kernel: semaphore service forward-jump completion, stats, sema_probe

Relax the in-order rule to forward progress (payload > current), with the
jump bounded and counted: UVM tolerates forward jumps up to 2M
(2 * UVM_CHANNEL_NUM_GPFIFO_ENTRIES_MAX), and push buffers are recycled so
intermediate pushes can be missed. Add miss/stale/jump counters to the
rebuild log and a sema_probe module parameter to dump the service's map
entry and live slot values for a GPU VA (diagnostics for tracker stalls).

1 files changed, 32 insertions, 0 deletions

diff --git a/kernel/main.c b/kernel/main.c
index 177c729..637c79b 100644
--- a/kernel/main.c
+++ b/kernel/main.c
@@ -2116,6 +2116,38 @@ module_param_cb(scan_seq, &nv_vgpu_scan_seq_ops, NULL, 0200);
 MODULE_PARM_DESC(scan_seq, "Scan vidmem for a u32 sequence (hex,comma separated)");
 
 /*
+ * Debug: dump the semaphore service's view of a GPU VA (VA->PA map entry and
+ * the current semaphore slot values):
+ *   echo 0x121000010[,count] > .../parameters/sema_probe
+ */
+static int nv_vgpu_sema_probe_set(const char *val, const struct kernel_param *kp)
+{
+	char buf[64];
+	char *comma;
+	u64 va = 0;
+	u32 n = 8;
+
+	strscpy(buf, val, sizeof(buf));
+	comma = strchr(buf, ',');
+	if (comma) {
+		*comma = '\0';
+		if (kstrtou32(comma + 1, 0, &n))
+			return -EINVAL;
+	}
+	if (kstrtoull(buf, 0, &va))
+		return -EINVAL;
+
+	return gsp_sema_service_probe(va, n);
+}
+
+static const struct kernel_param_ops nv_vgpu_sema_probe_ops = {
+	.set = nv_vgpu_sema_probe_set,
+};
+
+module_param_cb(sema_probe, &nv_vgpu_sema_probe_ops, NULL, 0200);
+MODULE_PARM_DESC(sema_probe, "Dump sema-service map + slot values for a GPU VA");
+
+/*
  * Debug: find page-table entries pointing at a physical address, in guest RAM
  * and emulated vidmem:
  *   echo 0x12442b000 > .../parameters/scan_pt