This chapter will present the process of creating a new driver to support another device.
Since NUT already supports many major power device protocols through several generic drivers (genericups, usbhid-ups, snmp-ups, blazer_* and nutdrv_qx), creation of new drivers has become rare.
So most of the time, this process will be limited to completing one of these generic drivers.
If your UPS only does contact closure readings over an RS-232 serial port, then go straight to the Contact closure hardware chapter for information on adding support. It’s a lot easier to add a few lines to a header file than it is to create a whole new driver.
If your UPS connects to your computer via a USB port, then it most likely appears as a USB HID device (this is the simplest way for the vendor to write a Windows control program for it). What comes next depends on whether the vendor implemented the HID PDC (Power Device Class) specification, or simply used the HID protocol to transport serial data to the UPS microcontroller.
A rough heuristic is to check the length of the HID Descriptor length
(wDescriptorLength in lsusb -v output). If it is less than 200 bytes long,
the UPS probably has a glorified USB-to-serial converter built in. Since the
query strings often start with the letter Q, this family of protocols is
often referred to as Q* in the NUT documentation. See the
Q* UPS chapter for more details.
Otherwise, if the HID Descriptor is longer, you can go to the HID subdrivers chapter. You can probably add support for your device by writing a new subdriver to the existing usbhid-ups driver, which is easier (and more maintainable) than writing an entire new driver.
If your USB UPS does not appear to fall into either of these two categories,
feel free to contact the nut-upsdev mailing list with details of your
device.
Similarly, if your UPS connects to your computer via an SNMP network card, you can probably add support for your device by adding a new subdriver to the existing snmp-ups driver. Instructions are provided in the SNMP subdrivers chapter.
The basic design of drivers is simple. main.c handles most of the work
for you. You don’t have to worry about arguments, config files, or
anything else like that. Your only concern is talking to the hardware
and providing data to the outside world.
Familiarize yourself with the design of skel.c in the drivers directory.
It shows a few examples of the functions that main.c will call to obtain
updated information from the hardware.
This structure tracks several description information about the driver:
status: the driver development status. The following values are allowed:
This information is currently used for the startup banner printing and tests.
Open the port (device_path) and do any low-level things that it may need
to start using that port. If you have to set DTR or RTS on a serial
port, do it here.
Don’t do any sort of hardware detection here, since you may be going into upsdrv_shutdown next.
Try to detect what kind of UPS is out there, if any, assuming that’s possible for your hardware. If there is a way to detect that hardware and it doesn’t appear to be connected, display an error and exit. This is the last time your driver is allowed to bail out.
This is usually a good place to create variables like ups.mfr,
ups.model, ups.serial, and other "one time only" items.
Poll the hardware, and update any variables that you care about
monitoring. Use dstate_setinfo() to store the new values.
Do at most one pass of the variables. You MUST return from this function or upsd will be unable to read data from your driver. main will call this function at regular intervals.
Don’t spent more than a couple of seconds in this function. Typically five (5) seconds is the maximum time allowed before you risk that the server declares the driver stale. If your UPS hardware requires a timeout period of several seconds before it answers, consider returning from this function after sending a command immediately and read the answer the next time it is called.
You must never abort from upsdrv_updateinfo(), even when the UPS doesn’t
seem to be attached anymore. If the connection with the UPS is lost, the
driver should retry to re-establish communication for as long as it is
running. Calling exit() or any of the fatal*() functions is specifically
not allowed anymore.
Do whatever you can to make the UPS power off the load but also return after the power comes back on. You may use a different command that keeps the UPS off if the user has requested that with a configuration setting.
You should attempt the UPS shutdown command even if the UPS detection fails. If the UPS does not shut down the load, then the user is vulnerable to a race if the power comes back on during the shutdown process.
This method should not directly exit() the driver program (neither
should it call fatalx() nor fatal_with_errno() methods). It can
upslogx(LOG_ERR, ...) or upslog_with_errno(LOG_ERR, ...), and then
set_exit_flag(N) if required (-1 for EXIT_FAILURE and -2 for
EXIT_SUCCESS which would be handled in the standard driver loop or
forceshutdown() method of main.c).
To be of any use, you must supply data in ups.status. That is the
minimum needed to let upsmon do its job. Whenever possible, you should
also provide anything else that can be monitored by the driver. Some
obvious things are the manufacturer name and model name, voltage data,
and so on.
If you can’t figure out some value automatically, use the ups.conf
options to let the user tell you. This can be useful when a driver
needs to support many similar hardware models, but can’t probe to see
what is actually attached.
All status data lives in structures that are managed by the dstate functions. All access and modifications must happen through those functions. Any other changes are forbidden, as they will not pushed out as updates to things like upsd.
dstate_setinfo("ups.model", "Mega-Zapper 1500");Many of these functions take format strings, so you can build the new values right there:
dstate_setinfo("ups.model", "Mega-Zapper %d", rating);Some variables have special properties. They can be writable, and some are strings. The ST_FLAG_* values can be used to tell upsd more about what it can do.
dstate_setflags("input.transfer.high", ST_FLAG_RW);UPS status flags like on line (OL) and on battery (OB) live in ups.status. Don’t manipulate this by hand. There are functions which will do this for you.
status_init() -- before doing anything else (clear internal buffers,
etc.)status_get(val) -- optionally check if a status word had been set
since the most-recent status_init()status_set(val) -- add a status word (OB, OL, etc)
status_commit() -- push out the update
Possible values for status_set:
OL -- On line (mains is present) OB -- On battery (mains is not present) LB -- Low battery HB -- High battery RB -- The battery needs to be replaced CHRG -- The battery is charging DISCHRG -- The battery is discharging (inverter is providing load power) BYPASS -- UPS bypass circuit is active -- no battery protection is available CAL -- UPS is currently performing runtime calibration (on battery) OFF -- UPS is offline and is not supplying power to the load OVER -- UPS is overloaded TRIM -- UPS is trimming incoming voltage (called "buck" in some hardware) BOOST -- UPS is boosting incoming voltage FSD -- Forced Shutdown (restricted use, see the note below)
Anything else will not be recognized by the usual clients. Coordinate with the nut-upsdev list before creating something new, since there will be duplication and ugliness otherwise.
FSD by itself following that command by a primary upsmon
process. Drivers must not set that value, apart from specific cases (see
below).
FSD when an imminent shutdown has been
detected. In this case, the "on battery + low battery" condition should not be
met. Otherwise, setting status to OB LB should be preferred.
OL and OB flags are an indication of the input line status only.
CHRG and DISCHRG flags are being replaced with
battery.charger.status. See the NUT command and variable naming scheme for more information.
These work like ups.status, and have three special functions which you must use to manage them.
alarm_init() -- before doing anything else
alarm_set() -- add an alarm word
alarm_commit() -- push the value into ups.alarm
the ALARM flag in ups.status is automatically set whenever you use alarm_set. To remove that flag from ups.status, call alarm_init and alarm_commit without calling alarm_set in the middle.
You should never try to set or unset the ALARM flag manually.
If you use UPS alarms, the call to status_commit() should be after alarm_commit(), otherwise there will be a delay in setting the ALARM flag in ups.status.
There is no official list of alarm words as of this writing, so don’t use these functions until you check with the upsdev list.
Also refer to the NUT daisychain support notes chapter of the user manual and developer guide for information related to alarms handling in daisychain mode.
If you’re not talking to a polled UPS, then you must ensure that it is still out there and is alive before calling dstate_dataok(). Even if nothing is changing, you should still "ping" it or do something else to ensure that it is really available. If the attempts to contact the UPS fail, you must call dstate_datastale() to inform the server and clients.
dstate_dataok()
You must call this if polls are succeeding. A good place to call this is the bottom of upsdrv_updateinfo().
dstate_datastale()
You must call this if your status is unusable. A good technique is to call this before exiting prematurely from upsdrv_updateinfo().
Don’t hide calls to these functions deep inside helper functions. It is very hard to find the origin of staleness warnings, if you call these from various places in your code. Basically, don’t call them from any other function than from within upsdrv_updateinfo(). There is no need to call either of these regularly as was stated in previous versions of this document (that requirement has long gone).
Drivers which use serial port functions should include serial.h and use these functions (and cross-platform data types) whenever possible:
Cross-platform data type to represent a serial-port connection.
Macro value representing an invalid serial-port connection.
This macro evaluates to true if fd currently has a "valid" value
(e.g. represents a connected device). You should invalidate the fd
when you initialize the variable or close the connection, by assigning
fd = ERROR_FD.
This macro evaluates to true if fd does not currently have a
"valid" value.
This opens the port and locks it if possible, using one of fcntl, lockf, or uu_lock depending on what may be available. If something fails, it calls fatal for you. If it succeeds, it always returns the fd that was opened.
This is a non-fatal version of ser_open(), that does not call fatal if something fails.
This sets the speed of the port and also does some basic configuring with tcgetattr and tcsetattr. If you have a special serial configuration (other than 8N1), then this may not be what you want.
The port name is provided again here so failures in tcgetattr() provide a useful error message. This is the only place that will generate a message if someone passes a non-serial port /dev entry to your driver, so it needs the extra detail.
