vgsp-gr: m152 scratch-PT alias cache + extended pool; m153 outcome-verifying launch retry

m152: the Tier-2 shadow alias pool consumed one never-reused 128 B fetch
line per mapping per launch (512 lines/boot), so a normal launch mix
exhausted it permanently: every later launch that needed a new mapping
was refused (the deterministic r152 'launch refused: no Tier-1 mapping'
loop). Two fixes: (a) an HPA-keyed alias cache in map_shadow_locked that
returns the existing never-rewritten VA for a page already mapped, so
repeat launches of the same buffers/sem pages consume nothing; a hit
reports line_out = ~0 so contiguity bookkeeping never extends an old
line; (b) a 48-window extension pool at VA 0x04000000 (PD0 indices
32..79, one 48-page PT chunk) raising capacity to 2048 lines/boot.

m153: the intermittent per-boot mis-execution classes (engine wrote
nothing; or wrote into a lower operand window while the output operand
stayed untouched - both complete with qmd_rel/ch_rel satisfied, live
evidence in r152 smiley: passes wrote b into the a window / nothing /
correct-after-10s-retry) are now detected by the m151 retirement wait
and retried: refill the bounce window from the guest inputs, re-publish
the same method block on a fresh ring entry, re-issue schedule, re-poll,
re-wait - bounded by the timeout path's attempt budget. Gate param
gsp_phys_rmcall_gr_nowrite_retry (default 1).

1 files changed, 25 insertions, 0 deletions

diff --git a/kernel/gsp_phys_gpu.c b/kernel/gsp_phys_gpu.c
index c5419a5..9b7235b 100644
--- a/kernel/gsp_phys_gpu.c
+++ b/kernel/gsp_phys_gpu.c
@@ -949,6 +949,31 @@ int gsp_phys_rmcall_gr_skedrec_value(void)
 EXPORT_SYMBOL_GPL(gsp_phys_rmcall_gr_skedrec_value);
 
 /*
+ * m153 no-write/wrong-place launch retry (gsp_phys_rmcall_gr_nowrite_retry;
+ * default 1). After a forwarded launch reports completion, the output
+ * retirement wait measures how many resolved data pages the engine actually
+ * changed. Two mis-execution signatures are known to be intermittent per
+ * boot: the engine writes nothing at all, or it writes into a lower operand
+ * window while the highest (output) operand stays untouched (a stale
+ * constant-source state). Both re-execute correctly on a fresh ring entry
+ * often enough to be worth a bounded retry: refill the bounce window from
+ * the guest inputs, re-publish the same method block, re-issue the schedule
+ * and re-poll, then re-wait. Set to 0 to disable the retry and accept the
+ * first outcome unconditionally.
+ */
+static int gsp_phys_rmcall_gr_nowrite_retry_param = 1;
+module_param_named(gsp_phys_rmcall_gr_nowrite_retry,
+		   gsp_phys_rmcall_gr_nowrite_retry_param, int, 0644);
+MODULE_PARM_DESC(gsp_phys_rmcall_gr_nowrite_retry,
+		 "Re-execute a forwarded GR launch whose retirement wait saw zero/wrong-place engine writes (default 1)");
+
+int gsp_phys_rmcall_gr_nowrite_retry_value(void)
+{
+	return gsp_phys_rmcall_gr_nowrite_retry_param;
+}
+EXPORT_SYMBOL_GPL(gsp_phys_rmcall_gr_nowrite_retry_value);
+
+/*
  * m69 window-0 shadow pool (gsp_phys_rmcall_gr_shadow_win0; default 0). When
  * set, the forwarded GR launch copies the QMD's program/param pages into the
  * window-0 shadow pool (module pages at VAs 0x108000..0x10f000, PTEs planted