<?xml version="1.0" encoding="latin1" ?>
<!DOCTYPE erlref SYSTEM "erlref.dtd">
<erlref>
<header>
<copyright>
<year>1997</year><year>2009</year>
<holder>Ericsson AB. All Rights Reserved.</holder>
</copyright>
<legalnotice>
The contents of this file are subject to the Erlang Public License,
Version 1.1, (the "License"); you may not use this file except in
compliance with the License. You should have received a copy of the
Erlang Public License along with this software. If not, it can be
retrieved online at http://www.erlang.org/.
Software distributed under the License is distributed on an "AS IS"
basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
the License for the specific language governing rights and limitations
under the License.
</legalnotice>
<title>erl_ddll</title>
<prepared></prepared>
<docno></docno>
<date></date>
<rev></rev>
</header>
<module>erl_ddll</module>
<modulesummary>Dynamic Driver Loader and Linker</modulesummary>
<description>
<p>The <c>erl_ddll</c> module provides an interface for loading
and unloading <em>erlang linked in drivers</em> in runtime.</p>
<note>
<p>This is a large reference document. For casual use of the
module, as well as for most real world applications, the
descriptions of the functions <seealso marker="#load/2">load/2</seealso> and <seealso marker="#unload/1">unload/1</seealso> are enough to get
going. </p>
</note>
<p>The driver should be provided as a dynamically linked library
in a object code format specific for the platform in use,
i. e. <c>.so</c> files on most Unix systems and <c>.ddl</c>
files on windows. An erlang linked in driver has to provide
specific interfaces to the emulator, so this module is not
designed for loading arbitrary dynamic libraries. For further
information about erlang drivers, refer to the ERTS reference
manual section <seealso marker="erts:erl_driver">erl_driver</seealso>.</p>
<marker id="users"></marker>
<p>When describing a set of functions, (i.e. a module, a part of a
module or an application) executing in a process and wanting to
use a ddll-driver, we use the term <em>user</em>. There can be
several users in one process (different modules needing the same
driver) and several processes running the same code, making up
several <em>users</em> of a driver. In the basic scenario, each
user loads the driver before starting to use it and unloads the
driver when done. The reference counting keeps track of
processes as well as the number of loads by each process, so that
the driver will only be unloaded when no one wants it
(it has no user). The driver also keeps track of ports that are
opened towards it, so that one can delay unloading until all
ports are closed or kill all ports using the driver when it is
unloaded. </p>
<marker id="scenarios"></marker>
<p>The interface supports two basic scenarios of loading and
unloading. Each scenario can also have the option of either
killing ports when the driver is unloading, or waiting for the
ports to close themselves. The scenarios are:</p>
<taglist>
<tag><em>Load and unload on a "when needed basis"</em></tag>
<item>
<p>This (most common) scenario simply supports that each
<seealso marker="#users">user</seealso> of the driver loads
it when it is needed and unloads it when the <seealso marker="#users">user</seealso> no longer have any use for
it. The driver is always reference counted and as long as a
process keeping the driver loaded is still alive, the driver
is present in the system.</p>
<p>Each <seealso marker="#users">user</seealso> of the driver
use <em>literally</em> the same pathname for the driver when
demanding load, but the <seealso marker="#users">users</seealso> are not really concerned
with if the driver is already loaded from the filesystem or
if the object code has to be loaded from filesystem.</p>
<p>Two pairs of functions support this scenario:</p>
<taglist>
<tag><em>load/2 and unload/1</em></tag>
<item>
<p>When using the <c>load/unload</c> interfaces, the
driver will not <em>actually</em> get unloaded until the
<em>last port</em> using the driver is closed. The function
<c>unload/1</c> can return immediately, as the <seealso marker="#users">users</seealso> are not really concerned
with when the actual unloading occurs. The
driver will actually get unloaded when no one needs it any longer.</p>
<p>If a process having the driver loaded dies, it will have
the same effect as if unloading was done. </p>
<p>When loading, the function <c>load/2</c> returns
<c>ok</c> as soon as there is any instance of the driver
present, so that if a driver is waiting to get unloaded
(due to open ports), it will simply change state to no
longer need unloading.</p>
</item>
<tag><em>load_driver/2 and unload_driver/1</em></tag>
<item>
<p>These interfaces is intended to be used when it is considered an
error that ports are open towards a driver that no <seealso marker="#users">user</seealso>
has loaded. The ports still open when the
last <seealso marker="#users">user</seealso> calls
<c>unload_driver/1</c> or when the last process having the
driver loaded dies, will get killed with reason
<c>driver_unloaded</c>.</p>