This is a non-fatal version of ser_set_speed(), that does not call fatal if something fails.
These functions can be used to set the modem control lines to provide cable power on the RS232 interface. Use state = 0 to set the line to 0 and any other value to set it to 1.
These functions read the state of the modem control lines. They will return 0 if the line is logic 0 and a non-zero value if the line is logic 1.
This function unlocks the port if possible and closes the fd. You should call this in your upsdrv_cleanup handler.
This attempts to write one character and returns the return value from write. You could call write directly, but using this function allows for future error handling in one place.
If you need to send a formatted buffer with an intercharacter delay, use this function. There are a number of UPS controllers which can’t take commands at the full speed that would normally be possible at a given bit rate. Adding a small delay usually turns a flaky UPS into a solid one.
The return value is the number of characters that was sent to the port, or -1 if something failed.
Like ser_send_pace, but without a delay. Only use this if you’re sure that your UPS can handle characters at the full line rate.
This sends a raw buffer to the fd. It is typically used for binary transmissions. It returns the results of the call to write.
This is just ser_send_buf with an intercharacter delay.
This will wait up to d_sec seconds + d_usec microseconds for one character to arrive, storing it at ch. It returns 1 on success, -1 if something fails and 0 on a timeout.
the delay value must not be too large, or your driver will not get back to the usual idle loop in main in time to answer the PINGs from upsd. That will cause an oscillation between staleness and normal behavior.
Like ser_get_char, but this one reads up to buflen bytes storing all of them in buf. The buffer is zeroed regardless of success or failure. It returns the number of bytes read, -1 on failure and 0 on a timeout.
This is essentially a single read() function with a timeout.
Like ser_get_buf, but this one waits for buflen bytes to arrive, storing all of them in buf. The buffer is zeroed regardless of success or failure. It returns the number of bytes read, -1 on failure and 0 on a timeout.
This should only be used for binary reads. See ser_get_line for protocols that are terminated by characters like CR or LF.
This is the reading function you should use if your UPS tends to send responses like "OK\r" or "1234\n". It reads up to buflen bytes and stores them in buf, but it will return immediately if it encounters endchar. The endchar will not be stored in the buffer. It will also return if it manages to collect a full buffer before reaching the endchar. It returns the number of bytes stored in the buffer, -1 on failure and 0 on a timeout.
If the character matches the ignset with strchr(), it will not be added
to the buffer. If you don’t need to ignore any characters, just pass it
an empty string — "".
The buffer is always cleared and is always null-terminated. It does
this by reading at most (buflen - 1) bytes.
any other data which is read after the endchar in the serial buffer will be lost forever. As a result, you should not use this unless your UPS uses a polled protocol.
Let’s say your endchar is \n and your UPS sends "OK\n1234\nabcd\n".
This function will read() all of that, find the first \n, and stop
there. Your driver will get "OK", and the rest is gone forever.
This also means that you should not "pipeline" commands to the UPS. Send a query, then read the response, then send the next query.
This is just like ser_get_line, but it allows you to specify a set of alert characters which may be received at any time. They are not added to the buffer, and this function will call your handler function, passing the character as an argument.
Implementation note: this function actually does all of the work, and ser_get_line is just a wrapper that sets an empty alertset and a NULL handler.
This function will drain the input buffer. If verbose is set to a positive number, then it will announce the characters which have been read in the syslog. You should not set verbose unless debugging is enabled, since it could be very noisy.
This function returns the number of characters which were read, so you can check for extra bytes by looking for a nonzero return value. Zero will also be returned if the read fails for some reason.
This function drains both the in- and output buffers. Return zero on success.
Call this whenever your serial communications fail for some reason. It takes a format string, so you can use variables and other things to clarify the error. This function does built-in rate-limiting so you can’t spam the syslog.
By default, it will write 10 messages, then it will stop and only write 1 in 100. This allows the driver to keep calling this function while the problem persists without filling the logs too quickly.
In the old days, drivers would report a failure once, and then would be silent until things were fixed again. Users had to figure out what was happening by finding that single error message, or by looking at the repeated complaints from upsd or the clients.
If your UPS frequently fails to acknowledge polls and this is a known situation, you should make a couple of attempts before calling this function.
this does not call dstate_datastale. You still need to do that.
This will clear the error counter and write a "re-established" message to the syslog after communications have been lost. Your driver should call this whenever it has successfully contacted the UPS. A good place for most drivers is where it calls dstate_dataok.
Drivers which use USB functions should include usb-common.h and use these:
You should use the usb_device_id_t structure, and the USB_DEVICE macro to declare the supported devices. This allows the automatic extraction of USB information, to generate the Hotplug, udev and UPower support files.
The structure allows to convey uint16_t values of VendorID and ProductID,
and an optional matching-function callback to interrogate the device in
more detail (constrain known supported firmware versions, OEM brands, etc.)
For example:
/* SomeVendor name */
#define SOMEVENDOR_VENDORID 0xXXXX
/* USB IDs device table */
static usb_device_id_t sv_usb_device_table [] = {
/* some models 1 */
{ USB_DEVICE(SOMEVENDOR_VENDORID, 0xYYYY), NULL },
/* various models */
{ USB_DEVICE(SOMEVENDOR_VENDORID, 0xZZZZ), NULL },
{ USB_DEVICE(SOMEVENDOR_VENDORID, 0xAAAA), NULL },
/* Terminating entry */
{ 0, 0, NULL }
};Call this in your device opening / matching function. Pass your usb_device_id_t list structure, and a set of VendorID, DeviceID, as well as Vendor, Product and Serial strings, possibly also Bus, bcdDevice (device release number) and Device name on the bus strings, in the USBDevice_t fields describing the specific piece of hardware you are inspecting.
This function returns one of the following value:
For implementation examples, refer to the various USB drivers, and search for the above patterns.
This set of USB helpers is due to expand is the near future…
PLEASE don’t make up new variables and commands just because you can. The new dstate functions give us the power to create just about anything, but that is a privilege and not a right. Imagine the mess that would happen if every developer decided on their own way to represent a common status element.
Check the NUT command and variable naming scheme section first to find the closest fit. If nothing matches, contact the upsdev list, and we’ll figure it out.
Patches which introduce unlisted names may be modified or dropped.
upsd can call drivers to store values in read/write variables and to kick off instant commands. This is how you register handlers for those events.
The driver core (drivers/main.c) has a structure called upsh. You should populate it with function pointers in your upsdrv_initinfo() function. Right now, there are only two possibilities:
If your driver’s function for handling variable set events is called my_ups_set(), then you’d do this to add the pointer:
upsh.setvar = my_ups_set;
my_ups_set() will receive two parameters:
const char * -- the variable being changed const char * -- the new value
You should return either STAT_SET_HANDLED if your driver recognizes the command, or STAT_SET_UNKNOWN if it doesn’t. Other possibilities will be added at some point in the future.
This works just like the set process, with slightly different values arriving from the server.
upsh.instcmd = my_ups_cmd;
Your function will receive two args:
const char * -- the command name const char * -- (reserved)
You should return either STAT_INSTCMD_HANDLED or STAT_INSTCMD_UNKNOWN depending on whether your driver can handle the requested command.
Use strcasecmp. The command names arriving from upsd should be treated without regards to case.
If you have a variable that can have several specific values, it is enumerated. You should add each one to make it available to the client:
dstate_addenum("input.transfer.low", "92");
dstate_addenum("input.transfer.low", "95");
dstate_addenum("input.transfer.low", "99");
dstate_addenum("input.transfer.low", "105");If you have a variable that support values comprised in one or more ranges, you should add each one to make it available to the client:
dstate_addrange("input.transfer.low", 90, 95);
dstate_addrange("input.transfer.low", 100, 105);Strings that may be changed by the client should have the ST_FLAG_STRING flag set, and a maximum length (in bytes) set in the auxdata.
dstate_setinfo("ups.id", "Big UPS");
dstate_setflags("ups.id", ST_FLAG_STRING | ST_FLAG_RW);
dstate_setaux("ups.id", 8);If the variable is not writable, don’t bother with the flags or the auxiliary data. It won’t be used.
If your hardware and driver can support a command, register it.
dstate_addcmd("load.on");The new ser_* functions may perform reads faster than the UPS is able to respond in some cases. This means that your driver will call select() and read() numerous times if your UPS responds in bursts. This also depends on how fast your system is.
You should check your driver with strace or its equivalent on your
system. If the driver is calling read() multiple times, consider adding
a call to usleep before going into the ser_read_* call. That will give
it a chance to accumulate so you get the whole thing with one call to
read without looping back for more.
This is not a request to save CPU time, even though it may do that. The important part here is making the strace/ktrace output easier to read.
write(4, "Q1\r", 3) = 3
nanosleep({0, 300000000}, NULL) = 0
select(5, [4], NULL, NULL, {3, 0}) = 1 (in [4], left {3, 0})
read(4, "(120.0 084.0 120.0 0 60.0 22.6"..., 64) = 47Without that delay, that turns into a mess of selects and reads. The select returns almost instantly, and read gets a tiny chunk of the data. Add the delay and you get a nice four-line status poll.
