vgsp: m76 - R2 lane tail (TSG PREEMPT bWait + device-sched off + perf boost)

The real client's context-creation tail after the lane-enable trio is
#1073 NVA06C_CTRL_CMD_PREEMPT on the TSG with bWait=1, #1075
NV0080 device-sched disable, #1077 PERF_BOOST_SET_2X. The explicit preempt
makes the RM perform a clean context save/switch-out at setup time so the
first dispatch loads the context normally; without it our lane's context sits
loaded from the ladder and task activation ends in a save/unload instead of
execution (m72/m73/m75 signature). Gated by gsp_phys_rmcall_gr_lane_tail
(default 0).

1 files changed, 21 insertions, 0 deletions

diff --git a/kernel/gsp_phys_gpu.c b/kernel/gsp_phys_gpu.c
index 2f8a127..a394425 100644
--- a/kernel/gsp_phys_gpu.c
+++ b/kernel/gsp_phys_gpu.c
@@ -796,6 +796,27 @@ unsigned int gsp_phys_rmcall_gr_lmem_tpc_size_value(void)
 EXPORT_SYMBOL_GPL(gsp_phys_rmcall_gr_lmem_tpc_size_value);
 
 /*
+ * m76 lane tail (gsp_phys_rmcall_gr_lane_tail; default 0). When set, the
+ * mode-35 lane enable also sends the R2 capture's tail after #1063: TSG
+ * PREEMPT (bWait=1, #1073), device-sched disable (#1075) and the
+ * PERF_BOOST_SET_2X control (#1077). The real client preempts its own TSG at
+ * context creation so the first dispatch performs a clean context load; our
+ * lane's context otherwise sits loaded from the ladder and task activation
+ * ends in a save/unload instead of execution.
+ */
+static int gsp_phys_rmcall_gr_lane_tail_param;
+module_param_named(gsp_phys_rmcall_gr_lane_tail,
+		   gsp_phys_rmcall_gr_lane_tail_param, int, 0644);
+MODULE_PARM_DESC(gsp_phys_rmcall_gr_lane_tail,
+		 "Send the R2 #1073/#1075/#1077 lane tail after the mode-35 lane enable (default 0)");
+
+int gsp_phys_rmcall_gr_lane_tail_value(void)
+{
+	return gsp_phys_rmcall_gr_lane_tail_param;
+}
+EXPORT_SYMBOL_GPL(gsp_phys_rmcall_gr_lane_tail_value);
+
+/*
  * m74 R1/F2 ordering (gsp_phys_rmcall_gr_r1_first; default 0). 1 = send the
  * VA-only booking promote (R1) before F2's mapped-MAIN promote so the
  * one-shot context-image initializer (0x52a94f8, pEngCtx+0x135) sees the