From: Wenjing Liu <wenjing.liu@xxxxxxx>
[Why & How]
There is a chance where the RX issues HPD deassert in the
middle of link training, this will cause our logic to
abort link training and turn off link. However our payload
allocation logic needs to use current link settings to
determine average time slot per MTP. This will need to
use current link bandwidth as divider. This causes divide
by zero error occasionally. The fix is to skip DP2.0 payload
allocation logic if current link is not in 128b/132b mode.
Reviewed-by: George Shen <George.Shen@xxxxxxx>
Acked-by: Stylon Wang <stylon.wang@xxxxxxx>
Signed-off-by: Wenjing Liu <wenjing.liu@xxxxxxx>
---
drivers/gpu/drm/amd/display/dc/core/dc_link.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
index 48e274f9ea84..07f154fc4e56 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c
@@ -3546,7 +3546,8 @@ static enum dc_status dc_link_update_sst_payload(struct pipe_ctx *pipe_ctx,
}
/* slot X.Y for SST payload allocate */
- if (allocate) {
+ if (allocate && dp_get_link_encoding_format(&link->cur_link_settings) ==
+ DP_128b_132b_ENCODING) {
avg_time_slots_per_mtp = calculate_sst_avg_time_slots_per_mtp(stream, link);
dc_log_vcp_x_y(link, avg_time_slots_per_mtp);
--
2.35.1