If your UPS uses "\n" and/or "\r" as endchar, consider the use of Canonical Input Mode Processing instead of the ser_get_line* functions.
Using a serial port in this mode means that select() will wait until a full line is received (or times out). This relieves you from waiting between sending a command and reading the reply. Another benefit is, that you no longer have to worry about the case that your UPS sends "OK\n1234\nabcd\n". This will be broken up cleanly in "OK\n", "1234\n" and "abcd\n" on consecutive reads, without risk of losing data (which is an often forgotten side effect of the ser_get_line* functions).
Currently, an example how this works can be found in the safenet and upscode2 drivers. The first uses a single "\r" as endchar, while the latter accepts either "\n", "\n\r" or "\r\n" as line termination. You can define other termination characters as well, but can’t undefine "\r" and "\n" (so if you need these as data, this is not for you).
In order to build your new driver, it needs to be added to
drivers/Makefile.am. At the moment, there are several driver list variables
corresponding to the general protocol of the driver (SERIAL_DRIVERLIST,
SNMP_DRIVERLIST, etc.). If your driver does not fit into one of these
categories, please discuss it on the nut-upsdev mailing list.
There are also *_SOURCES and optional *_LDADD variables to list the source
files, and any additional linker flags. If your driver uses the C math
library, be sure to add -lm, since this flag is not always included by
default on embedded systems.
When you add a driver to one of these lists, pay attention to the backslash
continuation characters (\\) at the end of the lines.
The automake program converts the Makefile.am files into Makefile.in
files to be processed by ./configure. See the discussion in Section 3.16, “Building the Code”
about automating the rebuild process for these files.
This is a collection of notes that apply to contact closure UPS hardware, specifically those monitored by the genericups driver.
"Contact closure" refers to a situation where one line is connected to another inside UPS hardware to indicate some sort of situation. These can be relays, or some other form of switching electronics. The generic idea is that you either have a signal on a line, or you don’t. Think binary.
Usually, the source for a signal is the host PC. It provides a high (logic level 1) from one of its outgoing lines, and the UPS returns it on one or more lines to communicate. The rest of the time, the UPS either lets it float or connects it to the ground to indicate a 0.
Other equipment generates the high and low signals internally, and does not require cable power. These signals just appear on the right lines without any special configuration on the PC side.
Some evil cabling and UPS equipment uses the transmit or receive lines as their reference points for these signals. This is not sufficient to register as a high signal on many serial ports. If you have problems reading certain signals on your system, make sure your UPS isn’t trying to do this.
Unlike their smarter cousins, this kind of UPS can only give you very simple yes/no answers. Due to the limited number of serial port lines that can be used for this purpose, you typically get two pieces of data:
That’s it. Some equipment actually swaps the second one for a notification about whether the battery needs to be replaced, which makes life interesting for those users.
Most hardware also supports an outgoing signal from the PC which means "shut down the load immediately". This is generally implemented in such a way that it only works when running on battery. Most hardware or cabling will ignore the shutdown signal when running on line power.
If none of the existing types in the genericups driver work completely, make a note of which ones (if any) manage to work partially. This can save you some work when creating support for your hardware.
Use that information to create a list of where the signals from your UPS appear on the serial port at the PC end, and whether they are active high or active low. You also need to know what outgoing lines, if any, need to be raised in order to supply power to the contacts. This is known as cable power. Finally, if your UPS can shut down the load, that line must also be identified.
There are only 4 incoming and 2 outgoing lines, so not many combinations are left. The other lines on a typical 9 pin port are transmit, receive, and the ground. Anything trying to do a high/low signal on those three is beyond the scope of the genericups driver. The only exception is an outgoing BREAK, which we already support.
When editing the genericups.h, the values have the following meanings:
Outgoing lines:
Incoming lines:
This may seem a bit confusing to have two variables per value that we want to read, but here’s how it works. If you set line_ol to TIOCM_RNG, then the value of TIOCM_RNG (0x080 on my box) will be anded with the value of the serial port whenever a poll occurs. If that flag exists, then the result of the and will be 0x80. If it does not exist, the result will be 0.
So, if line_ol = foo, then val_ol can only be foo or 0.
As a general case, if line_ol == val_ol, then the value you’re reading is active high. Otherwise, it’s active low. Check out the guts of upsdrv_updateinfo() to see how it really works.
Late in the 1.3 cycle, a feature was merged which allows you to create custom monitoring settings without editing the model table. Just set upstype to something close, then use settings in ups.conf to adjust the rest. See the genericups(8) man page for more details.
USB (Universal Serial Port) devices can be divided into several different classes (audio, imaging, mass storage etc). Almost all UPS devices belong to the "HID" class, which means "Human Interface Device", and also includes things like keyboards and mice. What HID devices have in common is a particular (and very flexible) interface for reading and writing information (such as X/Y coordinates and button states, in the case of a mouse, or voltages and status information, in the case of a UPS).
The NUT "usbhid-ups" driver is a meta-driver that handles all HID UPS devices. It consists of a core driver that handles most of the work of talking to the USB hardware, and several sub-drivers to handle specific UPS manufacturers (MGE, APC, and Belkin are currently supported). Adding support for a new HID UPS device is easy, because it requires only the creation of a new sub-driver.
There are a few USB UPS devices that are not true HID devices. These devices typically implement some version of the manufacturer’s serial protocol over USB (which is a really dumb idea, by the way). An example is the original Tripplite USB interface (USB idProduct = 0001). Its HID descriptor is only 52 bytes long (compared to several hundred bytes for a true PDC HID UPS). Such devices are not supported by the usbhid-ups driver, and are not covered in this document. If you need to add support for such a device, read new-drivers.txt and see the "tripplite_usb" driver for inspiration.
From the point of view of writing a HID subdriver, a HID device consists of a bunch of variables. Some variables (such as the current input voltage) are read-only, whereas other variables (such as the beeper enabled/disabled/muted status) can be read and written. These variables are usually grouped together and arranged in a hierarchical tree shape, similar to directories in a file system. This tree is called the "usage" tree. For example, here is part of the usage tree for a typical APC device. Variable components are separated by ".". Typical values for each variable are also shown for illustrative purposes.
UPS.Battery.Voltage | 11.4 V |
UPS.Battery.ConfigVoltage | 12 V |
UPS.Input.Voltage | 117 V |
UPS.Input.ConfigVoltage | 120 V |
UPS.AudibleAlarmControl | 2 (=enabled) |
UPS.PresentStatus.Charging | 1 (=yes) |
UPS.PresentStatus.Discharging | 0 (=no) |
UPS.PresentStatus.ACPresent | 1 (=yes) |
As you can see, variables that describe the battery status might be grouped together under "Battery", variables that describe the input power might be grouped together under "Input", and variables that describe the current UPS status might be grouped together under "PresentStatus". All of these variables are grouped together under "UPS".
This hierarchical organization of data has the advantage of being very flexible; for example, if some device has more than one battery, then similar information about each battery could be grouped under "Battery1", "Battery2" and so forth. If your UPS can also be used as a toaster, then information about the toaster function might be grouped under "Toaster", rather than "UPS".
However, the disadvantage is that each manufacturer will have their own idea about how the usage tree should be organized, and usbhid-ups needs to know about all of them. This is why manufacturer specific subdrivers are needed.
To make matters more complicated, usage tree components (such as "UPS", "Battery", or "Voltage") are internally represented not as strings, but as numbers (called "usages" in HID terminology). These numbers are defined in the "HID Usage Tables", available from http://www.usb.org/developers/hidpage/. The standard usages for UPS devices are defined in a document called "Usage Tables for HID Power Devices" (the Power Device Class [PDC] specification).
For example:
0x00840010 = UPS 0x00840012 = Battery 0x00840030 = Voltage 0x00840040 = ConfigVoltage 0x0084001a = Input 0x0084005a = AudibleAlarmControl 0x00840002 = PresentStatus 0x00850044 = Charging 0x00850045 = Discharging 0x008500d0 = ACPresent
Thus, the above usage tree is internally represented as:
00840010.00840012.00840030 00840010.00840012.00840040 00840010.0084001a.00840030 00840010.0084001a.00840040 00840010.0084005a 00840010.00840002.00850044 00840010.00840002.00850045 00840010.00840002.008500d0
To make matters worse, most manufacturers define their own additional
usages, even in cases where standard usages could have been used. for
example Belkin defines 00860040 = ConfigVoltage (which is incidentally
a violation of the USB PDC specification, as 00860040 is reserved for
future use).
Thus, subdrivers generally need to provide:
Moreover, subdrivers might have to provide additional functionality, such as custom implementations of specific instant commands (load.off, shutdown.restart), and conversions of manufacturer specific data formats.
The drivers/hidtypes.h header provides a number of macro names
for entries in the standard usage tables for Power Device
USAGE_POW_<SOMETHING> and Battery System USAGE_BAT_<SOMETHING>
data pages.
If NUT codebase would ever need to refresh those macros, here is some background information (based on NUT issue #1189 and PR #1290):
These data were parsed from (a very slightly updated version of) https://github.com/abend0c1/hidrdd/blob/master/rd.conf file, which incorporates the complete USB-IF usage definitions for Power Device and Battery System pages (among many others), so we didn’t have to extract the names and values from the USB-IF standards documents (did check it all by eye though).
