Tegra: Perform cache maintenance on video carveout memory

Currently, the non-overlapping video memory carveout region is cleared after
disabling the MMU at EL3. If at any exception level the carveout region is being
marked as cacheable, this zeroing of memory will not have an affect on the
cached lines. Hence, we first invalidate the dirty lines and update the memory
and invalidate again so that both caches and memory is zeroed out.

Change-Id: If3b2d139ab7227f6799c0911d59e079849dc86aa

diff --git a/plat/nvidia/tegra/common/drivers/memctrl/memctrl.c b/plat/nvidia/tegra/common/drivers/memctrl/memctrl.c
index 0d8e370f57f43f6bcbccaaa59546032e7e3b0585..40d1bab067335711c6680e1d0723abf8e888ef98 100644 (file)
--- a/plat/nvidia/tegra/common/drivers/memctrl/memctrl.c
+++ b/plat/nvidia/tegra/common/drivers/memctrl/memctrl.c
@@ -107,6 +107,23 @@ void tegra_memctrl_tzdram_setup(uint64_t phys_base, uint32_t size_in_bytes)
        tegra_mc_write_32(MC_SECURITY_CFG1_0, size_in_bytes >> 20);
 }
 
+static void tegra_clear_videomem(uintptr_t non_overlap_area_start,
+                                unsigned long long non_overlap_area_size)
+{
+       /*
+        * Perform cache maintenance to ensure that the non-overlapping area is
+        * zeroed out. The first invalidation of this range ensures that
+        * possible evictions of dirty cache lines do not interfere with the
+        * 'zeromem16' operation. Other CPUs could speculatively prefetch the
+        * main memory contents of this area between the first invalidation and
+        * the 'zeromem16' operation. The second invalidation ensures that any
+        * such cache lines are removed as well.
+        */
+       inv_dcache_range(non_overlap_area_start, non_overlap_area_size);
+       zeromem16((void *)non_overlap_area_start, non_overlap_area_size);
+       inv_dcache_range(non_overlap_area_start, non_overlap_area_size);
+}
+
 /*
  * Program the Video Memory carveout region
  *
@@ -118,7 +135,7 @@ void tegra_memctrl_videomem_setup(uint64_t phys_base, uint32_t size_in_bytes)
        uintptr_t vmem_end_old = video_mem_base + (video_mem_size << 20);
        uintptr_t vmem_end_new = phys_base + size_in_bytes;
        uint32_t regval;
-       uint64_t size;
+       unsigned long long non_overlap_area_size;
 
        /*
         * The GPU is the user of the Video Memory region. In order to
@@ -155,15 +172,15 @@ void tegra_memctrl_videomem_setup(uint64_t phys_base, uint32_t size_in_bytes)
 
        disable_mmu_el3();
        if (phys_base > vmem_end_old || video_mem_base > vmem_end_new) {
-               zeromem16((void *)video_mem_base, video_mem_size << 20);
+               tegra_clear_videomem(video_mem_base, video_mem_size << 20);
        } else {
                if (video_mem_base < phys_base) {
-                       size = phys_base - video_mem_base;
-                       zeromem16((void *)video_mem_base, size);
+                       non_overlap_area_size = phys_base - video_mem_base;
+                       tegra_clear_videomem(video_mem_base, non_overlap_area_size);
                }
                if (vmem_end_old > vmem_end_new) {
-                       size = vmem_end_old - vmem_end_new;
-                       zeromem16((void *)vmem_end_new, size);
+                       non_overlap_area_size = vmem_end_old - vmem_end_new;
+                       tegra_clear_videomem(vmem_end_new, non_overlap_area_size);
                }
        }
        enable_mmu_el3(0);