On Thu, Oct 28, 2021 at 11:19 AM Paul Menzel <[email protected]> wrote: > > Dear Alex, > > > On 28.10.21 16:21, Alex Deucher wrote: > > The DMA mask on SI parts is 40 bits not 44. Copy > > paste typo. > > > > Fixes: 244511f386ccb9 ("drm/amdgpu: simplify and cleanup setting the dma > > mask") > > This was present in Linux 5.4-rc1. Can it also be the cause of some of > the crashes with the AMD Ryzen systems like Bug 206487 - Random > freezes/crashes with enabled C-State C6 - AMD Ryzen [1]? I know, there > were also some things with Linux 4.19.57, but, if my memory serves > correctly, they were much more often with Linux 5.4. Some crashes had > MCE and other did not, some froze the system, some rebooted it.
I doubt it would cause an MCE. You should just see an IOMMU page fault. Alex > > > Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1762 > > Acked-by: Christian König <[email protected]> > > Tested-by: Paul Menzel <[email protected]> > > Signed-off-by: Alex Deucher <[email protected]> > > --- > > drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c | 4 ++-- > > 1 file changed, 2 insertions(+), 2 deletions(-) > > > > diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c > > b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c > > index 0e81e03e9b49..0fe714f54cca 100644 > > --- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c > > +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c > > @@ -841,12 +841,12 @@ static int gmc_v6_0_sw_init(void *handle) > > > > adev->gmc.mc_mask = 0xffffffffffULL; > > > > - r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(44)); > > + r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(40)); > > if (r) { > > dev_warn(adev->dev, "No suitable DMA available.\n"); > > return r; > > } > > - adev->need_swiotlb = drm_need_swiotlb(44); > > + adev->need_swiotlb = drm_need_swiotlb(40); > > > > r = gmc_v6_0_init_microcode(adev); > > if (r) { > > > > > Kind regards, > > Paul > > > [1]: https://bugzilla.kernel.org/show_bug.cgi?id=206487