The file was processed with the following chain of commands:
:; grep -e '^0084' -e '^0085' rd.conf \
| sed 's/,.*$//;s/ *$//' \
| sed 's/ /_/g;s/_/ /' \
| tr '[:lower:]' '[:upper:]' \
| sed 's/\(0085.... \)/\1USAGE_BAT_/;s/\(0084.... \)/\1USAGE_POW_/;s/\([A-Z_]*\)_PAGE/PAGE_\1/' \
| awk '{print "#define "$2" 0x"$1}'In preparation for writing a subdriver for a device that is currently unsupported, run usbhid-ups with the following command line:
drivers/usbhid-ups -DD -u root -x explore -x vendorid=XXXX \
-x port=auto -s ups(substitute your device’s 4-digit VendorID instead of "XXXX"). This will produce a bunch of debugging information, including a number of lines starting with "Path:" that describe the device’s usage tree. This information forms the initial basis for a new subdriver.
You should save this information to a file, e.g.:
drivers/usbhid-ups -DD -u root -x explore -x vendorid=XXXX \
-x port=auto -s ups -d1 2>&1 | tee /tmp/infoYou can now create an initial "stub" subdriver for your device by using
helper script scripts/subdriver/gen-usbhid-subdriver.sh.
this only creates a driver code "stub" which needs to be further customized to be actually useful (see "Customization" below).
Use the script as follows:
scripts/subdriver/gen-usbhid-subdriver.sh < /tmp/info
where /tmp/info is the file where you previously saved the debugging information.
This script prompts you for a name for the subdriver; use only letters and digits, and use natural capitalization such as "Belkin" (not "belkin" or "BELKIN"). The script may prompt you for additional information.
You should put the generated files into the drivers/ subdirectory, and
update usbhid-ups.c by adding the appropriate #include line and by
updating the definition of subdriver_list in usbhid-ups.c. You must
also add the subdriver to USBHID_UPS_SUBDRIVERS in drivers/Makefile.am
and call autoreconf and/or ./configure from the top-level NUT directory.
You can then recompile usbhid-ups, and start experimenting with the new
subdriver.
You may have a device from vendor (and maybe model) whose support usbhid-ups
already claims. However, you may feel that the driver does not represent all
data points that your device serves. This may be possible, as vendors tend to
use the same identifiers for unrelated products, as well as produce revisions
of devices with same marketed name but different internals (due to chip and
other components availability, cost optimization, etc.) Even without sinister
implications, UPS firmwares evolve and so bugs and features can get added,
fixed and removed over time with truly the same hardware being involved.
In this case you should follow the same instructions as above for "Writing a subdriver", but specify the same subdriver name as the one which supports your device family already.
Then compare the generated source file with the one already committed to NUT
codebase, paying close attention to ..._hid2nut[] table which maps "usage"
names to NUT data points. There may be several "usage" values served by
different device models or firmware versions, that provide same information
for a NUT data point, such as input.voltage. For the hid2nut mapping
tables, first hit wins (so you may e.g. prefer to check values with better
precision first).
Using a GUI tool with partial-line difference matching and highlighting, such as Meld or WinMerge, is recommended for this endeavour.
For new data points in hid2nut tables be sure to not invent new names,
but use standard ones from docs/nut-names.txt file. Temporarily, the
experimental.* namespace may be used.
If you need to standardize a name for some concept not addressed yet,
please do so via nut-upsdev mailing list discussion.
The initially generated subdriver code is only a stub, and will not implement any useful functionality (in particular, it will be unable to shut down the UPS). In the beginning, it simply attempts to monitor some UPS variables. To make this driver useful, you must examine the NUT variables of the form "unmapped.*" in the hid_info_t data structure, and map them to actual NUT variables and instant commands. There are currently no step-by-step instructions for how to do this. Please look at the files to see how the currently implemented subdrivers are written:
To test existing data points (including those not yet translated
to standard NUT mappings conforming to NUT command and variable naming scheme), you can use custom drivers built after you
./configure --with-unmapped-data-points.
Production driver builds must not include any non-standard names.
It is a fact of life that fellow developers make mistakes, and firmware authors do too. In some cases there are inconsistencies about bytes seen on the wire vs. their logical values, such value range and signedness if interpreting them according to standard.
NUT drivers now include a way to detect and fix up known issues in such
flawed USB report descriptors, side-stepping the standard similarly where
deemed needed. A pointer to such hook method is part of the subdriver_t
structure detailing each usbhid-ups subdriver nuances, defaulting to
a fix_report_desc() trivial implementation.
For some practical examples, see e.g. apc_fix_report_desc() method in the
drivers/apc-hid.c file, and cps_fix_report_desc() in drivers/cps-hid.c
file.
Finally note that such fix-ups may be not applicable to all devices or
firmware versions for what they assume their target audience is. If you
suspect that the fix-up method is actually causing problems, you can quickly
disable it with disable_fix_report_desc driver option for usbhid-ups.
If the problem does dissipate, please find a way to identify your "fixed"
hardware/firmware vs. those models where existing fix-up method should be
applied, and post a pull request so the NUT driver would handle both cases.
Beside looking for problems with report descriptor processing in NUT code, it is important to make sure what data the device actually serves on the wire, and if it is logically consistent with the protocol requirements.
While here, keep in mind that USB protocol on the wire has a specified
order of bytes involved, while processing on your computer may lay them
out differently due to bitness and endianness of the current binary build.
General NUT codebase (libhid.c, hidparser.c) aims to abstract this,
so application code like drivers can deal with their native numeric data
types, but when troubleshooting, do not rule out possibility of flaws
there as well. And certainly do not code any assumptions about ordered
multiple-byte ranges in a protocol buffer.
For a deep dive into the byte stream, you will need additional tools:
Typical troubleshooting of suspected firmware/protocol issues goes like this:
usbhid-ups driver debug verbosity level up to 5 (or more)
and restart the driver, so it would record the HEX dump of report descriptor
Example 1. Example direct use of REXX
Example adapted from https://github.com/networkupstools/nut/issues/2039
Run a NUT usb-hid driver with at least debug verbosity level 3 (-DDD)
to get a report descriptor dump starting with a line like this:
3.670755 [D3] Report Descriptor: (909 bytes) => 05 84 09 04 a1 01 ...
…and copy-paste those reported lines as input into rexx tool, which
would generate a C source file including human-worded description and
a relevant data structure:
:; rexx rd.rex -d --hex 05 84 09 04 a1 01 85 01 09 18 ... 55 b1 02 c0 c0 c0
//--------------------------------------------------------------------------------
// Decoded Application Collection
//--------------------------------------------------------------------------------
/*
05 84 (GLOBAL) USAGE_PAGE 0x0084 Power Device Page
09 04 (LOCAL) USAGE 0x00840004 UPS (Application Collection)
A1 01 (MAIN) COLLECTION 0x01 Application (Usage=0x00840004: Page=Power Device Page, Usage=UPS, Type=Application Collection)
85 01 (GLOBAL) REPORT_ID 0x01 (1)
09 18 (LOCAL) USAGE 0x00840018 Outlet System (Physical Collection)
...
*/
// All structure fields should be byte-aligned...
#pragma pack(push,1)
//--------------------------------------------------------------------------------
// Power Device Page featureReport 01 (Device <-> Host)
//--------------------------------------------------------------------------------
typedef struct
{
uint8_t reportId; // Report ID = 0x01 (1)
// Collection: CA:UPS CP:OutletSystem CP:Outlet
int8_t POW_UPSOutletSystemOutletSwitchable; // Usage 0x0084006C: Switchable, Value = to
int8_t POW_UPSOutletSystemOutletDelayBeforeStartup; // Usage 0x00840056: Delay Before Startup, Value = -1 to 60
int8_t POW_UPSOutletSystemOutletDelayBeforeShutdown; // Usage 0x00840057: Delay Before Shutdown, Value = -1 to 60
int8_t POW_UPSOutletSystemOutletDelayBeforeReboot; // Usage 0x00840055: Delay Before Reboot, Value = -1 to 60
int8_t POW_UPSOutletSystemOutletSwitchable_1; // Usage 0x0084006C: Switchable, Value = -1 to 60
int8_t POW_UPSOutletSystemOutletDelayBeforeStartup_1; // Usage 0x00840056: Delay Before Startup, Value = -1 to 60
int8_t POW_UPSOutletSystemOutletDelayBeforeShutdown_1; // Usage 0x00840057: Delay Before Shutdown, Value = -1 to 60
int8_t POW_UPSOutletSystemOutletDelayBeforeReboot_1; // Usage 0x00840055: Delay Before Reboot, Value = -1 to 60
} featureReport01_t;
#pragma pack(pop)It is desirable to support shutting down the UPS. Usually (for devices that follow the HID Power Device Class specification), this requires sending the UPS two commands. One for shutting down the UPS (with an offdelay) and one for restarting it (with an ondelay), where offdelay < ondelay. The two NUT commands for which this is relevant, are shutdown.return and shutdown.stayoff.
