gpu: nvgpu: Remove phys_addr_t from common code

Remove phys_addr_t from common code and replace it with u64. This
faciliates QNX compiling the common code since phys_addr_t is a
Linux specific type.

JIRA NVGPU-30
JIRA NVGPU-226

Change-Id: I15fe2078f9cd0b07c7e90ad6e359c493afa56714
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1576432
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

1 files changed, 2 insertions, 1 deletions

diff --git a/drivers/gpu/nvgpu/common/linux/platform_gk20a_tegra.c b/drivers/gpu/nvgpu/common/linux/platform_gk20a_tegra.c
index 6379d8ef..49700ebe 100644
--- a/drivers/gpu/nvgpu/common/linux/platform_gk20a_tegra.c
+++ b/drivers/gpu/nvgpu/common/linux/platform_gk20a_tegra.c
@@ -139,7 +139,8 @@ static void gk20a_tegra_secure_destroy(struct gk20a *g,
         DEFINE_DMA_ATTRS(attrs);
 
         if (desc->mem.priv.sgt) {
-                phys_addr_t pa = sg_phys(desc->mem.priv.sgt->sgl);
+                u64 pa = nvgpu_mem_get_phys_addr(g, &desc->mem);
+
                 dma_set_attr(DMA_ATTR_NO_KERNEL_MAPPING, __DMA_ATTR(attrs));
                 dma_free_attrs(&tegra_vpr_dev, desc->mem.size,
                         (void *)(uintptr_t)pa,