Pair tests that fail only on a CI runner give nothing to go on but the
output diff. The instance sync's and the relay's warnings say where an
instance stalled.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every hop through the relay waits a millisecond or so of real time, which
the speed multiplies: at 1600% on macOS CI a relay PTP receive in
pair/ptppoll timed out. Relay mode hardly gains above 800% anyway (26 s
against 22 s locally); P2P stays at 1600%.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Its entry in sendTargetPeers stayed, so the next blocking connect from
the same thread on a socket that got the same id returned BUSY
(pspautotests net/adhoc/pair/basic, connect to a closed port, after a
connected socket was closed).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Off the network the datagram reaches nobody and the send succeeds, but
the session was recovered first, and a send whose session wasn't up
returned WOULD_BLOCK (pspautotests net/adhoc/pair/vanish, PDP send after
leaving).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each instance paced itself against the wall clock, so on a busy machine
the one that got less CPU fell behind, and anything a test times across
instances (a peer leaving, a matching timeout) came out differently with
the load: pair/vanish and pair/matching3 failed on macOS CI.
With --sync-file and --sync-count, instances on one machine share a file
of one slot per instance. CoreTiming ends a slice at every emulated
millisecond, publishes it there, and waits until every other instance
has reached it too, so they're never more than a millisecond apart. It
works for any number of instances. One that stops moving (a debugger
pause, a crash) is waited for 2 s and then left out until it moves
again, and one that blocks the CPU thread in real time on something the
others must do (the relay PTP connect) stands aside meanwhile.
Tools/adhoc_multi.py and adhoc_session.py use it (--no-sync turns it
off). The host's 1 s head start is gone: they wait for the server's
port instead. adhoc_multi writes each instance's output to a file, as a
full pipe would now stop the other instances too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Fixes#19248
The server loop polled every socket and slept 10 ms, so it woke 100
times a second with nothing to do, and every message could wait up to
10 ms of real time before it was handled, which at a sped-up emulation
speed is many times that in game time. It now waits in select() on the
listening socket and every user, and wakes at least every 50 ms to notice
being stopped. It also no longer stops serving everyone while the hosting
instance is paused in the debugger.
pspautotests: pair/discover and pair/ptpmsg.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A PSP keeps what arrives in the socket's buffer, of the size given to
PdpCreate, and drops a datagram that doesn't fit: a buffer of 0 gets
nothing, one of 2048 bytes only some of ten 1000-byte datagrams
(pspautotests net/adhoc/pair/flow). Each transport now only hands over
the datagrams that have arrived (PdpTakeDatagram), into a queue per
socket that is filled every millisecond and whenever the game looks, and
PdpRecv, its blocking retry, GetPdpStat, PollSocket and game mode read
from that. This replaces each transport's own receive paths, so the
relay now reports the sender with NOT_ENOUGH_SPACE too, and P2P's
blocking receive no longer misreads a stale errno.
A datagram to ourselves goes straight into our own socket's queue, as on
a PSP, so it's there at once through the relay too (net/adhoc/term).
Also from comparing the transports:
- P2P: a blocking Accept of a connection from outside the group keeps
waiting instead of returning -1.
- Relay: an accepted socket gets an id of its own, not its listener's,
which blocking calls are told apart by.
- ALERT in PollSocket comes from the socket in both transports.
- No send/receive timeouts are set on the game's sockets: all host
sockets are non-blocking, and in relay mode the ids aren't host
sockets at all.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
It times the socket failures from when the peer drops off the peer list,
which the ad hoc server updates in wall time, so on macOS CI at 600% the
failure lands under a second after it instead of 1-2s. The relay run
still covers it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each instance runs on its own, so on a busy runner they drift apart in
emulated time, and a real second's lag is 8 emulated seconds at 800%:
long enough for pairSync (10 s) to give up on macOS CI. ADHOC_SPEED
still overrides it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Winsock never makes a socket whose connect failed writable; the failure
shows in select's exception set, with the reason in SO_ERROR. The poll
now checks both, and SO_ERROR on a writable socket too, which is where
other systems put a refusal.