Since the one-to-one mapping above doesn’t allow sending two HID commands to the UPS in response to sending one NUT command to the driver, this is handled by the driver. In order to make this work, you need to define the following four NUT values:
ups.delay.start (variable, R/W) ups.delay.shutdown (variable, R/W) load.off.delay (command) load.on.delay (command)
If the UPS supports it, the following variables can be used to show the countdown to start/shutdown:
ups.timer.start (variable, R/O) ups.timer.shutdown (variable, R/O)
The load.on and load.off commands will be defined implicitly by
the driver (using a delay value of 0). Define these commands
yourself, if your UPS requires a different value to switch on/off
the load without delay.
Note that the driver expects the load.off.delay and load.on.delay
to follow the HID Power Device Class specification, which means that
the load.on.delay command should NOT switch on the load in the
absence of mains power. If your UPS switches on the load regardless of
the mains status, DO NOT define this command. You probably want to
define the shutdown.return and/or shutdown.stayoff commands in
that case. Commands defined in the subdriver will take precedence over
the ones that are composed in the driver.
When running the driver with the -k flag, it will first attempt to
send a shutdown.return command and if that fails, will fallback to
shutdown.reboot.
The SNMP protocol allow for a common way to interact with devices over the network.
The NUT "snmp-ups" driver is a meta-driver that handles many SNMP devices, such as UPS and PDU. It consists of a core driver that handles most of the work of talking to the SNMP agent, and several sub-drivers to handle specific device manufacturers. Adding support for a new SNMP device is easy, because it requires only the creation of a new sub-driver.
From the point of view of writing an SNMP subdriver, an SNMP device consists of a bunch of variables, called OIDs (for Object IDentifiers). Some OIDs (such as the current input voltage) are read-only, whereas others (such as the beeper enabled/disabled/muted status) can be read and written. OID are grouped together and arranged in a hierarchical tree shape, similar to directories in a file system. OID components are separated by ".", and can be expressed in numeric or textual form. For example:
.iso.org.dod.internet.mgmt.mib-2.system.sysObjectID
is equivalent to:
.1.3.6.1.2.1.1.2.0
Here is an excerpt tree, showing only two OIDs, sysDescr and sysObjectID:
.iso
.org
.dod
.internet
.mgmt
.mib-2
.system
.sysDescr.0 = STRING: Dell UPS Tower 1920W HV
.sysObjectID.0 = OID: .iso.org.dod.internet.private.enterprises.674.10902.2
(...)
.upsMIB
.upsObjects
.upsIdent
.upsIdentModel = STRING: "Dell UPS Tower 1920W HV"
(...)
.private
.enterprises
.674
.10902
.2
.100
.1.0 = STRING: "Dell UPS Tower 1920W HV"
(...)As you can see in the above example, the device name is exposed three times, through three different MIBs:
Generic MIB-II (RFC 1213):
.iso.org.dod.internet.mgmt.mib-2.system.sysDescr.0 = STRING: Dell UPS Tower 1920W HV .1.3.6.1.2.1.1.1.0 = STRING: Dell UPS Tower 1920W HV
UPS MIB (RFC 1628):
.iso.org.dod.internet.mgmt.mib-2.upsMIB.upsObjects.upsIdent.upsIdentModel = STRING: "Dell UPS Tower 1920W HV" .1.3.6.1.2.1.33.1.1.2.0 = STRING: "Dell UPS Tower 1920W HV"
DELL SNMP UPS MIB:
.iso.org.dod.internet.private.enterprises.674.10902.2.100.1.0 = STRING: "Dell UPS Tower 1920W HV"
But only the two last can serve useful data for NUT.
An highly interesting OID is sysObjectID: its value is an OID that refers to the main MIB of the device. In the above example, the device points us at the Dell UPS MIB. sysObjectID, also called "sysOID" is used by snmp-ups to find the right mapping structure.
For more information on SNMP, refer to the Wikipedia article, or browse the Internet.
To be able to convert values, NUT SNMP subdrivers need to provide:
Moreover, subdrivers might have to provide additional functionality, such as custom implementations of specific instant commands (load.off, shutdown.restart), and conversions of manufacturer specific data formats. At the time of writing this document, snmp-ups doesn’t provide such mechanisms (only formatting ones), but it is planned in a future release.
In order to create a subdriver, you will need the following:
You can create an initial "stub" subdriver for your device by using the helper script scripts/subdriver/gen-snmp-subdriver.sh. Note that this only creates a "stub" which MUST be customized to be useful (see CUSTOMIZATION below).
You have two options to run gen-snmp-subdriver.sh:
This method requires to have a network access to the device, in order to automatically retrieve the needed information.
You have to specify the following parameters:
For example:
$ gen-snmp-subdriver.sh -H W.X.Y.Z -c foobar -n <MIB name>.mib
This method does not require direct access to the device, at least not for the one using gen-snmp-subdriver.sh.
The following SNMP data need to be dumped first:
a numeric SNMP walk (OIDs in dotted numeric format) of the tree pointed by sysOID. For example:
snmpwalk -On -c foobar W.X.Y.Z .1.3.6.1.4.1.705.1 > snmpwalk-On.log
a textual SNMP walk (OIDs in string format) of the tree pointed by sysOID. For example:
snmpwalk -Os -c foobar W.X.Y.Z .1.3.6.1.4.1.705.1 > snmpwalk-Os.log
if the OID are only partially resolved (i.e, there are still parts expressed in numeric form), then try using -M to point your .mib file.
Then call the script using:
$ gen-snmp-subdriver.sh -s <sysOID value> <numeric SNMP walk> <string SNMP walk>
For example:
$ gen-snmp-subdriver.sh -s .1.3.6.1.4.1.705.1 snmpwalk-On.log snmpwalk-Os.log
This script prompts you for a name for the subdriver if you don’t provide it with -n. Use only letters and digits, and use natural capitalization such as "Camel" (not "camel" or "CAMEL", apart if it natural). The script may prompt you for additional information.
Beside of the mandatory customization, there are a few things that you have to do, as mentioned at the end of the script:
finally call the following, from the top level directory, to test compilation:
$ autoreconf && configure && make
You can already start experimenting with the new subdriver; but all data will be prefixed by "unmapped.". You will now have to customize it.
The initially generated subdriver code is only a stub (mainly a big C structure to be precise), and will not implement any useful functionality (in particular, it will be unable to shut down the UPS). In the beginning, it simply attempts to monitor some UPS variables. To make this driver useful, you must examine the NUT variables of the form "unmapped.*" in the hid_info_t data structure, and map them to actual NUT variables and instant commands. There are currently no step-by-step instructions for how to do this. Please look at the files to see how the currently implemented subdrivers are written:
To help you, above each entry in <LDRIVER>-mib.c, there is a comment that displays the textual OID name. For example, the following entry:
/* upsMIB.upsObjects.upsIdent.upsIdentModel = STRING: "Dell UPS Tower 1920W HV" */
{ "unmapped.upsidentmodel", ST_FLAG_STRING, SU_INFOSIZE,
".1.3.6.1.4.1.2254.2.4.1.1.0", NULL, SU_FLAG_OK, NULL },Many times, only the first field will need to be modified, to map to an actual NUT variable name.
Check the NUT command and variable naming scheme section first to find a name that matches the OID name (closest fit). If nothing matches, contact the upsdev list, and we’ll figure it out.
In the above example, the right NUT variable is obviously "device.model".
The MIB definition file (.mib) also contains some description of these OIDs, along with the possible enumerated values.
To test existing data points (including those not yet translated
to standard NUT mappings conforming to NUT command and variable naming scheme), you can use custom drivers built after you
./configure --with-unmapped-data-points.
Production driver builds must not include any non-standard names.
The NUT "nutdrv_qx" driver is a meta-driver that handles Q* UPS devices.
It consists of a core driver that handles most of the work of talking to the hardware, and several sub-drivers to handle specific UPS manufacturers.
Adding support for a new UPS device is easy, because it requires only the creation of a new sub-driver.
In order to develop a new subdriver for a specific UPS you have to know the "idiom" (dialect of the protocol) spoken by that device.
This kind of devices speaks idioms that can be summed up as follows:
We send the UPS a query for one or more information
We send the UPS a command
ACK)
NAK)
To be supported by this driver the idiom spoken by the UPS must comply to these conditions.
You have to fill the subdriver_t structure:
typedef struct {
const char *name;
int (*claim)(void);
item_t *qx2nut;
void (*initups)(void);
void (*initinfo)(void);
void (*makevartable)(void);
const char *accepted;
const char *rejected;
#ifdef TESTING
testing_t *testing;
#endif /* TESTING */
} subdriver_t;Where:
name
protocol that will need to be set in the ups.conf file to use this subdriver plus the internal version of it separated by a space (e.g. "Megatec 0.01").
claim
1 if the device is supported by this subdriver, else 0.
qx2nut
item_t mapping a UPS idiom to NUT.
initups (optional)
upsdrv_initups.
This function will be called at the end of nutdrv_qx’s own upsdrv_initups.
initinfo (optional)
upsdrv_initinfo.
This function will be called at the end of nutdrv_qx’s own upsdrv_initinfo.
makevartable (optional)
ups.conf vars and flags.
