Re: [PATCH 1/4] drm/amdgpu: clear RB at ring init

On 07.06.2016 20:38, Liu, Monk wrote:

...

There's something else about these two patches I'm a bit worried about:

...

The GPU should only read data from ring buffers and IBs that we previously explicitly wrote there. I'm afraid these patches might just paper over bugs elsewhere, which might still bite us under different circumstances.

so do you mean we should use NOP to fulfill the ring buffer instead of 0 ?

I mean we should find out why the GPU seems to read something else than the packets we write to the ring buffer / IBs.