From mboxrd@z Thu Jan 1 00:00:00 1970 From: robherring2@gmail.com (Rob Herring) Date: Mon, 10 Dec 2012 10:30:01 -0600 Subject: [PATCH] OF: update coherent_dma_mask In-Reply-To: <1355141369-30030-1-git-send-email-subash.rp@samsung.com> References: <1355141369-30030-1-git-send-email-subash.rp@samsung.com> Message-ID: <50C60E09.8050805@gmail.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 12/10/2012 06:09 AM, Subash Patel wrote: > This patch is tested in ARM:exynos5250 with LPAE enabled. The coherent_dma_mask > needs to be defined to DMA_BIT_MASK(64) as dma-mapping API's check it against > 64-bit mask. > > Signed-off-by: Subash Patel > --- > drivers/of/platform.c | 2 +- > 1 files changed, 1 insertions(+), 1 deletions(-) > > diff --git a/drivers/of/platform.c b/drivers/of/platform.c > index 9bdeaf30..3c5417b 100644 > --- a/drivers/of/platform.c > +++ b/drivers/of/platform.c > @@ -214,7 +214,7 @@ struct platform_device *of_platform_device_create_pdata( > #if defined(CONFIG_MICROBLAZE) > dev->archdata.dma_mask = 0xffffffffUL; > #endif > - dev->dev.coherent_dma_mask = DMA_BIT_MASK(32); > + dev->dev.coherent_dma_mask = DMA_BIT_MASK(64); > dev->dev.bus = &platform_bus_type; > dev->dev.platform_data = platform_data; This is assuming all masters support 64-bit addressing which is probably not universally valid. This really needs to be determined from the DT and there's been recent patches to do this[1] although they need more work. [1] - https://lkml.org/lkml/2012/12/4/54 Rob