This adds a simple cp2130 userspace driver. Its probably unusable in the
real world but its a great base to build upon.
Signed-off-by: Anton Lundin <glance@acc.umu.se>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This rewrites the custom serial code to use the new api which I
implemented in the Subsurface-branch of libdivecomputer.
This is a bit to big patch but I haven't had the time to break it down
into more sensible patches.
This rewrite enables us to support more ftdi based divecomputer
communication and is tested with both a OSTC3, OSTC2N and a Suunto
Vyper, all over the libftdi driver.
The bluetooth code paths are tested to, and should work as before.
Signed-off-by: Anton Lundin <glance@acc.umu.se>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We used to clear the 'dc_tank_t' for each dive, but then only clear the
volume field in between each cylinder. That means that if the
libdivecomputer back-end does not touch a field, it might contain the
stale value from the previous tank information.
I'm not sure this is actually much of an issue, since I'd expect
back-ends do seem to initialize the fields fully (at least the EON Steel
back-end does). But it's inconsistent.
Also, the code was actually buggy because of the odd indentation: it
would only ask for new tank information up to 'ntanks' tanks, but
because of the final fixup that was done outside of the conditional, it
would actually update the cylinder begin/end pressure data *beyond*
'ntanks', and just re-use the last libdivecomputer data for the rest of
the cylinders.
Again, in practice, that probably never really happened, but it is a
real bug.
The fixed-up code actually looks better too, imho, and is one line
shorter because of the initialization now being done in one place rather
than two.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
In subsurface, we only have one cylinder pressure per sample (well,
technically two: we have a separate o2 pressure for rebreather diving).
Which makes things "interesting" if the dive computer can actually have
multiple pressure sensors, and can report them all concurrently. Like
the Suunto EON Steel.
We used to just take the last one (each sensor reading would just
overwrite any previous ones), and this quick hack just changes the logic
to prefer the "current" cylinder instead.
It's wrong, and it's stupid, but it's the best we can do without major
surgery.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Instead of having fixed numbers and trying to translate them into
strings, a dive computer could just give us the string directly. Like
the new EON Steel backend does.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
This uses the extended tank type information to fill in the cylinder use
(OC gas, CC Diluent or CC O2) from libdivecomputer when available.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
We have two different models for setting the deviceid associated with a
dive computer: either take the value from the libdivecomputer 'devinfo'
field (from the DC_EVENT_DEVINFO event), or generate the device ID by
just hashing the serial number string.
The one thing we do *not* want to have, is to use both methods, so that
the same device generates different device IDs. Because then we'll
think we have two different dive computers even though they are one and
the same.
Usually, this is not an issue, because libdivecomputer either sends the
DEVINFO event or gives us the serial number string, and we'll always
just pick one or the other.
However, in the case of at least the Suunto EON Steel, I intentionally
did *not* send the DC_EVENT_DEVINFO event, because it gives no useful
information. We used the serial number string to generate a device ID,
and everything was fine.
However, in commit d40cdb4755ee ("Add the devinfo event") in the
libdivecomputer tree, Jeff started generating those DC_EVENT_DEVINFO
events for the EON Steel too, and suddenly subsurface would start using
a device ID based on that instead.
The situation is inherently ambiguous - for the EON Steel, we want to
use the hash of the serial number (because that is what we've
historically done), but other dive computers might want to use the
DEVINFO data (because that is what _those_ backends have historically
done, even if they might also implement the new serial string model).
This commit makes subsurface resolve this ambiguity by simply preferring
whatever previous device ID it has associated with that particular
serial number string. If you have no previous device IDs, it doesn't
matter which one you pick.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
There are cases where we actually want to keep them, as exemplified by
this situation from Richard Yorke:
"I have just come across a situation when ignoring the surface marker
is a disadvantage. I have just had a problem with my BC feed
seeping, slowly filling my BC and as I control my buoyancy on the
bottom using the air in my drysuit, I did not notice, so that when I
came to ascend the expanding air in my BC caused a loss of control.
Fortunately not from a great depth and no untoward consequences.
However, the Subsurface profile only shows me rising to 4m and
descending to 5.5m for my safety stop. However I actually broke the
surface and descented to 5.5 but the frequency of recording depth was
not fast enough to show this as it was so brief"
so remove the code that ignores the surface events entirely.
I think we'll have to come up with some smarter filtering model for
showing them, but that is predicated on getting these events to come
through in the first place.
Reported-by: Richard Yorke <yorke.richard@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
Having subsurface-core as a directory name really messes with
autocomplete and is obviously redundant. Simmilarly, qt-mobile caused an
autocomplete conflict and also was inconsistent with the desktop-widget
name for the directory containing the "other" UI.
And while cleaning up the resulting change in the path name for include
files, I decided to clean up those even more to make them consistent
overall.
This could have been handled in more commits, but since this requires a
make clean before the build, it seemed more sensible to do it all in one.
Signed-off-by: Dirk Hohndel <dirk@hohndel.org>
