The 'status' field of send_socket_request is always either
STATUS_PENDING or STATUS_DEVICE_NOT_READY, and the 'total' field is
always zero.
Replace the 'status' field with 'force_async' boolean field, and get rid
of the 'total' field entirely.
Also, clean up the send_socket handler code a bit.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
If the type of the socket is SOCK_DGRAM, it shall always be bound to an
address if we ever attempt to send datagrams through the socket, whether
the attempt succeeds or not.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Make send_socket alert the async immediately if poll() call detects that
there are incoming data in the socket, bypassing the wineserver's main
polling loop.
For sock_transmit, we always mark the async as pending and set the IOSB
(unless async allocation has failed).
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This allows the initial I/O to be performed after the send_socket
handler is called.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This also makes async_handoff() behaviour more consistent with
async_set_result() for handling I/O failures.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Today, async_set_completion_callback() is used to register a function
that is called when the async I/O is completed. It is assumed that the
async will eventually be queued when such callback is registered.
However, this incurs extra complexity in future code that needs the
completion logic to be invoked even if the async is never actually
queued (e.g. when the I/O failed synchronously before async_handoff).
Generalise async completion callback by calling it in async_handoff()
when the I/O status indicates failure.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
The client can set mark_pending to indicate that the full-blown I/O
completion mechanism shall be triggered (asynchronous completion) even
if the status indicates failure.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Shift the resposibility of setting initial status from
set_async_direct_result request handler to async_set_result().
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Commit 15483b1a12 (server: Allow calling async_handoff() with status
code STATUS_ALERTED., 2022-02-10) introduced the set_async_direct_result
handler which calls async_set_initial_status().
However, the async_set_initial_status() call does nothing since
async->terminated is set, leaving the async in a confusing state
(unknown_status = 1 but pending/completed).
So far, this issue is unlikely to have been a problem in practice for
the following reasons:
1. async_set_initial_status() would have unset unknown_status, but it
remains set instead. This is usually not a problem, since
unknown_status is usually ever read by code paths effectively
unreachable for non-device (e.g. socket) asyncs.
It would still potentially allow set_async_direct_result to be called
multiple times, but it wouldn't actually happen in practice unless
something goes wrong.
2. async_set_initial_status() would have set initial_status; however,
it is left with the default value STATUS_PENDING. If the actual
status is something other than that, the handler closes the wait
handle and async_satisfied (the only real consumer of initial_status)
would never be called anyway.
For reasons above, this issue is not effectively observable or testable.
Nonetheless, the current code does leave the async object in an
inconsistent state.
Fix this by removing the !async->terminated check in
async_set_initial_status().
Also, remove assert( async->unknown_status ). The client can now
trigger the assert() by calling set_async_direct_result on a device
async, thereby causing async_set_initial_status() to be called twice.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
The 'status' field of recv_socket_request is always either
STATUS_PENDING or STATUS_DEVICE_NOT_READY, and the 'total' field is
always zero.
Replace the 'status' field with 'force_async' boolean field, and get rid
of the 'total' field entirely.
Also, clean up the recv_socket handler code a bit.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Make recv_socket alert the async immediately if poll() call detects that
there are incoming data in the socket, bypassing the wineserver's main
polling loop.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
If the server detects that an I/O request could be completed immediately
(e.g. the socket to read from already has incoming data), it can now
return STATUS_ALERTED to allow opportunistic synchronous I/O. The Unix
side will then attempt to perform I/O in nonblocking mode and report
back the I/O status to the server via the new server request
"set_async_direct_result". If the operation returns e.g. EAGAIN
or EWOULDBLOCK, the client can opt to either abandon the request (by
specifying an error status) or poll for it in the server as usual (by
waiting on the wait handle).
Without such mechanism in place, the client cannot safely perform
immediately satiable I/O operations synchronously, since it can
potentially conflict with other pending I/O operations that have already
been queued.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Fixes a regression introduced by
5c009c17b3.
For IPV6 socket already bound with bind() with IPV6_V6ONLY set to 0
the address returned by getsockname() may change after it is connected
to a IPV4 peer (e. g., from :: to ::ffff:192.0.2.128).
The effect of blamed commit is that before that Unix getsockname()
was called for each ws2_32 getsockname() and after the commit the
socket name is cached at bind and connect.
Signed-off-by: Paul Gofman <pgofman@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
So that the client allocates a larger receive buffer when needed and not
trigger the assert below when setting the reply message data.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Ideally we should be using sock_get_poll_events() and sock_poll_event() instead,
but that seems like an overly risky change for this late in code freeze.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=51442
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Otherwise, async_waiting() returns 0, leading the socket object to
believe that the previous async request has not yet been acknowledged.
This results in I/O hang for subsequent reads (until shutdown).
Also, async_destroy() calls async_reselect() only after removing the
async request from the queue. Make async_set_result() consistent with
this behaviour.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52332
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
We especially may want to support x86 on ARM, where long double is not an 80-bit
type.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Some PE executables (like mingw's gdb port) just do something like:
- WaitForSingleObject(GetStdHandle(STD_INPUT_HANDLE), INFINITE) and hang
for ever (the read operations are done *after* the wait operation
succeeds)
(of course, the real wait operation is more complex, but the problematic
part boils down to that)
This hangs for ever because conhost's main input thread hasn't been started
yet.
Signed-off-by: Eric Pouech <eric.pouech@gmail.com>
Signed-off-by: Jacek Caban <jacek@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Today, Wine uses NtQueryInformationProcess/ProcessDebugPort to detect
whether the current process is being debugged. If it is, the process
issues a breakpoint to yield control to the debugger.
