On 18.08.25 11:51, Sunil Khatri wrote: > Print PD address of VM root instead of GPU address in the debugfs
Better write: Print root PD address in PDE format instead of..." On modern GPUs this is what UMR expects in the registers as well. This way UMR can handle page tables both in VRAM as well as system memory. Regards, Christian. > so UMR tool need not workaround to get the right information of an > entry flags and other bits and not just an GPU address which hides > these bits. > > Signed-off-by: Sunil Khatri <[email protected]> > --- > drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c > b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c > index 127091de0f34..a70651050acf 100644 > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c > @@ -2155,7 +2155,7 @@ static int amdgpu_pt_info_read(struct seq_file *m, void > *unused) > return -EINVAL; > } > > - seq_printf(m, "gpu_address: 0x%llx\n", > amdgpu_bo_gpu_offset(fpriv->vm.root.bo)); > + seq_printf(m, "pd_address: 0x%llx\n", > amdgpu_gmc_pd_addr(fpriv->vm.root.bo)); > seq_printf(m, "max_pfn: 0x%llx\n", adev->vm_manager.max_pfn); > seq_printf(m, "num_level: 0x%x\n", adev->vm_manager.num_level); > seq_printf(m, "block_size: 0x%x\n", adev->vm_manager.block_size);
