gpu: nvgpu: Use coherent aperture flag

When using a coherent DMA API wee must make sure to program
any aperture fields with the coherent aperture setting. To
do this the nvgpu_aperture_mask() function was modified to
take a third aperture mask argument, a coherent setting, so
that code can use this function to generate coherent aperture
settings.

The aperture choice is some what tricky: the default version
of this function uses the state of the DMA API to determine
what aperture to use for SYSMEM: either coherent or
non-coherent internally. Thus a kernel user need only specify
the normal nvgpu_mem struct and the correct mask should be
chosen. Due to many uses of nvgpu_mem structs not created
directly from the DMA API wrapper it's easier to translate
SYSMEM to SYSMEM_COH after creation.

However, the GMMU mapping code, will encounter buffers from
userspace with difference coerency attributes than the DMA
API. Thus the __nvgpu_aperture_mask() really respects the
aperture setting passed in regardless of the DMA API state.
This aperture setting is pulled from NVGPU_VM_MAP_IO_COHERENT
since this is either passed in from userspace or set by the
kernel when using coherent DMA. The aperture field in attrs
is upgraded to coh if this flag is set.

This change also adds a coherent sysmem mask everywhere that
it can. There's a couple places that do not have a coherent
register field defined yet. These need to eventually be
defined and added.

Lastly the aperture mask code has been mvoed from the Linux
vm.c code to the general vm.c code since this function has
no Linux dependencies.

Note: depends on https://git-master.nvidia.com/r/1664536 for
new register fields.

JIRA EVLR-2333

Change-Id: I4b347911ecb7c511738563fe6c34d0e6aa380d71
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1655220
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

7 files changed, 45 insertions, 29 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/bus_gk20a.c b/drivers/gpu/nvgpu/gk20a/bus_gk20a.c
index 7f0cfe58..b2800772 100644
--- a/drivers/gpu/nvgpu/gk20a/bus_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/bus_gk20a.c
@@ -21,6 +21,7 @@
  */
 
 #include <nvgpu/page_allocator.h>
+#include <nvgpu/enabled.h>
 #include <nvgpu/log.h>
 #include <nvgpu/soc.h>
 #include <nvgpu/bus.h>
@@ -155,8 +156,9 @@ int gk20a_bus_bar1_bind(struct gk20a *g, struct nvgpu_mem *bar1_inst)
 
         gk20a_writel(g, bus_bar1_block_r(),
                      nvgpu_aperture_mask(g, bar1_inst,
-                       bus_bar1_block_target_sys_mem_ncoh_f(),
-                       bus_bar1_block_target_vid_mem_f()) |
+                                         bus_bar1_block_target_sys_mem_ncoh_f(),
+                                         bus_bar1_block_target_sys_mem_coh_f(),
+                                         bus_bar1_block_target_vid_mem_f()) |
                      bus_bar1_block_mode_virtual_f() |
                      bus_bar1_block_ptr_f(ptr_v));
 
diff --git a/drivers/gpu/nvgpu/gk20a/fb_gk20a.c b/drivers/gpu/nvgpu/gk20a/fb_gk20a.c
index a5a2cb51..e3052701 100644
--- a/drivers/gpu/nvgpu/gk20a/fb_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/fb_gk20a.c
@@ -98,8 +98,9 @@ void gk20a_fb_tlb_invalidate(struct gk20a *g, struct nvgpu_mem *pdb)
         gk20a_writel(g, fb_mmu_invalidate_pdb_r(),
                 fb_mmu_invalidate_pdb_addr_f(addr_lo) |
                 nvgpu_aperture_mask(g, pdb,
-                  fb_mmu_invalidate_pdb_aperture_sys_mem_f(),
-                  fb_mmu_invalidate_pdb_aperture_vid_mem_f()));
+                                fb_mmu_invalidate_pdb_aperture_sys_mem_f(),
+                                fb_mmu_invalidate_pdb_aperture_sys_mem_f(),
+                                fb_mmu_invalidate_pdb_aperture_vid_mem_f()));
 
