On 08/09/2023 03:52, Abhinav Kumar wrote:
On 9/7/2023 5:35 PM, Dmitry Baryshkov wrote:
On 08/09/2023 03:32, Abhinav Kumar wrote:
_dpu_plane_calc_bw() uses integer variables to calculate the bandwidth
used during plane bandwidth calculations. However for high resolution
displays this overflows easily and leads to below errors
Can we move the u64 conversion closer to the place where we actually
need it? Having u64 source width looks very strange.
Its this math
https://gitlab.freedesktop.org/drm/msm/-/blob/msm-next/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c#L139 which overflows.
So its not just one variable but I can even change this to u32 as that
also fixes the issue. Let me know if u32 works better.
Perhaps I went too far to go from int to u64.
I'd prefer to have the u64 conversion around the actual calculations -
so that we emphasise the issue, not the size of the width / etc.
[dpu error]crtc83 failed performance check -7
Promote the intermediate variables to u64 to avoid overflow.
Fixes: c33b7c0389e1 ("drm/msm/dpu: add support for clk and bw scaling
for display")
Closes: https://gitlab.freedesktop.org/drm/msm/-/issues/32
Signed-off-by: Abhinav Kumar <quic_abhinavk@xxxxxxxxxxx>
---
drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
b/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
index ae970af1154f..c6193131beec 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
@@ -118,7 +118,7 @@ static u64 _dpu_plane_calc_bw(const struct
dpu_mdss_cfg *catalog,
const struct drm_display_mode *mode,
struct dpu_sw_pipe_cfg *pipe_cfg)
{
- int src_width, src_height, dst_height, fps;
+ u64 src_width, src_height, dst_height, fps;
u64 plane_prefill_bw;
u64 plane_bw;
u32 hw_latency_lines;
--
With best wishes
Dmitry