Windows still retries a refused connect for about a second before
reporting it, far longer than a PSP at the tests' 800% speed, so
pair/basic is a known P2P failure on Windows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- A connect PtpOpen started counted as up as soon as getpeername
succeeded, which on Windows happens while it's still in progress, even
one about to be refused. It now needs the socket ready too, and a
pending connect's failure is read from SO_ERROR, the only place Windows
reports it (pspautotests net/adhoc/pair/basic).
- Out of the group, PdpRecv only dropped what its peek saw; on Windows a
peek can fail and leave the next datagram unchecked, so a vanished
client still got the host's data (pair/vanish).
pspautotests: pair/vanish times PTP's failure from when the peer left the
peer list.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The connect PtpOpen starts goes through the relay on a host thread, so
when it finished in emulated time depended on the emulation speed and on
how busy the host was. At --speed=800 on a loaded CI runner it often
wasn't up within the 100 ms net/adhoc/pair/ptppoll polls for, while on a
PSP it's up a few ms after PtpOpen.
The relay's per-millisecond poll now waits (at most a second, once per
connect) for a connect in progress to finish, filling local listeners'
backlogs meanwhile, since the listener may be on the same console.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Like the two-console tests, each runs through the built-in server's relay
and in P2P, with adhoc_known_failures for the ones that fail in one mode.
pair/basic passes in P2P now and moves to tests_good, a known failure
through the relay.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- GetPtpStat on a listener read its listen session as a connection,
which could write past it; a listener now reports nothing waiting.
Neither GetPtpStat nor GetPdpStat reports a dead session's negative
status as a byte count any more, and a new PDP socket starts with
nothing waiting instead of its buffer size.
- A poll of a connection whose peer has closed reports DISCONNECT, as
the P2P transport and a PSP do (pspautotests net/adhoc/poll).
- Only a send that went out counts as the last send for a vanished
peer's timeout.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A connect to a port nobody listens on is refused within a second on a
PSP, and every later connect on that socket is refused too, even once
something listens there (pspautotests net/adhoc/pair/basic,
pair/ptpclose). PPSSPP recreated the socket and tried again until the
game's timeout ran out, to cover a peer that wasn't listening yet.
Dissidia 012 and Fate/Unlimited Codes, named in the old comments, still
get to their matches.
Also set the established time wherever a connection becomes established,
which the timing of a vanished peer's failure counts from.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
How long an adhocctl request waits for the server before giving up, and
how long a joiner's GAME event waits for the host to show up, were
measured in wall time, so a game saw them shorter when sped up and
longer when slowed down.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A PSP completes a connect to a listener without the game calling Accept,
so an opened PTP connection is up a few ms after PtpOpen (pspautotests
net/adhoc/pair/ptppoll). The relay only answers the connecting side once
the listening side accepts, so the relay transport now accepts pending
connections into a queue as long as the listener's backlog, from a new
per-millisecond AdhocTransport::Poll() and from polls and accepts, and
PtpAccept takes them from there. A poll reports ACCEPT while the queue
isn't empty.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Out of the group, a PSP's sends reach nobody and nothing reaches it, and
a PTP connection to a peer that has left fails once its retransmits would
have run out. Only the relay transport did this; the P2P one kept sending
over the host's sockets, so neither side ever noticed the other was gone
(pspautotests net/adhoc/pair/vanish).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
--cpu-backend-only leaves out the tests that don't depend on the CPU
backend (net/ for now, the two-console ones taking about 40 s of a 60 s
run). The Linux CI jobs run everything with the JIT and pass it for
jit-ir, ir and the interpreter.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
wait, press --every, stick, type and until count the instances' emulated
time (the slowest one's), so a script means the same at any speed, and
start --speed PERCENT runs the instances sped up.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each started context had two host threads, polling every 10 ms of wall
time: one sent queued packets, received, and ran hellos, keepalives and
timeouts, and the other handed events to the callbacks. A single
CoreTiming event now does both for every running context, every 1 ms of
emulated time, and Stop does what the threads did on their way out.