         gk20a_writel(g, fb_mmu_invalidate_r(),
                 fb_mmu_invalidate_all_va_true_f() |
diff --git a/drivers/gpu/nvgpu/gk20a/fecs_trace_gk20a.c b/drivers/gpu/nvgpu/gk20a/fecs_trace_gk20a.c
index 409661fc..4fda0d2e 100644
--- a/drivers/gpu/nvgpu/gk20a/fecs_trace_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/fecs_trace_gk20a.c
@@ -653,6 +653,7 @@ int gk20a_fecs_trace_bind_channel(struct gk20a *g,
                 return -ENOMEM;
         aperture = nvgpu_aperture_mask(g, &trace->trace_buf,
                         ctxsw_prog_main_image_context_timestamp_buffer_ptr_hi_target_sys_mem_noncoherent_f(),
+                        ctxsw_prog_main_image_context_timestamp_buffer_ptr_hi_target_sys_mem_coherent_f(),
                         ctxsw_prog_main_image_context_timestamp_buffer_ptr_hi_target_vid_mem_f());
 
         if (nvgpu_mem_begin(g, mem))
diff --git a/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c b/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c
index dd0b78c0..247557aa 100644
--- a/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c
@@ -3240,8 +3240,9 @@ static int gk20a_fifo_update_runlist_locked(struct gk20a *g, u32 runlist_id,
                 gk20a_writel(g, fifo_runlist_base_r(),
                         fifo_runlist_base_ptr_f(u64_lo32(runlist_iova >> 12)) |
                         nvgpu_aperture_mask(g, &runlist->mem[new_buf],
-                          fifo_runlist_base_target_sys_mem_ncoh_f(),
-                          fifo_runlist_base_target_vid_mem_f()));
+                                fifo_runlist_base_target_sys_mem_ncoh_f(),
+                                fifo_runlist_base_target_sys_mem_coh_f(),
+                                fifo_runlist_base_target_vid_mem_f()));
         }
 
         gk20a_writel(g, fifo_runlist_r(),
@@ -3763,8 +3764,9 @@ static int gk20a_fifo_commit_userd(struct channel_gk20a *c)
         nvgpu_mem_wr32(g, &c->inst_block,
                        ram_in_ramfc_w() + ram_fc_userd_w(),
                        nvgpu_aperture_mask(g, &g->fifo.userd,
-                        pbdma_userd_target_sys_mem_ncoh_f(),
-                        pbdma_userd_target_vid_mem_f()) |
+                                           pbdma_userd_target_sys_mem_ncoh_f(),
+                                           pbdma_userd_target_sys_mem_coh_f(),
+                                           pbdma_userd_target_vid_mem_f()) |
                        pbdma_userd_addr_f(addr_lo));
 
         nvgpu_mem_wr32(g, &c->inst_block,
diff --git a/drivers/gpu/nvgpu/gk20a/gr_gk20a.c b/drivers/gpu/nvgpu/gk20a/gr_gk20a.c
index 6ae743ef..2cde10ec 100644
--- a/drivers/gpu/nvgpu/gk20a/gr_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/gr_gk20a.c
@@ -745,8 +745,9 @@ static u32 fecs_current_ctx_data(struct gk20a *g, struct nvgpu_mem *inst_block)
         u64 ptr = nvgpu_inst_block_addr(g, inst_block) >>
                 ram_in_base_shift_v();
         u32 aperture = nvgpu_aperture_mask(g, inst_block,
-                        gr_fecs_current_ctx_target_sys_mem_ncoh_f(),
-                        gr_fecs_current_ctx_target_vid_mem_f());
+                                gr_fecs_current_ctx_target_sys_mem_ncoh_f(),
+                                gr_fecs_current_ctx_target_sys_mem_coh_f(),
+                                gr_fecs_current_ctx_target_vid_mem_f());
 
         return gr_fecs_current_ctx_ptr_f(u64_lo32(ptr)) | aperture |
                 gr_fecs_current_ctx_valid_f(1);
@@ -2171,16 +2172,18 @@ void gr_gk20a_load_falcon_bind_instblk(struct gk20a *g)
 
