gpu: nvgpu: When powering down, abort if not idle

When trying to power down GPU the engine might be still busy. In this
case delay power down by returning -EBUSY from
gk20a_pm_runtime_suspend().

Bug 200224907

Change-Id: Ibad74c090add24a185bc1a7a02df367af9b95ced
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: http://git-master/r/1213042
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

4 files changed, 24 insertions, 9 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
index b4c132ce..085caec5 100644
--- a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
@@ -2789,16 +2789,10 @@ int gk20a_channel_suspend(struct gk20a *g)
         struct fifo_gk20a *f = &g->fifo;
         u32 chid;
         bool channels_in_use = false;
-        int err;
         u32 active_runlist_ids = 0;
 
         gk20a_dbg_fn("");
 
-        /* wait for engine idle */
-        err = g->ops.fifo.wait_engine_idle(g);
-        if (err)
-                return err;
-
         for (chid = 0; chid < f->num_channels; chid++) {
                 struct channel_gk20a *ch = &f->channel[chid];
                 if (gk20a_channel_get(ch)) {
diff --git a/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c b/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c
index ff052400..2e38c4b6 100644
--- a/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/fifo_gk20a.c
@@ -3016,6 +3016,19 @@ bool gk20a_fifo_mmu_fault_pending(struct gk20a *g)
                 return false;
 }
 
+bool gk20a_fifo_is_engine_busy(struct gk20a *g)
+{
+        int i;
+
+        for (i = 0; i < fifo_engine_status__size_1_v(); i++) {
+                u32 status = gk20a_readl(g, fifo_engine_status_r(i));
+                if (fifo_engine_status_engine_v(status) ==
+                        fifo_engine_status_engine_busy_v())
+                        return true;
+        }
+        return false;
+}
+
 int gk20a_fifo_wait_engine_idle(struct gk20a *g)
 {
         unsigned long end_jiffies = jiffies +
@@ -3023,7 +3036,6 @@ int gk20a_fifo_wait_engine_idle(struct gk20a *g)
         unsigned long delay = GR_IDLE_CHECK_DEFAULT;
         int ret = -ETIMEDOUT;
         u32 i;
-        struct device *d = dev_from_gk20a(g);
 
         gk20a_dbg_fn("");
 
@@ -3041,7 +3053,7 @@ int gk20a_fifo_wait_engine_idle(struct gk20a *g)
                 } while (time_before(jiffies, end_jiffies) ||
                                 !tegra_platform_is_silicon());
                 if (ret) {
-                        gk20a_err(d, "cannot idle engine %u\n", i);
+                        gk20a_dbg_info("cannot idle engine %u", i);
                         break;
                 }
         }
diff --git a/drivers/gpu/nvgpu/gk20a/fifo_gk20a.h b/drivers/gpu/nvgpu/gk20a/fifo_gk20a.h
index 17c6dbf6..71a9d35d 100644
--- a/drivers/gpu/nvgpu/gk20a/fifo_gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/fifo_gk20a.h
@@ -206,6 +206,7 @@ void gk20a_init_fifo(struct gpu_ops *gops);
 void fifo_gk20a_finish_mmu_fault_handling(struct gk20a *g,
                 unsigned long fault_id);
 int gk20a_fifo_wait_engine_idle(struct gk20a *g);
+bool gk20a_fifo_is_engine_busy(struct gk20a *g);
 u32 gk20a_fifo_engine_interrupt_mask(struct gk20a *g);
 u32 gk20a_fifo_get_pbdma_signature(struct gk20a *g);
 u32 gk20a_fifo_get_failing_engine_data(struct gk20a *g,
diff --git a/drivers/gpu/nvgpu/gk20a/gk20a.c b/drivers/gpu/nvgpu/gk20a/gk20a.c
index f6bb9445..10fa86a1 100644
--- a/drivers/gpu/nvgpu/gk20a/gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/gk20a.c
@@ -767,6 +767,10 @@ static int gk20a_pm_prepare_poweroff(struct device *dev)
         if (!g->power_on)
                 goto done;
 
+        if (gk20a_fifo_is_engine_busy(g)) {
+                mutex_unlock(&g->poweroff_lock);
+                return -EBUSY;
+        }
         gk20a_scale_suspend(dev);
 
         /* cancel any pending cde work */
@@ -1353,6 +1357,7 @@ static int gk20a_pm_runtime_suspend(struct device *dev)
 fail_railgate:
         gk20a_pm_finalize_poweron(dev);
 fail:
+        pm_runtime_mark_last_busy(dev);
         return err;
 }
 
@@ -1751,6 +1756,7 @@ int __gk20a_do_idle(struct device *dev, bool force_reset)
                 msecs_to_jiffies(GK20A_WAIT_FOR_IDLE_MS);
         int ref_cnt;
         bool is_railgated;
+        int err = 0;
 
         /* acquire busy lock to block other busy() calls */
         down_write(&g->busy_lock);
@@ -1825,7 +1831,9 @@ int __gk20a_do_idle(struct device *dev, bool force_reset)
                  */
 
                 /* Save the GPU state */
-                gk20a_pm_prepare_poweroff(dev);
+                err = gk20a_pm_prepare_poweroff(dev);
+                if (err)
+                        goto fail_drop_usage_count;
 
                 /* railgate GPU */
                 platform->railgate(dev);