gpu: nvgpu: Reorg css HAL initialization

Reorganize HAL initialization to remove inheritance and construct
the gpu_ops struct at compile time. This patch only covers the
css sub-module of the gpu_ops struct.

Perform HAL function assignments in hal_gxxxx.c through the
population of a chip-specific copy of gpu_ops.

Jira NVGPU-74

Change-Id: I3bf696e13d359982c964c7bc470500a30555c034
Signed-off-by: Sunny He <suhe@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1514205
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Alex Waterman <alexw@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>

5 files changed, 58 insertions, 27 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.c b/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.c
index 452bcd11..8f655b26 100644
--- a/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.c
@@ -72,7 +72,7 @@ static inline u32 css_hw_get_pending_snapshots(struct gk20a *g)
 }
 
 /* informs hw how many snapshots have been processed (frees up fifo space) */
-static inline void css_hw_set_handled_snapshots(struct gk20a *g, u32 done)
+inline void css_hw_set_handled_snapshots(struct gk20a *g, u32 done)
 {
         if (done > 0) {
                 gk20a_writel(g, perf_pmasys_mem_bump_r(),
@@ -127,7 +127,7 @@ static int css_gr_create_shared_data(struct gr_gk20a *gr)
         return 0;
 }
 
-static int css_hw_enable_snapshot(struct channel_gk20a *ch,
+int css_hw_enable_snapshot(struct channel_gk20a *ch,
                                 struct gk20a_cs_snapshot_client *cs_client)
 {
         struct gk20a *g = ch->g;
@@ -203,7 +203,7 @@ failed_allocation:
         return ret;
 }
 
-static void css_hw_disable_snapshot(struct gr_gk20a *gr)
+void css_hw_disable_snapshot(struct gr_gk20a *gr)
 {
         struct gk20a *g = gr->g;
         struct gk20a_cs_snapshot *data = gr->cs_data;
@@ -399,7 +399,7 @@ next_hw_fifo_entry:
         return 0;
 }
 
-static u32 css_gr_allocate_perfmon_ids(struct gk20a_cs_snapshot *data,
+u32 css_gr_allocate_perfmon_ids(struct gk20a_cs_snapshot *data,
                                        u32 count)
 {
         unsigned long *pids = data->perfmon_ids;
@@ -415,7 +415,7 @@ static u32 css_gr_allocate_perfmon_ids(struct gk20a_cs_snapshot *data,
         return f;
 }
 
-static u32 css_gr_release_perfmon_ids(struct gk20a_cs_snapshot *data,
+u32 css_gr_release_perfmon_ids(struct gk20a_cs_snapshot *data,
                                       u32 start,
                                       u32 count)
 {
@@ -653,7 +653,7 @@ void gr_gk20a_free_cyclestats_snapshot_data(struct gk20a *g)
         nvgpu_mutex_destroy(&gr->cs_lock);
 }
 
-static int css_hw_check_data_available(struct channel_gk20a *ch, u32 *pending,
+int css_hw_check_data_available(struct channel_gk20a *ch, u32 *pending,
                                         bool *hw_overflow)
 {
         struct gk20a *g = ch->g;
@@ -670,13 +670,3 @@ static int css_hw_check_data_available(struct channel_gk20a *ch, u32 *pending,
         *hw_overflow = css_hw_get_overflow_status(g);
         return 0;
 }
-
-void gk20a_init_css_ops(struct gpu_ops *gops)
-{
-        gops->css.enable_snapshot = css_hw_enable_snapshot;
-        gops->css.disable_snapshot = css_hw_disable_snapshot;
-        gops->css.check_data_available = css_hw_check_data_available;
-        gops->css.set_handled_snapshots = css_hw_set_handled_snapshots;
-        gops->css.allocate_perfmon_ids = css_gr_allocate_perfmon_ids;
-        gops->css.release_perfmon_ids = css_gr_release_perfmon_ids;
-}
diff --git a/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.h b/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.h
index a95eaeae..804308bc 100644
--- a/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/css_gr_gk20a.h
@@ -1,7 +1,7 @@
 /*
  * GK20A Cycle stats snapshots support (subsystem for gr_gk20a).
  *
- * Copyright (c) 2016, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2016-2017, NVIDIA CORPORATION.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms and conditions of the GNU General Public License,
@@ -121,6 +121,16 @@ struct gk20a_cs_snapshot {
         struct gk20a_cs_snapshot_fifo_entry     *hw_get;
 };
 
