Fixes#1335.
In file included from ./irp.h:27:
./chansrv_fuse.h:39:5: error: unknown type name 'time_t'
time_t atime; /* Time of last access. */
^
./chansrv_fuse.h:40:5: error: unknown type name 'time_t'
time_t mtime; /* Time of last modification. */
^
./chansrv_fuse.h:41:5: error: unknown type name 'time_t'
time_t ctime; /* Time of last status change. */
^
3 errors generated.
*** Error code 1
- Replace xfuse_cb_enum_dir() directory callback for adding files with
more general xfuse_devredir_add_file_or_dir() to be called from a
directory or a lookup operation.
- Moved XRDP_INODE out of public interface for chansrv_fuse, and replaced
with simpler struct file_attr to pass to
xfuse_devredir_add_file_or_dir()
- Allow a struct file_attr to be placed in an IRP for assembly of file
attributes over multiple IRP_MJ_QUERY_INFORMATION requats.
- Add dev_redir_lookup_entry() to public interface for devredir.c
- Add xfuse_devredir_cb_lookup_entry() callback to public interface for
chansrv-fuse.c
- Remove unused 'is_synced' member from struct xrdp_inode
- Move prototype for xfuse_devredir_cb_write_file() to correct file
- Add const correctness to dev_redir_strings_ends_with() function
- Add const correctness to fuse_reverse_pathname() function
- Moved devredir_proc_cid_* functions out of devredir.h and made static
- Added XFUSE_DUMP_ADDED_ENTRIES maro for debugging
- Removed duplicate code path in xfuse_remove_dir_or_file()
- Removed duplicate code path in xfuse_cb_rename()
- Removed duplicate code path in xfuse_create_dir_or_file()
- Removed duplicate code path in xfuse_cb_open()
- Removed duplicate code path in xfuse_proc_opendir_req()
- Remove unused USE_SYNC_FLAG macro
- Remove unused members invoke_fuse, off, dirbuf1 from XFUSE_INFO
- Clear f_fifo_opendir entries on deinit
- Added some comments and fixed some others
as it is already deprecated. x11rdp is complicated to build and very few
people using it actually. However, some people still select "X11rdp"
session and get stuck despite not installing x11rdp.
https://github.com/neutrinolabs/xrdp/issues/962#issuecomment-430545526
People who really want to use x11rdp should revert this commit.
remove not used chansrv <-> xrdp messages
move static channel disable control into libxrdp
remove some blocking read, write chansrv calls
add drdynvc calls to libxrdp
add drdynvc calls to chansrv
channel cleanup
Unless X server failures are caught, these can cause a premature
exit of chansrv, giving it no chance to clean up. This is currently a
particular problem for fuser mounts.
As the Debian patch[1] expresses, spitting messages on the console when
a process starts in background is a bad idea. Everything should be
written to log file and daemon should start silently. This is a first
step to shut up daemons.
Got some idea from Debian Remote Maintainers and Thorsten Glaser,
thanks!
[1] 2751ad4d62/debian/patches/shutup-daemon.diff
Pull request #650 is not valid to avoid run session twice.
It certainly stops running session twice but causes #1016.
In FreeBSD, sesman process will run like this. The intermediate
sesman is needed to detect session termination correctly.
xrdp-sesman (daemon)
|
+- xrdp-sesman (FreeBSD specific intermediate sesman)
|
+- xrdp-sesman (bsd sesion leader & each session)
|
+- Xorg
+- startwm.sh
+- xrdp-chansrv
To stop runninng session twice correctly, just exit before the
intermediate sesman executes Xorg, WM and chansrv.
* Initialise inode table in `xfuse_create_share()` if necessary
* Add guard to `xfuse_init_xrdp_fs()` to prevent double initialisation of the inode table
Test case: On a system running xrdp with no sessions running run:
xrdp-sesadmin -u=<user> -p=<password> -c=list
Expected result: "No sessions." (ignoring debug output)
Observed result: "Error getting session list."
In the SCP_SERVER_STATE_MNG_LISTREQ case in scp_v1_mng_process() if
there are no sessions it ends the scp session, which causes an error in
the client. In commit 0017081d the client was changed to report errors,
giving the result above.
Fix by calling scp_v1s_mng_list_sessions() from scp_v1_mng_process()
even when there are no sessions, and if so sending a packet with a count
of zero so that the client gets what it expects.
as it was undocumented and few people know reconnectwm.sh is executed on
client reconnect. The behaviour of startwm.sh / reconnectwm.sh should
be documented. This is a first step of documenting them.
