A whole-GPU reset can cover every device in an XGMI hive, while a KFD process may be bound to only one of those devices. The reset notifier currently looks up a user GPU ID for every process on every reset node. For an unrelated node the lookup returns -EINVAL and triggers a false WARN_ONCE().
Look up the process device data first and skip processes that are not bound to the reset node. Keep the warning for a missing user GPU ID when a process device mapping does exist, since that remains an inconsistent state. Signed-off-by: Xiang Liu <[email protected]> --- drivers/gpu/drm/amd/amdkfd/kfd_events.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c index eb6fa4af295c..05ce2fbfa4f4 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c @@ -1162,17 +1162,18 @@ void kfd_signal_reset_event(struct kfd_node *dev) idx = srcu_read_lock(&kfd_processes_srcu); hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { - int user_gpu_id = kfd_process_get_user_gpu_id(p, dev->id); - struct kfd_process_device *pdd = kfd_get_process_device_data(dev, p); + struct kfd_process_device *pdd; + int user_gpu_id; - if (unlikely(user_gpu_id == -EINVAL)) { - WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", dev->id); + pdd = kfd_get_process_device_data(dev, p); + /* A hive reset can include nodes this process never opened. */ + if (!pdd) continue; - } - if (unlikely(!pdd)) { - WARN_ONCE(1, "Could not get device data from process pid:%d\n", - p->lead_thread->pid); + user_gpu_id = kfd_process_get_user_gpu_id(p, dev->id); + + if (unlikely(user_gpu_id == -EINVAL)) { + WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", dev->id); continue; } -- 2.34.1