-void gk20a_init_css_ops(struct gpu_ops *gops);
+inline void css_hw_set_handled_snapshots(struct gk20a *g, u32 done);
+int css_hw_enable_snapshot(struct channel_gk20a *ch,
+                                struct gk20a_cs_snapshot_client *cs_client);
+void css_hw_disable_snapshot(struct gr_gk20a *gr);
+u32 css_gr_allocate_perfmon_ids(struct gk20a_cs_snapshot *data,
+                                       u32 count);
+u32 css_gr_release_perfmon_ids(struct gk20a_cs_snapshot *data,
+                                      u32 start,
+                                      u32 count);
+int css_hw_check_data_available(struct channel_gk20a *ch, u32 *pending,
+                                        bool *hw_overflow);
 
 #endif /* CSS_GR_GK20A_H */
diff --git a/drivers/gpu/nvgpu/gm20b/hal_gm20b.c b/drivers/gpu/nvgpu/gm20b/hal_gm20b.c
index 2d8ed9ec..300f0872 100644
--- a/drivers/gpu/nvgpu/gm20b/hal_gm20b.c
+++ b/drivers/gpu/nvgpu/gm20b/hal_gm20b.c
@@ -234,6 +234,16 @@ static const struct gpu_ops gm20b_ops = {
         .cde = {
                 .get_program_numbers = gm20b_cde_get_program_numbers,
         },
+#if defined(CONFIG_GK20A_CYCLE_STATS)
+        .css = {
+                .enable_snapshot = css_hw_enable_snapshot,
+                .disable_snapshot = css_hw_disable_snapshot,
+                .check_data_available = css_hw_check_data_available,
+                .set_handled_snapshots = css_hw_set_handled_snapshots,
+                .allocate_perfmon_ids = css_gr_allocate_perfmon_ids,
+                .release_perfmon_ids = css_gr_release_perfmon_ids,
+        },
+#endif
         .falcon = {
                 .falcon_hal_sw_init = gk20a_falcon_hal_sw_init,
         },
@@ -253,6 +263,9 @@ int gm20b_init_hal(struct gk20a *g)
         gops->dbg_session_ops = gm20b_ops.dbg_session_ops;
         gops->debug = gm20b_ops.debug;
         gops->cde = gm20b_ops.cde;
+#if defined(CONFIG_GK20A_CYCLE_STATS)
+        gops->css = gm20b_ops.css;
+#endif
         gops->falcon = gm20b_ops.falcon;
 
         /* Lone functions */
@@ -303,9 +316,6 @@ int gm20b_init_hal(struct gk20a *g)
         gm20b_init_regops(gops);
         gm20b_init_therm_ops(gops);
         gk20a_init_tsg_ops(gops);
-#if defined(CONFIG_GK20A_CYCLE_STATS)
-        gk20a_init_css_ops(gops);
-#endif
         g->name = "gm20b";
 
