diff options
Diffstat (limited to 'tools/guest-diag/pin_gr_stream.py')
| -rw-r--r-- | tools/guest-diag/pin_gr_stream.py | 57 |
1 files changed, 53 insertions, 4 deletions
diff --git a/tools/guest-diag/pin_gr_stream.py b/tools/guest-diag/pin_gr_stream.py index 2e45bbf..e1f5195 100644 --- a/tools/guest-diag/pin_gr_stream.py +++ b/tools/guest-diag/pin_gr_stream.py @@ -16,6 +16,7 @@ import re import socket import subprocess import sys +import threading import time HERE = os.path.dirname(os.path.abspath(__file__)) @@ -66,10 +67,13 @@ def _start_cmd(): # a partially-failed first attempt (pid file not written yet) used to # leave the first scanner alive, and multiple scanners each re-assert # the whole CE pair set - r148 ran three and drove the guest to load 19 - # (clients wedged in userspace spins). The pattern is assembled at - # runtime via printf so this command's own cmdline (which contains the - # literal file path) cannot match the pkill. - return ("P=$(printf 'guest_pin_%%s.py' stream); " + # (clients wedged in userspace spins). The pattern is anchored + # (^python3) and assembled at runtime via printf: an unanchored pattern + # or a literal one would also match this command's own shell (its + # cmdline contains the script path), and pkill would kill the shell + # before it could start the new scanner - every start attempt then + # failed silently (r149). + return ("P=$(printf '^python3 /tmp/guest_pin_%%s.py' stream); " "sudo pkill -f \"$P\" 2>/dev/null; sleep 1; " "echo '%s' | base64 -d > /tmp/guest_pin_stream.py; " "sudo rm -f /tmp/guest_pin_stream.pid; " @@ -121,9 +125,54 @@ def scrub_register(fin_pa, hsem_pa): return False +def _demand_loop(): + """Push the host's unreadable CE push VAs to the guest scanner. + + The service logs each distinct stalled VA once per boot ('stalled VA + #N: 0x...'). Each sighting is tracked and re-sent every 20 s for five + minutes: the guest resolves the VA by content against the + nv_vm_map_pages areas (event-independent, no object chain), so a page + that was not yet written when first asked is picked up on a retry. + This closes the stall class where the one-shot #sem kprobe event was + lost and the chain resolver never learns the client. + """ + track = {} # va -> [first_seen, last_sent] + while True: + time.sleep(3) + try: + r = subprocess.run(["sudo", "-n", "dmesg"], + capture_output=True, text=True, timeout=20) + except Exception: + continue + now = time.time() + for m in re.finditer(r"stalled VA #\d+: (0x[0-9a-f]+)", + (r.stdout or "")[-200000:]): + va = int(m.group(1), 16) + track.setdefault(va, [now, 0.0]) + for va in list(track): + first, last = track[va] + if now - first > 300.0: + del track[va] + todo = [va for va, (f, l) in track.items() if now - l > 20.0] + if not todo: + continue + lines = "".join("0x%x\n" % va for va in todo) + try: + vexec("printf '%%s' '%s' >> /tmp/guest_want_vas.txt" % lines, 30) + except Exception: + continue + for va in todo: + track[va][1] = now + log("demand: sent %d stalled VA(s) to guest" % len(todo)) + + def main(): if not install_and_start(): log("WARN: could not start guest scanner (will retry via vsock wait)") + try: + threading.Thread(target=_demand_loop, daemon=True).start() + except Exception as e: + log("WARN: demand thread failed to start: %s" % e) srv = socket.socket(socket.AF_VSOCK, socket.SOCK_STREAM) srv.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1) srv.bind((socket.VMADDR_CID_ANY, PORT)) |