 +1582,9 @@ int amdgpu_ualink_sysfs_init(struct amdgpu_device *adev)
 	if (r)
 		goto err_config;
 
+	amdgpu_ualink_xcp_sysfs_init(adev);
 	adev->ualink.sysfs_init = true;
+
 	return 0;
 
 err_config:
@@ -1515,6 +1603,7 @@ void amdgpu_ualink_sysfs_fini(struct amdgpu_device *adev)
 	if (!adev->ualink.sysfs_init)
 		return;
 
+	amdgpu_ualink_xcp_sysfs_fini(adev);
 	kobject_del(&adev->ualink.stations->kobj);
 	kobject_del(&adev->ualink.config->kobj);
 	kobject_del(&adev->ualink.setup->kobj);
@@ -1522,6 +1611,82 @@ void amdgpu_ualink_sysfs_fini(struct amdgpu_device *adev)
 	adev->ualink.sysfs_init = false;
 }
 
+static int amdgpu_ualink_xcp_sysfs_add(struct amdgpu_xcp *xcp)
+{
+	struct amdgpu_device *adev = xcp->xcp_mgr->adev;
+	int r;
+
+	r = kobject_init_and_add(&xcp->ualink.kobj, &ualink_xcp_info_ktype,
+				 &xcp->ddev->dev->kobj, "ualink");
+	if (r)
+		goto err;
+
+	r = sysfs_create_group(&xcp->ualink.kobj, &ualink_xcp_info_group);
+	if (r)
+		goto err;
+
+	/* pin info so it outlives this node regardless of teardown order */
+	kobject_get(&adev->ualink.info->kobj);
+	xcp->ualink.sysfs = true;
+	return 0;
+err:
+	kobject_put(&xcp->ualink.kobj);
+	return r;
+}
+
+void amdgpu_ualink_xcp_sysfs_update(struct amdgpu_xcp *xcp)
+{
+	if (!xcp->ualink.sysfs)
+		return;
+
+	sysfs_update_group(&xcp->ualink.kobj, &ualink_xcp_info_group);
+}
+
+static void amdgpu_ualink_xcp_sysfs_remove(struct amdgpu_xcp *xcp)
+{
+	struct amdgpu_device *adev = xcp->xcp_mgr->adev;
+
+	if (!xcp->ualink.sysfs)
+		return;
+
+	/* group is only populated for valid partitions */
+	if (xcp->valid)
+		sysfs_remove_group(&xcp->ualink.kobj, &ualink_xcp_info_group);
+	kobject_put(&adev->ualink.info->kobj);
+	kobject_put(&xcp->ualink.kobj);
+	xcp->ualink.sysfs = false;
+}
+
+static void amdgpu_ualink_xcp_sysfs_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_xcp *xcp;
+	int i;
+
+	if (!adev->xcp_mgr)
+		return;
+
+	for (i = 0; i < MAX_XCP; i++) {
+		xcp = &adev->xcp_mgr->xcp[i];
+		if (!xcp->ddev || amdgpu_xcp_is_primary(xcp))
+			continue;
+		amdgpu_ualink_xcp_sysfs_add(xcp);
+	}
+}
+
+static void amdgpu_ualink_xcp_sysfs_fini(struct amdgpu_device *adev)
+{
+	struct amdgpu_xcp *xcp;
+	int i;
+
+	if (!adev->xcp_mgr)
+		return;
+
+	for (i = 0; i < MAX_XCP; i++) {
+		xcp = &adev->xcp_mgr->xcp[i];
+		amdgpu_ualink_xcp_sysfs_remove(xcp);
+	}
+}
+
 static int amdgpu_ualink_npa_alloc_va(struct amdgpu_device *adev,
 			       struct drm_mm_node *mm_node,
 			       u64 va, u64 range_start,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.h
index 2e579d0bbe9e6..63710b484c6a5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.h
@@ -27,6 +27,8 @@
 
 #include <linux/uuid.h>
 
+struct amdgpu_xcp;
+
 #define AMDGPU_UALINK_ACCEL_MAX 256
 #define AMDGPU_UALINK_LOCAL_ACCELS_MAX 8
 #define AMDGPU_UALINK_STATIONS_MAX 64
@@ -398,6 +400,7 @@ int amdgpu_ualink_resume_handler(struct amdgpu_device *adev);
 
 int amdgpu_ualink_sysfs_init(struct amdgpu_device *adev);
 void amdgpu_ualink_sysfs_fini(struct amdgpu_device *adev);
+void amdgpu_ualink_xcp_sysfs_update(struct amdgpu_xcp *xcp);
 int amdgpu_ualink_manager_start(struct amdgpu_device *adev);
 void amdgpu_ualink_manager_stop(struct amdgpu_device *adev);
 int amdgpu_ualink_export_handle(struct drm_device *dev, struct drm_file *filp,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
index 18f4455ef0a8f..a8c79a66df1d8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
@@ -1061,6 +1061,11 @@ static const struct kobj_type xcp_sysfs_ktype = {
 	.sysfs_ops = &kobj_sysfs_ops,
 };
 
+bool amdgpu_xcp_is_primary(struct amdgpu_xcp *xcp)
+{
+	return xcp->ddev == adev_to_drm(xcp->xcp_mgr->adev);
+}
+
 static void amdgpu_xcp_sysfs_entries_fini(struct amdgpu_xcp_mgr *xcp_mgr, int n)
 {
 	struct amdgpu_xcp *xcp;
@@ -1110,6 +1115,7 @@ static void amdgpu_xcp_sysfs_entries_update(struct amdgpu_xcp_mgr *xcp_mgr)
 		if (!xcp->ddev)
 			continue;
 		sysfs_update_group(&xcp->kobj, &amdgpu_xcp_attrs_group);
+		amdgpu_ualink_xcp_sysfs_update(xcp);
 	}
 
 	return;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h
index 878c1c422893c..33157409eda0c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h
@@ -100,6 +100,11 @@ struct amdgpu_xcp_ip {
 	bool valid;
 };
 
+struct amdgpu_xcp_ualink {
+	struct kobject kobj;
+	bool sysfs;
+};
+
 struct amdgpu_xcp {
 	struct amdgpu_xcp_ip ip[AMDGPU_XCP_MAX_BLOCKS];
 
@@ -115,6 +120,7 @@ struct amdgpu_xcp {
 	struct amdgpu_sched	gpu_sched[AMDGPU_HW_IP_NUM][AMDGPU_RING_PRIO_MAX];
 	struct amdgpu_xcp_mgr *xcp_mgr;
 	struct kobject kobj;
+	struct amdgpu_xcp_ualink ualink;
 	uint64_t unique_id;
 };
 
@@ -191,6 +197,7 @@ int amdgpu_xcp_pre_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags);
 int amdgpu_xcp_post_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags);
 void amdgpu_xcp_sysfs_init(struct amdgpu_device *adev);
 void amdgpu_xcp_sysfs_fini(struct amdgpu_device *adev);
+bool amdgpu_xcp_is_primary(struct amdgpu_xcp *xcp);
 
 static inline int amdgpu_xcp_get_num_xcp(struct amdgpu_xcp_mgr *xcp_mgr)
 {
-- 
2.55.0