and might return before setting it to anything else.

For some reason, MSVC was initializing the time_t (second element of the
struct) to -1, not the int fd (the first element), which was initialized to
0. Weird.

The auto-remove param was only needed for fmutex(), which immediately closes
(and thus don't want auto-removal), so create an internal function that's
used by both fmutex() and fmutex_open().

This is needed for *nix where normal (i.e. not Samba-mount) file behavior
requires that we manually unlink() the file, but not until just before closing
it. So we have to keep track of the file name and the 'auto_remove' desire of
the caller. This is not necessary when the file is on a Linux-smbfs, but we
need to work as expected on normal/local *nix file systems too.

This solves atomic lock file removal for *nix, but does not gaurantee lock file
cleanup upon ungraceful process termination (e.g. crash). At least in this
particular case, Windows beats UNIX.

This reverts commit e55183c0.

fmutex_open()'s auto-remove feature doesn't work on normal/local Linux file
systems: unlink() of a file immediately removes its directory entry allowing
subsequent opens of the same file name, defeating the lock.

I had tested fmutex_open() on a Samba mount (from Linux) and it worked as I
wanted (like the Windows version does). Still searching for a good solution:
if a process terminates, gracefully or not, I'd like the file to be removed
at that time and until then, other processes or threads can't create or open
the same lock file (using O_EXCL and file locking to insure this).

Oh, so much copy/paste eliminated!

This appears to be the best we can do when Windows+M (minimize all)
is pressed.  We don't seem to be notified that we should minimize,
only that a different application is being activated.

This also makes Alt-TAB behave a bit more like expected.

This condition caused aborting BBS operations via Ctrl-C to fail when
connecting from a traditional Telnet client using binary transmission
mode (e.g. telnet -8) or SyncTERM v1.2rc6.

The check of rio_abortable was probably enough to disable this function
during file transfers (e.g. we've had no issues with SSH file transfers
that might include byte 0x03 received from the client), but I added the
check for SS_FILEXFER system status check too, perhaps unnecessary.

On *nix, you need the command shell to do wildcard expansion of filenames

Add the errno value to help root-cause.

We really should be using safe_strerror() everywhere.

This appears to be the only way to 100% provide mutual exclusivity between
Windows and Linux client sharing a Samba file system.

Also, make fmutex_close() take a pointer to the file descriptor so that we can
better manage the resource (i.e. don't close a descriptor that's already been
closed and possibly valid in another thread).

... which simplifies the fmutex_close() function and signature.

By default, the duration is infinite, but some sysops may not want this
file to be ever-grower but rather periodically pruned by the new trashman
utility.

Sysop that don't use spambait.cfg won't benefit from this change.

Also, don't get hostname unless we're surely going to write to the file.

The 'atomic_remove' argument isn't used in non-Windows builds since we can
remove() before close() on non-Windows OSes already.

O_EXCL doesn't appear to be working very reliably over Samba, so let's add
file locking to the equation.

glob() results might be out of date by the time we look at each file, just
skip missing QWK pack semaphore files silently. This should resolve the
".now already gone" messages too. Hopefully.

Removed some extraneous quotes around logged event command-lines.

Also when detecting a blocked event_thread, log the event_code that is/was
running last.

This removes unnecessary some cruft and complexity, most especially from the
event_thread().

Using sbbs_t::fremove() instead of remove(). This probably resolves some
Coverity issues where we weren't checking the return value.

ENOENT = "No such file or directory"

Passing 'true' for the optional 'log_all_errors' parameter overrides this
(logs all errors, including ENOENT).

Add log message when spawning background timed event

Only log messages about QWK mutex files existing when > 60 seconds old

First, fix the logged username bug in the "Lock exists" log message.
Next, double check that the trigger file (sem file or REP file) still exists
after acquiring the file mutex.

Use a temp varible (fname) to make the code easier to read.

Vertrauen's FTP server gets abused by QWKnet logins sometimes and handling
the race conditions around QWK packet creation attempts is silly - there's
no legit reason why a QWKnet account needs to be logged-in multiple times
concurrently to the hub's FTP server, so reject the subsequent logins even
when they're on different hosts (as is the case with Vertrauen).

As part of this change:
- fmutex() now takes an new time_t* argument to (optionally) store the
  time of the mutex file for helping logging (locked since when?).
- time_as_hhmm() created to format a string as either HH:MM or HH:MM[a|p]
  (depending on system configuration for 12 or 24 hour time formatting).
- renamed the old hhmmtostr ()to tm_as_hhmm() (since it takes a struct tm arg)
  and have it return a non-padded string (useful in more situations without
  requiring truncation) when the sysop prefers 24-hour time.

Pretty remedial option parsing here, the order is significant