On Fri, Mar 09, 2018 at 11:12:37AM -0300, Fabio Estevam wrote:
Hi Russell,
On Fri, Mar 9, 2018 at 8:34 AM, Russell King - ARM Linux linux@armlinux.org.uk wrote:
On Fri, Mar 09, 2018 at 12:20:59PM +0100, Lucas Stach wrote:
All the DRM GEM dma-buf import/export operations are done through the virtual DRM master device. As this isn't instanciated from DT anymore we need to make sure the DMA ops are set up correctly.
Signed-off-by: Lucas Stach l.stach@pengutronix.de
drivers/gpu/drm/etnaviv/etnaviv_drv.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c index ab50090d066c..d7666aed943b 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c @@ -655,7 +655,8 @@ static int etnaviv_pdev_probe(struct platform_device *pdev) struct device *dev = &pdev->dev; struct component_match *match = NULL;
dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(32));
arch_setup_dma_ops(dev, 0, 0x100000000, NULL, false);dma_coerce_mask_and_coherent(dev, DMA_BIT_MASK(32));NAK. dma_coerce_mask_and_coherent() exists for broken devices. Please instead ensure that the device is created with the proper default DMA mask.
Should it be like this (like vc4_drv.c) ?
--- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c @@ -578,6 +578,8 @@ static int etnaviv_bind(struct device *dev) struct drm_device *drm; int ret;
dev->coherent_dma_mask = DMA_BIT_MASK(32);
No. Drivers should be using dma_set_coherent_mask().
Whoever is creating this struct device has the responsibility to initialise the DMA mask to something sensible.