vgsp+guest-diag: kcore-truth pin resolver (pool/pushbuffer/channel ring+userd) + GR ring userd-derived candidates

guest_pin_stream.py now resolves translations entirely from guest truth via
/proc/kcore + a manual page-table walk (no QMP races): the CE semaphore pool
and pushbuffer memory from the UVM rm_mem structures, and each channel's
GPFIFO ring + GPPUT from the tracking semaphore's embedding uvm_channel_t
(channel_info located by its gpFifoGpuVa value). This removes the watcher
pipeline's flakiness on the cudaMalloc stall class (missing pool/pushbuffer
pins).

gsp_sema_service.c: svc_gr_ring_autopin tries the guest-registered USERD
memdesc base as a ring candidate (the fixed FB convention only matches when
the guest's VRAM allocator happens to reproduce the old layout), plus an
optional (gsp_sema_dbg) one-shot FB scan + raw-word diagnostics when every
candidate fails.

tools/cuda-ab/p2: self-checking arbitrary-kernel acceptance suite (10
programs: sequential launches, shapes, params, smem/GEMM, streams, malloc
churn, printf+error, FMA benchmark, mixed, managed probe) with host oracle
expectations and a guest runner.

1 files changed, 39 insertions, 0 deletions

diff --git a/tools/cuda-ab/p2/build.sh b/tools/cuda-ab/p2/build.sh
new file mode 100755
index 0000000..867182c
--- /dev/null
+++ b/tools/cuda-ab/p2/build.sh
@@ -0,0 +1,39 @@
+#!/bin/bash
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# p2/build.sh - build the P2 "arbitrary CUDA kernels" acceptance programs.
+#
+# Target is sm_75 (Turing TU104) to match the project's GPU. The CUDA
+# runtime is linked statically (--cudart=static) so the guest needs only the
+# driver's libcuda.so, exactly like ab.sh/victory_mail.
+#
+# Usage: ./build.sh [program ...]     (default: all programs)
+set -e
+
+HERE="$(cd "$(dirname "$0")" && pwd)"
+NVCC="${NVCC:-nvcc}"
+FLAGS="${P2_NVCC_FLAGS:--O2 -arch=sm_75 --cudart=static}"
+
+command -v "$NVCC" >/dev/null || { echo "nvcc not found"; exit 1; }
+
+mkdir -p "$HERE/bin"
+
+build_one() {
+	local src="$HERE/src/$1.cu"
+	local bin="$HERE/bin/$1"
+
+	[ -f "$src" ] || { echo "[build] missing source: $src" >&2; return 1; }
+	echo "[build] $1"
+	# shellcheck disable=SC2086
+	"$NVCC" $FLAGS -I"$HERE/src" -o "$bin" "$src"
+}
+
+if [ "$#" -gt 0 ]; then
+	for p in "$@"; do build_one "$p"; done
+else
+	for src in "$HERE"/src/p2_*.cu; do
+		build_one "$(basename "$src" .cu)"
+	done
+fi
+
+echo "[build] OK: $(ls "$HERE"/bin/p2_* 2>/dev/null | wc -l) program(s) in $HERE/bin"