Make sure not to collide with other subdrivers' vars and flags.
accepted (optional)
\r) and you can decide at which index of the answer the value should start or end setting the appropriate from and to in the item_t (see Mapping an idiom to NUT).
rejected (optional)
\r) and whatever character is expected.
testing
Testing table (an array of testing_t) that will hold the commands and the replies used for testing the subdriver.
testing_t:
typedef struct {
const char *cmd;
const char answer[SMALLBUF];
const int answer_len;
} testing_t;Where:
cmd
answer
Answer for that command.
If answer contains inner \0s, in order to preserve them, answer_len as well as an item_t's preprocess_answer() function must be set.
answer_len
Answer length:
-1 → auto calculate answer length (treat answer as a null-terminated string),
\0s (treat answer as a sequence of bytes till the item_t's preprocess_answer() function gets called).
For more information, see Mapping an idiom to NUT.
If you understand the idiom spoken by your device, you can easily map it to NUT variables and instant commands, filling qx2nut with an array of item_t data structure:
typedef struct item_t {
const char *info_type;
const int info_flags;
info_rw_t *info_rw;
const char *command;
char answer[SMALLBUF];
const int answer_len;
const char leading;
char value[SMALLBUF];
const int from;
const int to;
const char *dfl;
unsigned long qxflags;
int (*preprocess_command)(struct item_t *item, char *command, const size_t commandlen);
int (*preprocess_answer)(struct item_t *item, const int len);
int (*preprocess)(struct item_t *item, char *value, const size_t valuelen);
} item_t;Where:
info_type
QX_FLAG_NONUT is set, name to print to logs and if both QX_FLAG_NONUT and QX_FLAG_SETVAR are set, name of the var to retrieve from ups.conf.
info_flags
ST_FLAG_* values to set in dstate_addinfo).
info_rw
An array of info_rw_t to handle r/w variables:
ST_FLAG_STRING is set in info_flags it’ll be used to set the length of the string (in dstate_setaux)
QX_FLAG_ENUM is set in qxflags it’ll be used to set enumerated values (in dstate_addenum)
QX_FLAG_RANGE is set in qxflags it’ll be used to set range boundaries (in dstate_addrange)
If QX_FLAG_SETVAR is set the value given by the user will be checked against these infos.
info_rw_t:
typedef struct {
char value[SMALLBUF];
int (*preprocess)(char *value, const size_t len);
} info_rw_t;Where:
value
ST_FLAG_STRING.
preprocess(value, len)
Optional function to preprocess range/enum value.
This function will be given value and its size_t and must return either 0 if value is supported or -1 if not supported.
command
answer
Answer from the UPS, filled at runtime.
If you expect a non-valid C string (e.g.: inner \0s) or need to perform actions before the answer is used (and treated as a null-terminated string), you should set a preprocess_answer() function.
answer_len
0 if there’s no minimum length to look after.
leading
#, ( …
value
answer in the interval [from to to]).
from
to
0 if all the remaining of the line is needed.
dfl
printf format to store value from the UPS in NUT variables.
Set it either to %s for strings or to a floating point specifier (e.g. %.1f) for numbers.
Otherwise:
QX_FLAG_ABSENT → default value
QX_FLAG_CMD → default command value
qxflags
Driver’s own flags.
| Retrieve this variable only once. |
| Retrieve this info smartly, i.e. only when a command/setvar is executed and we expect that data could have been changed. |
| Data is absent in the device, use default value. |
| Mandatory vars. |
| Instant command. |
| The var is settable and the actual item stores info on how to set it. |
| This var’s value need to be trimmed of leading/trailing spaces/hashes. |
| Enum values exist. |
| Ranges for this var are available. |
| This var doesn’t have a corresponding var in NUT. |
| Skip this var: this item won’t be processed. |
The driver will run a so-called QX_WALKMODE_INIT in initinfo walking through all the items in qx2nut, adding instant commands and the like.
From then on it’ll run a so-called QX_WALKMODE_QUICK_UPDATE just to see if the UPS is still there and then it’ll do a so-called QX_WALKMODE_FULL_UPDATE to update all the vars.
If there’s a problem with a var in QX_WALKMODE_INIT, the driver will automagically set QX_FLAG_SKIP on it and then it’ll skip that item in QX_WALKMODE_QUICK_UPDATE/QX_WALKMODE_FULL_UPDATE, provided that the item has not the flag QX_FLAG_QUICK_POLL set, in that case the driver will set datastale.
preprocess_command(item, command, commandlen)
item), the command to be sent to the UPS (command) and its size_t (commandlen).
Return -1 in case of errors, else 0.
command must be filled with the actual command to be sent to the UPS.
preprocess_answer(item, len)
item) with the answer we got from the UPS unmolested and already stored in item's answer and the length of that answer (len).
Return -1 in case of errors, else the length of the newly allocated item's answer (from now on, treated as a null-terminated string).
preprocess(item, value, valuelen)
Function to preprocess the data from/to the UPS: you are given the currently processed item (item), a char array (value) and its size_t (valuelen).
Return -1 in case of errors, else 0.
If QX_FLAG_SETVAR/QX_FLAG_CMD is set then the item is processed before the command is sent to the UPS so that you can fill it with the value provided by the user.
In this case value must be filled with the command to be sent to the UPS.
Otherwise the function will be used to process the value we got from the answer of the UPS before it’ll get stored in a NUT variable.
In this case value must be filled with the processed value already compliant to NUT standards.
You must provide an item_t with QX_FLAG_SETVAR and its boundaries set for both ups.delay.start and ups.delay.shutdown to map the driver variables ondelay and offdelay, as they will be used in the shutdown sequence.
In order to keep the data flow at minimum you should keep together the items in qx2nut that need data from the same query (i.e. command): doing so the driver will send the query only once and then every item_t processed after the one that got the answer, provided that it’s filled with the same command and that the answer wasn’t NULL, will get that answer.
The following examples are from the voltronic subdriver.
We know that when the UPS is queried for status with QGS\r, it replies with something like (234.9 50.0 229.8 50.0 000.0 000 369.1 ---.- 026.5 ---.- 018.8 100000000001\r and we want to access the output voltage (the third token, in this case 229.8).
> [QGS\r] < [(234.9 50.0 229.8 50.0 000.0 000 369.1 ---.- 026.5 ---.- 018.8 100000000001\r] 0123456789012345678901234567890123456789012345678901234567890123456789012345 0 1 2 3 4 5 6 7
Here’s the item_t:
{ "output.voltage", 0, NULL, "QGS\r", "", 76, '(', "", 12, 16, "%.1f", 0, NULL, NULL, NULL },|
|
|
|
|
|
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
We are expecting a number, so at first the core driver will check if it’s made up entirely of digits/points/spaces, then it’ll convert it into a double. Because of that we need to provide a floating point specifier. |
|
|
|
|
|
|
|
|
|
|
|
|
Also from QGS\r, we want to process the 9th status bit 100000000001 that tells us whether the UPS is shutting down or not.
> [QGS\r] < [(234.9 50.0 229.8 50.0 000.0 000 369.1 ---.- 026.5 ---.- 018.8 100000000001\r] 0123456789012345678901234567890123456789012345678901234567890123456789012345 0 1 2 3 4 5 6 7
Here’s the item_t:
{ "ups.status", 0, NULL, "QGS\r", "", 76, '(', "", 71, 71, "%s", QX_FLAG_QUICK_POLL, NULL, NULL, voltronic_status },|
|
|
|
|
|
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Since a |
|
|
|
|
|
|
|
|
|
|
|
This function will be called after the |
So your UPS reports its battery type when queried for QBT\r; we are expecting an answer like (01\r and we know that the values can be mapped as follows: 00 → "Li", 01 → "Flooded" and 02 → "AGM".
> [QBT\r] < [(01\r] <- 00="Li", 01="Flooded" or 02="AGM" 0123 0
Here’s the item_t:
{ "battery.type", ST_FLAG_RW, voltronic_e_batt_type, "QBT\r", "", 4, '(', "", 1, 2, "%s",
QX_FLAG_SEMI_STATIC | QX_FLAG_ENUM, NULL, NULL, voltronic_p31b },|
|
|
|
|
|
|
|
The values stored here will be added to the NUT variable, setting its boundaries: in this case |
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Since a |
|
|
|
|
|
|
|
|
|
|
|
This function will be called after the |
We also know that we can change battery type with the PBTnn\r command; we are expecting either (ACK\r if the command succeeded or (NAK\r if the command is rejected.
> [PBTnn\r] nn = 00/01/02 < [(ACK\r] 01234 0
Here’s the item_t:
{ "battery.type", 0, voltronic_e_batt_type, "PBT%02.0f\r", "", 5, '(', "", 1, 4, NULL,
QX_FLAG_SETVAR | QX_FLAG_ENUM, NULL, NULL, voltronic_p31b_set },|
|
|
|
|
|
|
|
The value provided by the user will be automagically checked by the core nutdrv_qx driver against the enumerated values already set by the non setvar item (i.e. |
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Not used for |
|
|
|
|
|
|
|
|
|
|
|
This function will be called before the |
We know that we have to send to the UPS Tnn\r or T.n\r in order to start a battery test lasting nn minutes or .n minutes: we are expecting either (ACK\r on success or (NAK\r if the command is rejected.
