| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
Last parameter is a vector of 2 ints, not a pointer to int
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
| |
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
| |
Use macro AF_UNIX_CONNECT_TIMEOUT instead of numerical constant
for connect timeout.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
| |
This fsctl might come in handy at one point...
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
| |
All of these functions just call recvfrom/sendmsg which are still TODO
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
| |
* Use correct cygwait/WFSO invocation to not die on cancel and signals
uncontrolled.
* Manage io handles under io_lock.
* Copy peer address to user space under SEH to avoid a resource leak.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Using TerminateThread potentially leaks resources. In our case,
the connect wait thread may be forcefully terminated after
having successfully opened a client side pipe handle. If this
occurs, we have a stale pipe server instance, so the pipe will
never be closed as long as the process lives.
Avoid this by changing the npfs handle to non-blocking, so we can
wait on a termination event object from inside the thread itself
and cleanly exit from the thread instead of terminating.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
| |
Depending on the exact circumstances, some erros are indicated
by different status codes. Add helper macros to handle them
together.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
| |
The handle to the device is never needed. As the name impies,
FSCTL_PIPE_WAIT works on the file system, not on the device level.
Drop opening the device and make sure to open only one handle to NPFS.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
| |
Minimize overhead in creating a nameless event object.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
|
|
| |
* Implement helper functions
* Improve bind
* Implement setting blocking, ioctl(FIONBIO), fcntl(F_SETFL)
* Implement close_on_exec and fixup_after_fork
* Allow overriding sun_path and peer_sun_path
* Improve comments
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
| |
Only store and manipulate pipe basename.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
| |
Needed it new AF_UNIX socket code
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
| |
Linux and FreeBSD use int as well. In addition, this fixes an Ada
incompatiblity problem on 64-bit targets. See also GCC:
gcc/ada/libgnarl/s-osinte__rtems.ads
Signed-off-by: Sebastian Huber <sebastian.huber@embedded-brains.de>
|
|
|
|
|
|
|
| |
Let systems optionally provide the _CLOCKID_T_ type via
<machine/_types.h>.
Signed-off-by: Sebastian Huber <sebastian.huber@embedded-brains.de>
|
|
|
|
|
|
|
|
|
|
| |
They are only used there anyway and it allows to use the AF_UNIX
macro without tweaking header files. While at it, improve
both constructors. The default constructor now creates the name
of an unnamed socket, the constructor taking parameters carefully
checks its input.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
| |
operation
The connect state was stored in a bitfield which is not safe
to do atomic operations on.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
| |
start_transaction is now defined in ntdll.h and we can't rely
on debug_printf being available
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
|
| |
Locale modifier @cjkwide makes Unicode "ambiguous width" characters
wide. So ambiguous width characters can be enforced to have width 2
even in non-CJK locales. This gives e.g. users of "Powerline symbols"
the opportunity to adjust their width to the desired behaviour (and the
behaviour apparently expected by some tools) without having to set a CJK
locale and without losing consistence of terminal character width with
wcwidth/wcswidth locale width.
|
|
|
|
|
|
| |
The caller already does it anyway.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
| |
The new implementation neglected to release the file descriptor
in case of error.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
| |
...plus lots of helper functions. Add comment to explain how everything
works. This comment will be improved while implementing the yet missing
parts.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
| |
This function is going to be used for transposing sun_path of
abstract sockets. This also adds a transposition of the NUL
character to tfx_chars since NUL-bytes in abstract socket names
are perfectly valid.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
| |
Also remove redundant declaration of RtlInitEmptyUnicodeString
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
| |
|
|
|
|
|
|
|
| |
Add constructors and new/delete operators to make sure sun_name_t
objects are allocated on the cygheap.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
|
| |
* check_reparse_point_target returns a path flag mask, rather than
just 1. Return PATH_SYMLINK | PATH_REP for symlinks and directory
mount points, PATH_SOCKET | PATH_REP for AF_UNIX sockets.
* Define Cygwin AF_UNIX socket reparse tag and GUID in ntdll.h.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
| |
We'll need them elsewhere in future.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
| |
...in preparation of reusing this flag for other types of
reparse points, not only symlinks.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
| |
Define new struct sun_name_t and use throughout internally.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
| |
* Split out cygwin/_ucred.h file
* drop local credentials
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
|
|
| |
* Rearrange includes and drop unneccessary ones.
* Don't pull in cygwin/socket.h into sys/un.h just to get
sa_family_t. Include sys/types.h and use __sa_family_t instead.
* start including Windows headers using the w32api/ path prefix
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
| |
Calling fhandler_socket::fstat from fhandler_socket::fstat recursively
is not a good idea...
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
|
|
|
|
|
|
|
| |
fhandler_cygdrive has a size of 696 bytes on x86_64, while the next
biggest fhandler type, fhandler_pty_master, is 584 bytes. The members
responsible for the size are private to opendir/readdir/closedir usage.
fhandler_disk_file stores private readdir data in DIR->__d_internal
instead. Use equivalent method with fhandler_cygdrive. This drops
the size to 464 bytes.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|
|
|
|
| |
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
|