reset_queues_mes() describes the hung queue to
amdgpu_gfx_reset_mes_compute() with a remove queue input that lacks the
queue's gang context slot. If MES finds no hung queue,
amdgpu_gfx_reset_mes_compute() removes this queue with
remove_hw_queue(), and in MES RS64 mode MES then removes the gang in
slot 0 instead of the queue's.

Fixes: e32b68c6b135 ("drm/amdkfd: enable rs64mem for kfd queue")
Signed-off-by: Prike Liang <[email protected]>
---
 drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c 
b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 86b70748617b..878159275841 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -479,6 +479,7 @@ static int reset_queues_mes(struct device_queue_manager 
*dqm, struct queue *q)
        memset(&queue_input, 0x0, sizeof(struct mes_remove_queue_input));
        queue_input.doorbell_offset = q->properties.doorbell_off;
        queue_input.gang_context_addr = q->gang_ctx_gpu_addr;
+       queue_input.gang_context_array_index = q->gang_ctx_array_index;
        queue_input.queue_type = 
convert_to_amdgpu_ring_type(q->properties.type);
        queue_input.remove_queue_after_reset = false;
        queue_input.xcc_id = ffs(dqm->dev->xcc_mask) - 1;
-- 
2.34.1

Reply via email to