Matching no longer depends on host scheduling or wall-clock delays, so an
accept's ESTABLISHED comes 1-10 ms after SelectTarget at any emulation
speed, as on a PSP (pspautotests net/adhoc/pair/matchingaccept).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The pair tests that pass go in tests_good, and each runs twice: through
the built-in server's relay and P2P. pair_known_failures lists the ones
that fail in only one mode. A P2P test that needs a loopback address
the machine lacks (macOS has only 127.0.0.1) is skipped with a warning,
and the run fails at the end with the command that adds it. The macOS
CI runners add 127.0.0.2. Each run gets a random port offset.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Polling an unconnected or listening PTP socket for SEND took the PDP
path in postoffice_can_send. For an opened connection that created a
PDP session and stored it as the socket's handle, so its sends and
receives failed afterwards (pspautotests net/adhoc/pair/ptppoll). A PTP
socket that isn't established now just isn't writable.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Fixes#22857
PES 2009 calls sceNetApctlGetState first and only calls sceNetApctlInit
when it fails with 0x80410a0d. We returned success, so Init was never
called, apctl's state never advanced and the Netconf dialog waited on
"Connecting to the access point" forever. A PSP returns 0x80410a0d before
Init and after Term (pspautotests net/adhoc/apctlinit).
After that the game polls sceNetGetLocalEtherAddr until it succeeds, which
we only allowed in ad hoc mode. Also allow it once apctl has connected. It
still fails with apctl initialized but not connected, as the test shows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
RelayTransport, with the aemu_postoffice helpers only it uses, goes to
AdhocTransportRelay.cpp, and P2PTransport to AdhocTransportP2P.cpp. The
helpers they share with sceNetAdhoc.cpp are declared in
AdhocTransport.h. Moved as they were.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The relay and P2P halves of PtpOpen, PtpListen, PtpClose, PtpSend,
PtpRecv, PtpAccept and their blocking retries, GetPtpStat, PtpFlush,
PollSocket and deleting all sockets move into RelayTransport and
P2PTransport. Connecting stays shared, with the transports answering
the steps that differ (start, check on, confirm, still pending).
No change in behaviour, except that a relay poll asks the postoffice
whether a socket is readable once instead of twice.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The relay and P2P halves of PdpCreate, PdpDelete, PdpSend, PdpRecv, the
blocking PDP receive, GetPdpStat and game mode's readiness checks move into
RelayTransport and P2PTransport, which Transport() picks between by
serverHasRelay. No change in behaviour.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
As on a PSP (pspautotests net/adhoc/abandon): the Disconnect fails and
the join goes on to CONNECT. Before, it was accepted, and the join's late
answer from the server could still connect us afterwards.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
On Android the render thread runs the UI, and surfaceDestroyed waits for
it, so slow storage (SAF especially) there stutters and can become an ANR.
- Store: only check whether the game is installed when an install or
uninstall finishes, not every frame.
- Savedata: read the modification time with SIZE on the GameInfo worker.
The popup asked for it every frame, and each button on first draw.
- FirmwareScreen: scan the NAND, and erase it, on a worker thread. It
rescanned on every RecreateViews, which is after every popup.
- GameScreen: look up the installed update, plugins and what the firmware
can play on a worker thread. They ran on every RecreateViews, which
happens as each piece of game info arrives.
- LaunchFile: wait only for the file type, and for PARAM.SFO just in the
cases that use it.
Also add --start-screen=firmware.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
With offsets of up to 30000, a port net/adhoc/limits listened on could
be one an outgoing connection was using, and the listen failed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
When the game gives up on a join (game mode timed out, or it
disconnected), a join still queued isn't sent anymore, and the rest of
the server's answer to one already sent is dropped: its players, and the
BSSID that came last and connected us. Players joining a group we've
just left (before the server knew) are dropped too, instead of staying
in our peer list.
Replaces the previous fix, which only dropped a late BSSID: on Windows
the join itself was still queued when game mode timed out in
net/adhoc/gamemode, and went out after it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
When entering game mode timed out before the server's group BSSID came
back, the late BSSID was taken for a normal group's, and the state went
to CONNECTED: every later CreateEnterGameMode then got ALREADY_CONNECTED.
At the ad hoc tests' 8x speed the 1 s timeouts in net/adhoc/gamemode are
125 ms of wall time, short enough for that on Windows; at 50x it happens
on macOS too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Fixes#22842
D3D11 init only took the failure path on E_FAIL, so any other failed
HRESULT from D3D11CreateDevice reported success without a device. Also
stop the generic message overwriting the more specific ones.
The build-x86/x64.cmd scripts targeted VS2015, checked misspelled
variables so never caught a failed build, and nothing uses them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
