guest-diag: adaptive miss-driven pinner; cdt holds on failure for inspection

pin_miss_pages.sh watches the host journal for release-miss VAs, resolves
each page for the live cdt pid through the guest pagemap and injects it (the
injection rewinds the pump so releases consumed under a stale pin are
re-read); it re-pins when a new cdt pid appears.  cuda_drv_test prints
cuGetErrorName/String on failure and, with CDT_HOLD=<seconds>, stays alive so
the RM error record and tracker state can be inspected after cuCtxCreate has
given up.

1 files changed, 24 insertions, 1 deletions

diff --git a/tools/test/cuda_drv_test.c b/tools/test/cuda_drv_test.c
index 20ad88a..8995a55 100644
--- a/tools/test/cuda_drv_test.c
+++ b/tools/test/cuda_drv_test.c
@@ -15,7 +15,9 @@
  *   gcc -O1 -o /tmp/cdt cuda_drv_test.c -ldl
  */
 #include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
+#include <unistd.h>
 #include <dlfcn.h>
 #include <stdint.h>
 
@@ -24,9 +26,30 @@ typedef int CUdevice;
 typedef void *CUcontext;
 typedef uint64_t CUdeviceptr;
 
+/* On failure: print cuGetErrorName/String and, when CDT_HOLD=<seconds> is
+ * set, keep the process alive so the live state (trackers, RM error record)
+ * can be inspected with gdb after cuCtxCreate has given up. */
+static void fail_info(CUresult r)
+{
+    const char *(*errname)(CUresult, const char **) = dlsym(RTLD_DEFAULT, "cuGetErrorName");
+    const char *(*errstr)(CUresult, const char **) = dlsym(RTLD_DEFAULT, "cuGetErrorString");
+    const char *s = NULL;
+    const char *hold = getenv("CDT_HOLD");
+
+    if (errname && errname(r, &s) == 0 && s)
+        printf("cuGetErrorName: %s\n", s);
+    if (errstr && errstr(r, &s) == 0 && s)
+        printf("cuGetErrorString: %s\n", s);
+    if (hold && atoi(hold) > 0) {
+        printf("CDT_HOLD: sleeping %d s for inspection\n", atoi(hold));
+        fflush(stdout);
+        sleep((unsigned)atoi(hold));
+    }
+}
+
 #define CHECK(expr) do { \
     CUresult r = (expr); \
-    if (r) { printf("%-24s FAILED rc=%d\n", #expr, r); return 1; } \
+    if (r) { printf("%-24s FAILED rc=%d\n", #expr, r); fail_info(r); return 1; } \
     printf("%-24s OK\n", #expr); \
 } while (0)