vgsp-sema: bounded sema-norel diagnostic for readable CE pushes without a release

The entries the pump holds for the 8 s CE grace are exactly the ones whose
decode found no release. Under SEMA_DBG this logs the decode state
(have_sem, rel va/payload, ops) and the first 64 push bytes for every
content-bearing CE push that decoded no release (2048-line budget,
all-zero reads are filtered so they cannot wash it out). Used in r129 to
show that a push whose tail crosses the CE page boundary reads 0xFF
garbage for its second page - the sem+LAUNCH_DMA live exactly in that
tail, so only boundary-crossing pushes were undecodable.

1 files changed, 38 insertions, 0 deletions

diff --git a/kernel/gsp_sema_service.c b/kernel/gsp_sema_service.c
index 9ab609a..45a9843 100644
--- a/kernel/gsp_sema_service.c
+++ b/kernel/gsp_sema_service.c
@@ -602,6 +602,14 @@ struct sema_svc {
 	u64			grce_log;		/* GRCE inline-write budget */
 	u32			cur_ch;			/* channel of the op being run */
 	u64			entry_log;		/* per-push summary budget */
+	/*
+	 * m151: bounded log for CE pushes that were READABLE (nonzero first
+	 * bytes) yet decoded no release. These are the entries the pump holds
+	 * for the 8 s grace; naming their decode state (have_sem, rel) and
+	 * head bytes is the Class-2 diagnostic (r128: correct content, right
+	 * sem block at +0xC0 with RELEASE_ONE_WORD, yet rel_seen false).
+	 */
+	u64			entry_norel_log;
 	u64			ops_denied;		/* work refused: implausible stream */
 	u64			false_trans;		/* resolvable VA mapped to non-push */
 	u64			sem_write_reresolve;	/* writes re-resolved after read-back */
@@ -6936,6 +6944,36 @@ static void svc_decode_stream(struct nv_vgpu_device *vgpu, const u8 *buf,
 				tlen ? tail : NULL);
 		}
 	}
+
+	/*
+	 * m151 Class-2 diagnostic: a CE push with real content that decoded no
+	 * release is the one that pays the 8 s grace. Log its decode state
+	 * (bounded, content-gated so all-zero reads do not wash it out).
+	 */
+	if (gsp_sema_dbg && ch->engine_type && ch->h_object && !st.have_rel) {
+		u8 head[64];
+		u32 hlen = len < sizeof(head) ? len : (u32)sizeof(head);
+		bool nz = false;
+		u32 i;
+
+		for (i = 0; i + 4 <= hlen; i += 4) {
+			u32 w;
+
+			memcpy(&w, buf + i, 4);
+			if (w) {
+				nz = true;
+				break;
+			}
+		}
+		if (nz && svc.entry_norel_log < 2048) {
+			svc.entry_norel_log++;
+			memcpy(head, buf, hlen);
+			pr_info("sema-norel: h=0x%x eng=%u va=0x%llx len=%u ops=%u have_sem=%u rel=0x%llx pay=%u head=%*phN\n",
+				ch->h_object, ch->engine_type, push_va, len,
+				ops_runs, st.have_sem, st.rel_va,
+				st.rel_payload, hlen, head);
+		}
+	}
 }
 
 /* ------------------------------------------------------------------ */