         c->twod_class = FERMI_TWOD_A;
diff --git a/drivers/gpu/nvgpu/gp106/hal_gp106.c b/drivers/gpu/nvgpu/gp106/hal_gp106.c
index 2ebf93e2..c54232aa 100644
--- a/drivers/gpu/nvgpu/gp106/hal_gp106.c
+++ b/drivers/gpu/nvgpu/gp106/hal_gp106.c
@@ -278,6 +278,16 @@ static const struct gpu_ops gp106_ops = {
                 .need_scatter_buffer = gp10b_need_scatter_buffer,
                 .populate_scatter_buffer = gp10b_populate_scatter_buffer,
         },
+#if defined(CONFIG_GK20A_CYCLE_STATS)
+        .css = {
+                .enable_snapshot = css_hw_enable_snapshot,
+                .disable_snapshot = css_hw_disable_snapshot,
+                .check_data_available = css_hw_check_data_available,
+                .set_handled_snapshots = css_hw_set_handled_snapshots,
+                .allocate_perfmon_ids = css_gr_allocate_perfmon_ids,
+                .release_perfmon_ids = css_gr_release_perfmon_ids,
+        },
+#endif
         .xve = {
                 .sw_init          = xve_sw_init_gp106,
                 .get_speed        = xve_get_speed_gp106,
@@ -314,6 +324,9 @@ int gp106_init_hal(struct gk20a *g)
         gops->debug = gp106_ops.debug;
         gops->dbg_session_ops = gp106_ops.dbg_session_ops;
         gops->cde = gp106_ops.cde;
+#if defined(CONFIG_GK20A_CYCLE_STATS)
+        gops->css = gp106_ops.css;
+#endif
         gops->xve = gp106_ops.xve;
         gops->falcon = gp106_ops.falcon;
 
@@ -342,9 +355,6 @@ int gp106_init_hal(struct gk20a *g)
         gp106_init_regops(gops);
         gk20a_init_tsg_ops(gops);
         gk20a_init_pramin_ops(gops);
-#if defined(CONFIG_GK20A_CYCLE_STATS)
-        gk20a_init_css_ops(gops);
-#endif
         gp106_init_therm_ops(gops);
 
         g->name = "gp10x";
diff --git a/drivers/gpu/nvgpu/gp10b/hal_gp10b.c b/drivers/gpu/nvgpu/gp10b/hal_gp10b.c
index 796d26dd..e526af4c 100644
--- a/drivers/gpu/nvgpu/gp10b/hal_gp10b.c
+++ b/drivers/gpu/nvgpu/gp10b/hal_gp10b.c
@@ -245,6 +245,16 @@ static const struct gpu_ops gp10b_ops = {
                 .need_scatter_buffer = gp10b_need_scatter_buffer,
                 .populate_scatter_buffer = gp10b_populate_scatter_buffer,
         },
+#if defined(CONFIG_GK20A_CYCLE_STATS)
+        .css = {
+                .enable_snapshot = css_hw_enable_snapshot,
+                .disable_snapshot = css_hw_disable_snapshot,
+                .check_data_available = css_hw_check_data_available,
+                .set_handled_snapshots = css_hw_set_handled_snapshots,
+                .allocate_perfmon_ids = css_gr_allocate_perfmon_ids,
+                .release_perfmon_ids = css_gr_release_perfmon_ids,
+        },
+#endif
         .falcon = {
                 .falcon_hal_sw_init = gk20a_falcon_hal_sw_init,
         },
@@ -264,6 +274,9 @@ int gp10b_init_hal(struct gk20a *g)
         gops->debug = gp10b_ops.debug;
         gops->dbg_session_ops = gp10b_ops.dbg_session_ops;
         gops->cde = gp10b_ops.cde;
+#if defined(CONFIG_GK20A_CYCLE_STATS)
+        gops->css = gp10b_ops.css;
+#endif
         gops->falcon = gp10b_ops.falcon;
 
         /* Lone Functions */
@@ -323,9 +336,7 @@ int gp10b_init_hal(struct gk20a *g)
         gp10b_init_therm_ops(gops);
         gk20a_init_tsg_ops(gops);
         gk20a_init_pramin_ops(gops);
-#if defined(CONFIG_GK20A_CYCLE_STATS)
-        gk20a_init_css_ops(gops);
-#endif
+
         g->name = "gp10b";
 
         c->twod_class = FERMI_TWOD_A;