         inst_ptr = nvgpu_inst_block_addr(g, &ucode_info->inst_blk_desc);
         gk20a_writel(g, gr_fecs_new_ctx_r(),
-                        gr_fecs_new_ctx_ptr_f(inst_ptr >> 12) |
-                        nvgpu_aperture_mask(g, &ucode_info->inst_blk_desc,
+                     gr_fecs_new_ctx_ptr_f(inst_ptr >> 12) |
+                     nvgpu_aperture_mask(g, &ucode_info->inst_blk_desc,
                                 gr_fecs_new_ctx_target_sys_mem_ncoh_f(),
+                                gr_fecs_new_ctx_target_sys_mem_coh_f(),
                                 gr_fecs_new_ctx_target_vid_mem_f()) |
-                        gr_fecs_new_ctx_valid_m());
+                     gr_fecs_new_ctx_valid_m());
 
         gk20a_writel(g, gr_fecs_arb_ctx_ptr_r(),
-                        gr_fecs_arb_ctx_ptr_ptr_f(inst_ptr >> 12) |
-                        nvgpu_aperture_mask(g, &ucode_info->inst_blk_desc,
+                     gr_fecs_arb_ctx_ptr_ptr_f(inst_ptr >> 12) |
+                     nvgpu_aperture_mask(g, &ucode_info->inst_blk_desc,
                                 gr_fecs_arb_ctx_ptr_target_sys_mem_ncoh_f(),
+                                gr_fecs_arb_ctx_ptr_target_sys_mem_coh_f(),
                                 gr_fecs_arb_ctx_ptr_target_vid_mem_f()));
 
         gk20a_writel(g, gr_fecs_arb_ctx_cmd_r(), 0x7);
@@ -4379,8 +4382,9 @@ static int gk20a_init_gr_setup_hw(struct gk20a *g)
 
         gk20a_writel(g, fb_mmu_debug_wr_r(),
                      nvgpu_aperture_mask(g, &gr->mmu_wr_mem,
-                       fb_mmu_debug_wr_aperture_sys_mem_ncoh_f(),
-                       fb_mmu_debug_wr_aperture_vid_mem_f()) |
+                                fb_mmu_debug_wr_aperture_sys_mem_ncoh_f(),
+                                fb_mmu_debug_wr_aperture_sys_mem_coh_f(),
+                                fb_mmu_debug_wr_aperture_vid_mem_f()) |
                      fb_mmu_debug_wr_vol_false_f() |
                      fb_mmu_debug_wr_addr_f(addr));
 
@@ -4389,8 +4393,9 @@ static int gk20a_init_gr_setup_hw(struct gk20a *g)
 
         gk20a_writel(g, fb_mmu_debug_rd_r(),
                      nvgpu_aperture_mask(g, &gr->mmu_rd_mem,
-                       fb_mmu_debug_wr_aperture_sys_mem_ncoh_f(),
-                       fb_mmu_debug_rd_aperture_vid_mem_f()) |
+                                fb_mmu_debug_wr_aperture_sys_mem_ncoh_f(),
+                                fb_mmu_debug_wr_aperture_sys_mem_coh_f(),
+                                fb_mmu_debug_rd_aperture_vid_mem_f()) |
                      fb_mmu_debug_rd_vol_false_f() |
                      fb_mmu_debug_rd_addr_f(addr));
 
diff --git a/drivers/gpu/nvgpu/gk20a/mm_gk20a.c b/drivers/gpu/nvgpu/gk20a/mm_gk20a.c
index b27d1109..4ff6125b 100644
--- a/drivers/gpu/nvgpu/gk20a/mm_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/mm_gk20a.c
@@ -122,8 +122,9 @@ static inline u32 big_valid_pde0_bits(struct gk20a *g,
 {
         u32 pde0_bits =
                 nvgpu_aperture_mask(g, pd->mem,
-                  gmmu_pde_aperture_big_sys_mem_ncoh_f(),
-                  gmmu_pde_aperture_big_video_memory_f()) |
+                                    gmmu_pde_aperture_big_sys_mem_ncoh_f(),
+                                    gmmu_pde_aperture_big_sys_mem_coh_f(),
+                                    gmmu_pde_aperture_big_video_memory_f()) |
                 gmmu_pde_address_big_sys_f(
                            (u32)(addr >> gmmu_pde_address_shift_v()));
 
