From: Harish Kasiviswanathan <Harish.Kasiviswanathan@xxxxxxx>
Some AMD motherboards without an APU have a broken CRAT table which
causes KFD initialization failures or incorrect information about
NUMA nodes, CPU cores or system memory. Ignore CRAT tables without
GPUs and rely on KFD's code to create a CRAT table for the CPU.
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan at amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling at amd.com>
---
drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 25 ++++++++++++++++++++++---
1 file changed, 22 insertions(+), 3 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
index e7daf2c..7f0d41e 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
@@ -904,6 +904,25 @@ static void kfd_add_non_crat_information(struct kfd_topology_device *kdev)
/* TODO: For GPU node, rearrange code from kfd_topology_add_device */
}
+/* kfd_is_acpi_crat_invalid - CRAT from ACPI is valid only for AMD APU devices.
+ * Ignore CRAT for all other devices. AMD APU is identified if both CPU
+ * and GPU cores are present.
+ * @device_list - topology device list created by parsing ACPI CRAT table.
+ * @return - TRUE if invalid, FALSE is valid.
+ */
+static bool kfd_is_acpi_crat_invalid(struct list_head *device_list)
+{
+ struct kfd_topology_device *dev;
+
+ list_for_each_entry(dev, device_list, list) {
+ if (dev->node_props.cpu_cores_count &&
+ dev->node_props.simd_count)
+ return false;
+ }
+ pr_info("Ignoring ACPI CRAT on non-APU system\n");
+ return true;
+}
+
int kfd_topology_init(void)
{
void *crat_image = NULL;
@@ -936,7 +955,7 @@ int kfd_topology_init(void)
/*
* Get the CRAT image from the ACPI. If ACPI doesn't have one
- * create a virtual CRAT.
+ * or if ACPI CRAT is invalid create a virtual CRAT.
* NOTE: The current implementation expects all AMD APUs to have
* CRAT. If no CRAT is available, it is assumed to be a CPU
*/
@@ -945,7 +964,8 @@ int kfd_topology_init(void)
ret = kfd_parse_crat_table(crat_image,
&temp_topology_device_list,
proximity_domain);
- if (ret) {
+ if (ret ||
+ kfd_is_acpi_crat_invalid(&temp_topology_device_list)) {
kfd_release_topology_device_list(
&temp_topology_device_list);
kfd_destroy_crat_image(crat_image);
@@ -971,7 +991,6 @@ int kfd_topology_init(void)
goto err;
}
}
-
kdev = list_first_entry(&temp_topology_device_list,
struct kfd_topology_device, list);
kfd_add_perf_to_topology(kdev);
--
2.7.4