Some debuggers (e.g. latest CDB) appear to create debug handles with
restricted DACL, which causes querying debug port to fail with
STATUS_ACCESS_DENIED. This results in the debuggee erroneously
skipping the initial breakpoint.
Fix this by not requiring DEBUG_ALL_ACCESS when opening the debug port
object. Instead, use MAXIMUM_ALLOWED for the access mask.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52184
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
FreeBSD 14-CURRENT got the non-standard sched_setaffinity() function, but
it uses FreeBSD's cpuset_t type instead of Linux's cpu_set_t (also
non-standard). This breaks the build. We have to define _WITH_CPU_SET_T
before including <sched.h> to get that type named cpu_set_t instead.
Signed-off-by: Damjan Jovanovic <damjan.jov@gmail.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Makes Cygwin mintty.exe or script.exe show output, if the
stack issues got worked around.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=47808
Signed-off-by: Bernhard Übelacker <bernhardu@mailbox.org>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
\??\GLOBALROOT is a well-known NT symbolic link that allows applications
to access the NT object manager's root namespace via Win32 APIs.
Signed-off-by: Jinoh Kang <jinoh.kang.kr@gmail.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Explicitly include <sys/types.h> to make uid_t available to users of
server/security.h.
This is a problem on musl, and possibly other non-glibc systems.
Signed-off-by: Ismael Luceno <ismael@iodev.co.uk>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Instead of waiting for 1s before considering it dead, when most of the
time the process dies quickly, delaying desktop or wineserver shutdown.
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
ALT and F10 key generate WM_SYSKEYDOWN message.
They should also have the ability to queue hotkey message.
Signed-off-by: Ziqing Hui <zhui@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Resetting their length to make sure we pass alignment checks both for
their current values and for the new values we'll write.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=46472
Signed-off-by: Rémi Bernon <rbernon@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Only really an optimization (plus it makes the code a little conceptually simpler).
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Based on a patch by Chip Davis.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=30155
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Although there is arguably an advantage to saving a server request, the impetus
for this patch is make it easier for the server to process the IRP return status
before (or at the same time as) it processes the IOSB status. This allows
simpler handling of the case where the IRP handler returns STATUS_PENDING but
completes the IRP before returning.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Instead of manually specifying success or failure.
Based on test_return_status() in ntoskrnl. The changes in this patch don't
affect device IRPs, but the tests show the heuristic that Windows uses, and in
practice it turns out to be correct for all known asyncs.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Mostly just to simplify the interface, so that we don't need to use the return
value to communicate this.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
In particular, don't mark asyncs with output data as "alerted", i.e.
restartable.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
For convenience, and to centralize the STATUS_ALERTED logic into one place.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
We still don't quite handle this completely correctly, but we should at least
avoid crashing the server due to a failed assertion.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
At best this is a no-op. At worst this will emit an extra APC_ASYNC_IO where
the termination (e.g. cancellation) should already have emitted one.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
If the async has not been terminated yet, async->fd cannot be NULL.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This was initially used to let the client wait on IRP completion;
c56c42ff79 changed it to use the async handle
instead.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Synthesized formats must be removed too if the format they depend on has
been removed.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=51496
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Don't use rd_shutdown and wr_shutdown to determine this. On the one hand, it's
possible to have pending asyncs even if rd_shutdown && wr_shutdown, which will
be cheerfully completed upon receiving data. On the other hand, RST doesn't
cause WSAESHUTDOWN, but rather WSAECONNRESET.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Based on a patch by Torge Matthies.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=51319
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This patch does result in one functional change: if we are selecting for
AFD_POLL_READ on a socket which has had SD_RECEIVE and there are no asyncs, we
will now respond to POLLIN instead of ignoring it. Neither this nor the previous
behaviour matches Windows, which instead puts the socket into an aborted state
and sends RST to the peer if any data is received after SD_RECEIVE or if
SD_RECEIVE is done while there is pending data.
Apart from this there is no functional change, as the places where rd_shutdown
alone is checked can't be reached if there was a hangup. It is instead for
semantic clarity.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This was used to implement AcceptEx() using multiple APCs, and made obsolete by
0bbd3f6617.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This fixes a regression introduced by 97afac469f.
If we make a request on an asynchronous device handle, and the IRP handler
returns STATUS_PENDING, wait_async() will return STATUS_PENDING, as intended.
However, if the async object is signaled before the user has a chance to call
wait_async() [e.g. if get_next_device_request is called quickly enough], select
will return STATUS_PENDING immediately, which causes server_select() to think
the object is not signaled, and wait for a select reply forever.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=51277
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=51295
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Several server objects check if the last handle is being closed in their
close_handle callback. If a process holds the last two handles to an object,
this code path currently won't be triggered.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This semantically reverts 481517178f.
That commit was used to implement NtFlushBuffersFile, which at the time didn't
use a callback function. 9050b58f07 changed it to
use irp_completion(), since the result of a blocking flush needed to be taken
from the IOSB.
As of 97afac469f that's not true anymore, but on
the other hand it is theoretically possible for a device driver to touch the
Information member of the IOSB, and we don't seem to lose anything by making
all asyncs take a common path.
Since all asyncs pass user data and there's no clear reason for them not to,
let's get rid of a bit of extra code complexity that's no longer used.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This fixes a server crash that can be triggered by deleting a key and then
trying to retrieve its name. In that case key->parent is NULL.
Signed-off-by: Zebediah Figura <z.figura12@gmail.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
A deleted key returns STATUS_KEY_DELETED when ObjectNameInformation is
requested, but succeeds when ObjectBasicInformation is requested.
Signed-off-by: Zebediah Figura <z.figura12@gmail.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