@@ -135,8 +136,9 @@ static inline u32 small_valid_pde1_bits(struct gk20a *g,
 {
         u32 pde1_bits =
                 nvgpu_aperture_mask(g, pd->mem,
-                  gmmu_pde_aperture_small_sys_mem_ncoh_f(),
-                  gmmu_pde_aperture_small_video_memory_f()) |
+                                    gmmu_pde_aperture_small_sys_mem_ncoh_f(),
+                                    gmmu_pde_aperture_small_sys_mem_coh_f(),
+                                    gmmu_pde_aperture_small_video_memory_f()) |
                 gmmu_pde_vol_small_true_f() | /* tbd: why? */
                 gmmu_pde_address_small_sys_f(
                            (u32)(addr >> gmmu_pde_address_shift_v()));
@@ -215,6 +217,7 @@ static void __update_pte(struct vm_gk20a *vm,
 
         pte_w[1] = __nvgpu_aperture_mask(g, attrs->aperture,
                                          gmmu_pte_aperture_sys_mem_ncoh_f(),
+                                         gmmu_pte_aperture_sys_mem_coh_f(),
                                          gmmu_pte_aperture_video_memory_f()) |
                 gmmu_pte_kind_f(attrs->kind_v) |
                 gmmu_pte_comptagline_f((u32)(attrs->ctag >> ctag_shift));
@@ -268,7 +271,7 @@ static void update_gmmu_pte_locked(struct vm_gk20a *vm,
                 page_size >> 10,
                 nvgpu_gmmu_perm_str(attrs->rw_flag),
                 attrs->kind_v,
-                nvgpu_aperture_str(attrs->aperture),
+                nvgpu_aperture_str(g, attrs->aperture),
                 attrs->cacheable ? 'C' : '-',
                 attrs->sparse    ? 'S' : '-',
                 attrs->priv      ? 'P' : '-',
@@ -363,11 +366,12 @@ void gk20a_mm_init_pdb(struct gk20a *g, struct nvgpu_mem *inst_block,
         gk20a_dbg_info("pde pa=0x%llx", pdb_addr);
 
         nvgpu_mem_wr32(g, inst_block, ram_in_page_dir_base_lo_w(),
-                nvgpu_aperture_mask(g, vm->pdb.mem,
-                  ram_in_page_dir_base_target_sys_mem_ncoh_f(),
-                  ram_in_page_dir_base_target_vid_mem_f()) |
-                ram_in_page_dir_base_vol_true_f() |
-                ram_in_page_dir_base_lo_f(pdb_addr_lo));
+                       nvgpu_aperture_mask(g, vm->pdb.mem,
+                                ram_in_page_dir_base_target_sys_mem_ncoh_f(),
+                                ram_in_page_dir_base_target_sys_mem_coh_f(),
+                                ram_in_page_dir_base_target_vid_mem_f()) |
+                       ram_in_page_dir_base_vol_true_f() |
+                       ram_in_page_dir_base_lo_f(pdb_addr_lo));
 
         nvgpu_mem_wr32(g, inst_block, ram_in_page_dir_base_hi_w(),
                 ram_in_page_dir_base_hi_f(pdb_addr_hi));
diff --git a/drivers/gpu/nvgpu/gk20a/pramin_gk20a.c b/drivers/gpu/nvgpu/gk20a/pramin_gk20a.c
index 05d0473e..711aeb0d 100644
--- a/drivers/gpu/nvgpu/gk20a/pramin_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/pramin_gk20a.c
@@ -41,6 +41,7 @@ u32 gk20a_pramin_enter(struct gk20a *g, struct nvgpu_mem *mem,
         u32 lo = (u32)(addr & 0xfffff);
         u32 win = nvgpu_aperture_mask(g, mem,
                         bus_bar0_window_target_sys_mem_noncoherent_f(),
+                        bus_bar0_window_target_sys_mem_coherent_f(),
                         bus_bar0_window_target_vid_mem_f()) |
                 bus_bar0_window_base_f(hi);