vgsp: ctxshare_force probe - share on the proven mode-31 execution channel (V6+sibling A attach via params init)

3 files changed, 48 insertions, 0 deletions

diff --git a/kernel/gsp_phys_booter.c b/kernel/gsp_phys_booter.c
index 10ef3b7..4fc8b4f 100644
--- a/kernel/gsp_phys_booter.c
+++ b/kernel/gsp_phys_booter.c
@@ -13144,6 +13144,27 @@ static int gsp_phys_rmcall_stage3(void __iomem *bar0)
 		gsp_phys_rmcall_user_lane_kgr_queries(bar0);
 
 	/*
+	 * Mode-31 share probe (gsp_phys_rmcall_ctxshare_force=1): allocate the
+	 * share before the replay so every ladder-TSG channel (V6 and the mode-31
+	 * sibling A) attaches to it via the channel-params helper, while the
+	 * proven mode-31 construct/commit flow stays intact.
+	 */
+	if (gsp_phys_rmcall_engine_is_gr() &&
+	    gsp_phys_rmcall_ctxshare_force_value() &&
+	    !boot_stage.rmcall_ctxshare) {
+		if (!gsp_phys_rmcall_vas_shared_value()) {
+			pr_err("vgsp-boot: rm-call: ctxshare_force needs the ladder SHARED_MANAGEMENT VAS (gsp_phys_rmcall_vas_shared=1); aborting with no retry\n");
+			return -EINVAL;
+		}
+		ret = gsp_phys_rmcall_alloc_ctxshare(bar0);
+		if (ret) {
+			pr_err("vgsp-boot: rm-call: stage 3 aborted: ctxshare_force FERMI_CONTEXT_SHARE_A alloc failed (%d); aborting with no retry\n",
+			       ret);
+			return ret;
+		}
+	}
+
+	/*
 	 * Golden-image replay (gsp_phys_rmcall_golden=1 + gsp_phys_gpu=1):
 	 * replay the stock client's post-load golden channel allocations after
 	 * the ladder's VAS/TSG exist and before the ladder's own channel/GR
diff --git a/kernel/gsp_phys_gpu.c b/kernel/gsp_phys_gpu.c
index e2fffc9..974100a 100644
--- a/kernel/gsp_phys_gpu.c
+++ b/kernel/gsp_phys_gpu.c
@@ -309,6 +309,25 @@ bool gsp_phys_rmcall_golden_enabled(void)
 EXPORT_SYMBOL_GPL(gsp_phys_rmcall_golden_enabled);
 
 /*
+ * Mode-31 share probe (default 0): allocate FERMI_CONTEXT_SHARE_A on the
+ * ladder TSG before the golden replay for ANY GR promote mode, so the proven
+ * mode-31 execution channel V6 (and the sibling A) attach to the share while
+ * keeping mode 31's construct/commit flow. Tests whether the context share is
+ * the QMD-dispatch discriminator. Requires gsp_phys_rmcall_vas_shared=1.
+ */
+static bool gsp_phys_rmcall_ctxshare_force_param;
+module_param_named(gsp_phys_rmcall_ctxshare_force,
+		   gsp_phys_rmcall_ctxshare_force_param, bool, 0644);
+MODULE_PARM_DESC(gsp_phys_rmcall_ctxshare_force,
+		 "Force FERMI_CONTEXT_SHARE_A allocation before the golden replay for any GR promote mode (default 0; requires gsp_phys_rmcall_vas_shared=1)");
+
+bool gsp_phys_rmcall_ctxshare_force_value(void)
+{
+	return gsp_phys_rmcall_ctxshare_force_param;
+}
+EXPORT_SYMBOL_GPL(gsp_phys_rmcall_ctxshare_force_value);
+
+/*
  * P5 wave-1 shadow work forwarding (default 0). When set, the sema service
  * requests one real CE copy on the debugfs ladder's channel for each decoded
  * guest CE push, using only the ladder's module-owned buffers; the software
diff --git a/kernel/gsp_phys_gpu.h b/kernel/gsp_phys_gpu.h
index 8cfb5e9..2087e33 100644
--- a/kernel/gsp_phys_gpu.h
+++ b/kernel/gsp_phys_gpu.h
@@ -128,6 +128,14 @@ u32 gsp_phys_rmcall_golden_exec_shape_value(void);
 bool gsp_phys_rmcall_golden_enabled(void);
 
 /*
+ * True when gsp_phys_rmcall_ctxshare_force is set (default 0): allocate
+ * FERMI_CONTEXT_SHARE_A on the ladder TSG before the golden replay for any GR
+ * promote mode, so the mode-31 execution channel V6 attaches to the share
+ * while the mode-31 construct/commit flow stays intact.
+ */
+bool gsp_phys_rmcall_ctxshare_force_value(void);
+
+/*
  * True when the gsp_phys_work_forward module parameter is set (default 0: the
  * guest's CE pushes stay software-emulated exactly as before). When set, the
  * sema service shadows each decoded CE push as one copy on the debugfs ladder's