gpu: nvgpu: Add NVGPU_IOCTL_CHANNEL_SET_BOOSTED_CTX

This ioctl can be used on gp10b to set a flag in the context header
indicating this context should be run at elevated clock
frequency. FECS ctxsw ucode will read this flag as part of the context
switch and will request higher GPU clock frequencies from BPMP for the
duration of the context execution.

Bug 1819874

Change-Id: I84bf580923d95585095716d49cea24e58c9440ed
Signed-off-by: Peter Boonstoppel <pboonstoppel@nvidia.com>
Reviewed-on: http://git-master/r/1292746
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

4 files changed, 26 insertions, 0 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
index b963ad96..3dda1cbf 100644
--- a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
@@ -3874,6 +3874,24 @@ long gk20a_channel_ioctl(struct file *filp,
                         err = -EINVAL;
                 }
                 break;
+        case NVGPU_IOCTL_CHANNEL_SET_BOOSTED_CTX:
+                if (ch->g->ops.gr.set_boosted_ctx) {
+                        bool boost =
+                                ((struct nvgpu_boosted_ctx_args *)buf)->boost;
+
+                        err = gk20a_busy(dev);
+                        if (err) {
+                                dev_err(dev,
+                                        "%s: failed to host gk20a for ioctl cmd: 0x%x",
+                                        __func__, cmd);
+                                break;
+                        }
+                        err = ch->g->ops.gr.set_boosted_ctx(ch, boost);
+                        gk20a_idle(dev);
+                } else {
+                        err = -EINVAL;
+                }
+                break;
         default:
                 dev_dbg(dev, "unrecognized ioctl cmd: 0x%x", cmd);
                 err = -ENOTTY;
diff --git a/drivers/gpu/nvgpu/gk20a/gk20a.h b/drivers/gpu/nvgpu/gk20a/gk20a.h
index 8d6d36fc..7f2383d9 100644
--- a/drivers/gpu/nvgpu/gk20a/gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/gk20a.h
@@ -319,6 +319,10 @@ struct gpu_ops {
                                 struct vm_gk20a *vm, u32 class,
                                 u32 graphics_preempt_mode,
                                 u32 compute_preempt_mode);
+                int (*set_boosted_ctx)(struct channel_gk20a *ch, bool boost);
+                void (*update_boosted_ctx)(struct gk20a *g,
+                                           struct mem_desc *mem,
+                                           struct gr_ctx_desc *gr_ctx);
                 int (*fuse_override)(struct gk20a *g);
                 void (*init_sm_id_table)(struct gk20a *g);
                 int (*load_smid_config)(struct gk20a *g);
diff --git a/drivers/gpu/nvgpu/gk20a/gr_gk20a.c b/drivers/gpu/nvgpu/gk20a/gr_gk20a.c
index 0e1c88a4..170bfc7f 100644
--- a/drivers/gpu/nvgpu/gk20a/gr_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/gr_gk20a.c
@@ -2109,6 +2109,9 @@ int gr_gk20a_load_golden_ctx_image(struct gk20a *g,
         if (g->ops.gr.update_ctxsw_preemption_mode)
                 g->ops.gr.update_ctxsw_preemption_mode(g, ch_ctx, mem);
 
+        if (g->ops.gr.update_boosted_ctx)
+                g->ops.gr.update_boosted_ctx(g, mem, ch_ctx->gr_ctx);
+
         virt_addr_lo = u64_lo32(ch_ctx->patch_ctx.mem.gpu_va);
         virt_addr_hi = u64_hi32(ch_ctx->patch_ctx.mem.gpu_va);
 
diff --git a/drivers/gpu/nvgpu/gk20a/gr_gk20a.h b/drivers/gpu/nvgpu/gk20a/gr_gk20a.h
index 19ab2ec0..5a987a82 100644
--- a/drivers/gpu/nvgpu/gk20a/gr_gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/gr_gk20a.h
@@ -391,6 +391,7 @@ struct gr_ctx_desc {
 
         u32 graphics_preempt_mode;
         u32 compute_preempt_mode;
+        bool boosted_ctx;
 #ifdef CONFIG_ARCH_TEGRA_18x_SOC
         struct gr_ctx_desc_t18x t18x;
 #endif