> [Tnn\r] < [(ACK\r] 01234 0
Here’s the item_t:
{ "test.battery.start", 0, NULL, "T%s\r", "", 5, '(', "", 1, 4, NULL, QX_FLAG_CMD, NULL, NULL, voltronic_process_command },|
|
|
|
|
|
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Not used for |
|
|
|
|
|
|
|
|
|
|
|
This function will be called before the |
In order to set the server-side var ups.delay.start, that will be then used by the driver, we have to provide the following item_t:
{ "ups.delay.start", ST_FLAG_RW, voltronic_r_ondelay, NULL, "", 0, 0, "", 0, 0, "180",
QX_FLAG_ABSENT | QX_FLAG_SETVAR | QX_FLAG_RANGE, NULL, NULL, voltronic_process_setvar },|
|
|
|
|
|
|
|
The values stored here will be added to the NUT variable, setting its boundaries: in this case |
|
|
Not used for |
|
|
Not used for |
|
|
Not used for |
|
|
Not used for |
|
|
Not used for |
|
|
Not used for |
|
|
Not used for |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This function will be called, in setvar, before the driver stores the value in the NUT var: here it’s used to truncate the user-provided value to the nearest settable interval. |
If your UPS reports some data items that are not yet available as NUT variables and you need to process them, you can add them in item_t data structure adding the QX_FLAG_NONUT flag to its qxflags: the info will then be printed to the logs.
So we know that the UPS reports actual input/output phase angles when queried for QPD\r:
> [QPD\r] < [(000 120\r] <- Input Phase Angle -- Output Phase Angle 012345678 0
Here’s the item_t for input phase angle:
{ "input_phase_angle", 0, NULL, "QPD\r", "", 9, '(', "", 1, 3, "%03.0f",
QX_FLAG_STATIC | QX_FLAG_NONUT, NULL, NULL, voltronic_phase },|
|
This information will be used to print the value we got back from the UPS in the logs. |
|
|
|
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
If there’s no |
|
|
|
|
|
|
|
|
|
|
|
This function will be called after the |
Here’s the item_t for output phase angle:
{ "output_phase_angle", ST_FLAG_RW, voltronic_e_phase, "QPD\r", "", 9, '(', "", 5, 7, "%03.0f",
QX_FLAG_SEMI_STATIC | QX_FLAG_ENUM | QX_FLAG_NONUT, NULL, NULL, voltronic_phase },|
|
This information will be used to print the value we got back from the UPS in the logs. |
|
|
This could also be |
|
|
Enumerated list of available value (here: |
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
If there’s no |
|
|
|
|
|
|
|
|
|
|
|
This function will be called after the |
If you need also to change some values in the UPS you can add a ups.conf var/flag in the subdriver’s own makevartable and then process it adding to its qxflags both QX_FLAG_NONUT and QX_FLAG_SETVAR: this item will be processed only once in QX_WALKMODE_INIT.
The driver will check if the var/flag is defined in ups.conf: if so, it’ll then call setvar passing to this item the defined value, if any, and then it’ll print the results in the logs.
We know we can set output phase angle sending PPDnnn\r to the UPS:
> [PPDn\r] n = (000, 120, 180 or 240) < [(ACK\r] 01234 0
Here’s the item_t
{ "output_phase_angle", 0, voltronic_e_phase, "PPD%03.0f\r", "", 5, '(', "", 1, 4, NULL,
QX_FLAG_SETVAR | QX_FLAG_ENUM | QX_FLAG_NONUT, NULL, NULL, voltronic_phase_set },|
|
This information will be used to print the value we got back from the UPS in the logs and to retrieve the user-provided value in |
|
|
|
|
|
Enumerated list of available values (here: |
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Filled at runtime |
|
|
|
|
|
|
|
|
Not used for |
|
|
|
|
|
|
|
|
|
|
|
This function will be called before the |
You are already given the following functions:
int instcmd(const char *cmdname, const char *extradata)
Execute an instant command. In detail:
cmdname in the qx2nut data structure (if not found, try to fallback to commonly known commands);
cmdname is found, call its preprocess function, passing to it extradata, if any, otherwise its dfl value, if any;
Return STAT_INSTCMD_INVALID if the command is invalid, STAT_INSTCMD_FAILED if it failed, STAT_INSTCMD_HANDLED on success.
int setvar(const char *varname, const char *val)
info_rw structure.
Return STAT_SET_HANDLED on success, otherwise STAT_SET_UNKNOWN.
item_t *find_nut_info(const char *varname, const unsigned long flag, const unsigned long noflag)
Find an item of item_t type in qx2nut data structure by its info_type, optionally filtered by its qxflags, and return it if found, otherwise return NULL.
flag: flags that have to be set in the item, i.e. if one of the flags is absent in the item it won’t be returned.
noflag: flags that have to be absent in the item, i.e. if at least one of the flags is set in the item it won’t be returned.
int qx_process(item_t *item, const char *command)
command (a null-terminated byte string) or, if it is NULL, send the command stored in the item to the UPS and process the reply, saving it in item's answer.
Return -1 on errors, 0 on success.
int ups_infoval_set(item_t *item)
item-specific preprocess function, if any, otherwise executing the standard preprocessing (including trimming if QX_FLAG_TRIM is set).
Return -1 on failure, 0 for a status update and 1 in all other cases.
int qx_status(void)
status_bit_t passed to the STATUS() macro (see nutdrv_qx.h).
void update_status(const char *nutvalue)
OB are supported, simply set it as not OL); prefix them with an exclamation mark if you want to clear them from the status (e.g. !OL).
Armac subdriver is based on reverse engineering of Power Manager II software by Richcomm Technologies written in 2005 that is still (as of 2023) being distributed as a valid software for freshly sold UPS of various manufacturers. It uses commands as defined for Megatec protocol - but has a different communication mechanism.
It uses two types of USB interrupt transfers: - 4 bytes to send a command (usually single transfer). - 6 byte chunk to read a reply (multiple transfers).
Transfers are similar to those of the richcomm nut driver, but the transferred data is not short binary commands. Instead, serial text data is overlaid in these transfers in a way that creates a badly made USB serial interface. UPS reply looks similar to this:
0 1 2 3 4 5 HL 00 00 00 00 00
HL is a control byte. Its high nibble meaning is unknown. It changes between two possible values during transmission. Low nibble encodes number of bytes that have a meaning in the transaction. For example there are 5 bytes that might contain ASCII serial data, but only some might be valid, and other might be random, stale buffer data, etc.
What follows is set of observed transmissions by various UPSes gathered from Github issues.
## Vultech V2000
419.987514 [D4] armac command Q1 419.988307 [D4] armac cleanup ret i=0 ret=6 ctrl=c0 420.119402 [D4] read: ret 6 buf 81: 28 30 31 30 30 >(0100< 420.130383 [D4] read: ret 6 buf c1: 32 30 31 30 30 >20100< 420.141408 [D4] read: ret 6 buf 82: 33 33 31 30 30 >33100< 420.152201 [D4] read: ret 6 buf c3: 2e 30 20 30 30 >.0 00< 420.153237 [D4] read: ret 6 buf 82: 30 30 20 30 30 >00 00< 420.164299 [D4] read: ret 6 buf c1: 30 30 20 30 30 >00 00< 420.175293 [D4] read: ret 6 buf 82: 2e 30 20 30 30 >.0 00< 420.186358 [D4] read: ret 6 buf c3: 20 32 33 30 30 > 2300< 420.190322 [D4] read: ret 6 buf 83: 33 2e 30 30 30 >3.000< 420.194323 [D4] read: ret 6 buf c1: 20 2e 30 30 30 > .000< 420.205358 [D4] read: ret 6 buf 81: 30 2e 30 30 30 >0.000< 420.216318 [D4] read: ret 6 buf c2: 31 34 30 30 30 >14000< 420.227445 [D4] read: ret 6 buf 83: 20 34 39 30 30 > 4900< 420.228334 [D4] read: ret 6 buf c2: 2e 30 39 30 30 >.0900< 420.239461 [D4] read: ret 6 buf 81: 20 30 39 30 30 > 0900< 420.250411 [D4] read: ret 6 buf c2: 32 37 39 30 30 >27900< 420.261405 [D4] read: ret 6 buf 83: 2e 30 20 30 30 >.0 00< 420.265468 [D4] read: ret 6 buf c3: 32 30 2e 30 30 >20.00< 420.269465 [D4] read: ret 6 buf 81: 38 30 2e 30 30 >80.00< 420.280322 [D4] read: ret 6 buf c1: 20 30 2e 30 30 > 0.00< 420.291469 [D4] read: ret 6 buf 82: 30 30 2e 30 30 >00.00< 420.302465 [D4] read: ret 6 buf c3: 30 30 31 30 30 >00100< 420.303511 [D4] read: ret 6 buf 82: 00 30 31 30 30 > <- This has 0x00 and '0', will be read as "00" 420.303515 [D3] found null byte in status bits at 43 byte, assuming 0. 420.314425 [D4] read: ret 6 buf c1: 31 30 31 30 30 >10100< <- this has '1' 420.325432 [D4] read: ret 6 buf 81: 0d 30 31 30 30 >.0100< <- and this finishes with `\r`. 420.325442 [D3] armac command Q1 response read: '(233.0 000.0 233.0 014 49.0 27.0 20.8 00001001'
1.185164 [D4] armac command ID 1.316257 [D4] read: ret 6 buf c1: 23 31 00 30 30 >#1 1.327309 [D4] read: ret 6 buf 81: 20 31 00 30 30 > 1 1.338264 [D4] read: ret 6 buf c2: 20 20 00 30 30 > 1.349151 [D4] read: ret 6 buf 83: 20 20 20 30 30 > 00< 1.360277 [D4] read: ret 6 buf c2: 20 20 20 30 30 > 00< 1.371322 [D4] read: ret 6 buf 83: 20 20 20 30 30 > 00< 1.382265 [D4] read: ret 6 buf c3: 20 20 20 30 30 > 00< 1.393156 [D4] read: ret 6 buf 82: 20 20 20 30 30 > 00< 1.404324 [D4] read: ret 6 buf c3: 20 20 20 30 30 > 00< 1.415342 [D4] read: ret 6 buf 83: 20 20 20 30 30 > 00< 1.426292 [D4] read: ret 6 buf c2: 20 20 20 30 30 > 00< 1.437203 [D4] read: ret 6 buf 83: 20 20 20 30 30 > 00< 1.448328 [D4] read: ret 6 buf c3: 56 34 2e 30 30 >V4.00< 1.459293 [D4] read: ret 6 buf 82: 31 30 2e 30 30 >10.00< 1.470274 [D4] read: ret 6 buf c3: 20 20 20 30 30 > 00< 1.481208 [D4] read: ret 6 buf 82: 20 20 20 30 30 > 00< 1.492261 [D4] read: ret 6 buf c1: 0d 20 20 30 30 > 1.492270 [D3] armac command ID response read: '# V4.10 '
4.749667 [D4] armac command F 4.876638 [D4] read: ret 6 buf 81: 23 31 00 30 30 >#1 4.887614 [D4] read: ret 6 buf c1: 32 31 00 30 30 >21 4.898644 [D4] read: ret 6 buf 82: 32 30 00 30 30 >20 4.909595 [D4] read: ret 6 buf c3: 2e 30 20 30 30 >.0 00< 4.920648 [D4] read: ret 6 buf 82: 30 30 20 30 30 >00 00< 4.931629 [D4] read: ret 6 buf c3: 35 20 32 30 30 >5 200< 4.942601 [D4] read: ret 6 buf 83: 34 2e 30 30 30 >4.000< 4.953666 [D4] read: ret 6 buf c2: 30 20 30 30 30 >0 000< 4.964535 [D4] read: ret 6 buf 83: 35 30 2e 30 30 >50.00< 4.975540 [D4] read: ret 6 buf c2: 30 0d 2e 30 30 >0 4.975546 [D3] armac command F response read: '#220.0 005 24.00 50.0'
## Armac R/2000I/PSW
112.966856 [D4] armac command Q1 112.968197 [D4] armac cleanup ret i=0 ret=6 ctrl=c0 <- Cleanups required. 113.091193 [D4] read: ret 6 buf 81: 28 30 0d 2e 30 >(0 <- Usually 1-3 bytes available in transfer. 113.103211 [D4] read: ret 6 buf c1: 30 30 0d 2e 30 >00 113.115180 [D4] read: ret 6 buf 82: 30 30 0d 2e 30 >00 113.117144 [D4] read: ret 6 buf c3: 2e 30 20 2e 30 >.0 .0< 113.120150 [D4] read: ret 6 buf 81: 31 30 20 2e 30 >10 .0< 113.132178 [D4] read: ret 6 buf c1: 34 30 20 2e 30 >40 .0< 113.144159 [D4] read: ret 6 buf 82: 30 2e 20 2e 30 >0. .0< 113.146149 [D4] read: ret 6 buf c3: 30 20 32 2e 30 >0 2.0< 113.149173 [D4] read: ret 6 buf 81: 32 20 32 2e 30 >2 2.0< 113.161167 [D4] read: ret 6 buf c1: 37 20 32 2e 30 >7 2.0< 113.173159 [D4] read: ret 6 buf 82: 2e 30 32 2e 30 >.02.0< 113.175157 [D4] read: ret 6 buf c3: 20 30 30 2e 30 > 00.0< 113.178158 [D4] read: ret 6 buf 81: 32 30 30 2e 30 >200.0< 113.190157 [D4] read: ret 6 buf c1: 20 30 30 2e 30 > 00.0< 113.202161 [D4] read: ret 6 buf 82: 30 30 30 2e 30 >000.0< 113.204154 [D4] read: ret 6 buf c3: 2e 30 20 2e 30 >.0 .0< 113.207150 [D4] read: ret 6 buf 81: 34 30 20 2e 30 >40 .0< 113.219174 [D4] read: ret 6 buf c1: 36 30 20 2e 30 >60 .0< 113.231165 [D4] read: ret 6 buf 82: 2e 38 20 2e 30 >.8 .0< 113.233157 [D4] read: ret 6 buf c3: 20 35 36 2e 30 > 56.0< 113.237149 [D4] read: ret 6 buf 81: 2e 35 36 2e 30 >.56.0< 113.249168 [D4] read: ret 6 buf c1: 30 35 36 2e 30 >056.0< 113.261155 [D4] read: ret 6 buf 83: 20 31 30 2e 30 > 10.0< 113.263151 [D4] read: ret 6 buf c2: 30 30 30 2e 30 >000.0< 113.266152 [D4] read: ret 6 buf 81: 31 30 30 2e 30 >100.0< 113.278161 [D4] read: ret 6 buf c1: 30 30 30 2e 30 >000.0< <- No Null bytes. 113.290155 [D4] read: ret 6 buf 82: 30 30 30 2e 30 >000.0< 113.292159 [D4] read: ret 6 buf c1: 0d 30 30 2e 30 > 113.292169 [D3] armac command Q1 response read: '(000.0 140.0 227.0 002 00.0 46.8 56.0 10001000'
Next query would return 0x80 control byte - 0 available bytes. This used to terminate transmission, but some UPS don’t work like that.
## Armac R/3000I/PF1
0.083301 [D4] armac command Q1 0.164847 [D4] read: ret 6 buf a6: 28 32 34 31 2e >(241.< 0.184839 [D4] read: ret 6 buf 86: 35 20 30 30 30 >5 000< 0.205851 [D4] read: ret 6 buf a6: 2e 30 20 32 33 >.0 23< 0.226849 [D4] read: ret 6 buf 86: 30 2e 33 20 30 >0.3 0< 0.247859 [D4] read: ret 6 buf a6: 30 30 20 34 39 >00 49< 0.268862 [D4] read: ret 6 buf 86: 2e 39 20 32 2e >.9 2.< 0.289857 [D4] read: ret 6 buf a6: 32 35 20 34 38 >25 48< 0.309866 [D4] read: ret 6 buf 86: 2e 30 20 30 30 >.0 00< 0.330863 [D4] read: ret 6 buf a6: 30 30 30 30 30 >00000< 0.827913 [D4] read: ret 6 buf 83: 31 0d 30 30 30 >1 000< 0.827927 [D3] armac command Q1 response read: '(241.5 000.0 230.3 000 49.9 2.25 48.0 00000001' 0.827954 [D4] armac command ID 1.394985 [D4] read: ret 6 buf a5: 4e 41 4b 0d 30 >NAK < 1.395001 [D3] armac command ID response read: 'NAK'
This UPS sends higher nibble set to 6 often, which exceeds available bytes.
Maybe means that more are available. Its serial-USB bridge is probably faster.
We read 5 bytes in case 6 nibble is sent. End of transmission is marked by \r,
no 0 nibble is sent.
You must put the generated files into the drivers/ subdirectory, with the name of your subdriver preceded by nutdrv_qx_, and update nutdrv_qx.c by adding the appropriate #include line and by updating the definition of subdriver_list.
Please, make sure to add your driver in that list in a smart way: if your device supports also the basic commands used by the other subdrivers to claim a device, add something that is unique (i.e. not supported by the other subdrivers) to your device in your claim function and then add it on top of the slightly supported ones in that list.
You must also add the subdriver to NUTDRV_QX_SUBDRIVERS list variable in the drivers/Makefile.am and call "autoreconf" and/or "./configure" from the top level NUT directory.
You can then recompile nutdrv_qx, and start experimenting with the new subdriver.
For more details, have a look at the currently available subdrivers:
nutdrv_qx_bestups.{c,h}
nutdrv_qx_masterguard.{c,h}
nutdrv_qx_mecer.{c,h}
nutdrv_qx_megatec.{c,h}
nutdrv_qx_megatec-old.{c,h}
nutdrv_qx_mustek.{c,h}
nutdrv_qx_q1.{c,h}
nutdrv_qx_voltronic.{c,h}
nutdrv_qx_voltronic-qs.{c,h}
nutdrv_qx_voltronic-qs-hex.{c,h}
nutdrv_qx_zinto.{c,h}
nutdrv_qx_ablerex.{c,h}