<p>The function names <c>load_driver</c> and
<c>unload_driver</c> are kept for backward
compatibility.</p>
</item>
</taglist>
</item>
<tag><em>Loading and reloading for code replacement</em></tag>
<item>
<p>This scenario occurs when the driver code might need
replacement during operation of the Erlang
emulator. Implementing driver code replacement is somewhat
more tedious than beam code replacement, as one driver
cannot be loaded as both "old" and "new" code. All <seealso marker="#users">users</seealso> of a driver must have it
closed (no open ports) before the old code can be unloaded
and the new code can be loaded.</p>
<p>The actual unloading/loading is done as one atomic
operation, blocking all processes in the system from using
the driver concerned while in progress.</p>
<p>The preferred way to do driver code replacement is to let
<em>one single process</em> keep track of the driver. When
the process start, the driver is loaded. When replacement
is required, the driver is reloaded. Unload is probably never
done, or done when the process exits. If more than one <seealso marker="#users">user</seealso> has a driver loaded when code
replacement is demanded, the replacement cannot occur until
the last "other" <seealso marker="#users">user</seealso> has
unloaded the driver.</p>
<p>Demanding reload when a reload is already in progress is
always an error. Using the high level functions, it is also
an error to demand reloading when more than one <seealso marker="#users">user</seealso> has the driver loaded. To
simplify driver replacement, avoid designing your system so
that more than than one <seealso marker="#users">user</seealso> has the driver loaded.</p>
<p>The two functions for reloading drivers should be used
together with corresponding load functions, to support the two
different behaviors concerning open ports:</p>
<taglist>
<tag><em>load/2 and reload/2</em></tag>
<item>
<p>This pair of functions is used when reloading should be
done after the last open port towards the driver is
closed.</p>
<p>As <c>reload/2</c> actually waits for the reloading to
occur, a misbehaving process keeping open ports towards
the driver (or keeping the driver loaded) might cause
infinite waiting for reload. Timeouts has to be provided
outside of the process demanding the reload or by using
the low-level interface <seealso marker="#try_load/3">try_load/3</seealso> in combination
with driver monitors (see below).</p>
</item>
<tag><em>load_driver/2 and reload_driver/2</em></tag>
<item>
<p>This pair of functions are used when open ports towards
the driver should be killed with reason
<c>driver_unloaded</c> to allow for new driver code to
get loaded.</p>
<p>If, however, another process has the driver loaded,
calling <c>reload_driver</c> returns the error code
<c>pending_process</c>. As stated earlier,
the recommended design is to not allow other <seealso marker="#users">users</seealso> than the "driver
reloader" to actually demand loading of the concerned
driver.</p>
</item>
</taglist>
</item>
</taglist>
</description>
<funcs>
<func>
<name>demonitor(MonitorRef) -> ok</name>
<fsummary>Remove a monitor for a driver</fsummary>
<type>
<v>MonitorRef = ref()</v>
</type>
<desc>
<p>Removes a driver monitor in much the same way as
<seealso marker="erts:erlang#erlang:demonitor/1">erlang:demonitor/1</seealso> does with process
monitors. See <seealso marker="#monitor/2">monitor/2</seealso>, <seealso marker="#try_load/3">try_load/3</seealso> and <seealso marker="#try_unload/2">try_unload/2</seealso> for details
about how to create driver monitors.</p>
<p>The function throws a <c>badarg</c> exception if the
parameter is not a ref(). </p>
</desc>
</func>
<func>
<name>info() -> AllInfoList</name>
<fsummary>Retrieve information about all drivers</fsummary>
<type>
<v>AllInfoList = [ DriverInfo ]</v>
<v>DriverInfo = {DriverName, InfoList}</v>
<v>DriverName = string()</v>
<v>InfoList = [ InfoItem ]</v>
<v>InfoItem = {Tag, Value}</v>
<v>Tag = atom()</v>
<v>Value = term()</v>
</type>
<desc>
<p>Returns a list of tuples <c>{DriverName, InfoList}</c>, where
<c>InfoList</c> is the result of calling <seealso marker="#info/1">info/1</seealso> for that
<c>DriverName</c>. Only dynamically linked in drivers are
included in the list.</p>
</desc>
</func>
<func>
<name>info(Name) -> InfoList</name>
<fsummary>Retrieve information about one driver</fsummary>
<type>
<v>Name = string() | atom()</v>
<v>InfoList = [ InfoItem ]</v>
<v>InfoItem = {Tag, Value}</v>
<v>Tag = atom()</v>
<v>Value = term()</v>
</type>
<desc>
<p>Returns a list of tuples <c>{Tag, Value}</c>, where
<c>Tag</c> is the information item and <c>Value</c> is the result
of calling <seealso marker="#info/2">info/2</seealso> with this driver name and
this tag. The result being a tuple list containing all
information available about a driver. </p>
<p>The different tags that will appear in the list are:</p>
<list type="bulleted">
<item>processes</item>
<item>driver_options</item>
<item>port_count</item>
<item>linked_in_driver</item>
<item>permanent</item>
<item>awaiting_load</item>
<item>awaiting_unload</item>
</list>
<p>For a detailed description of each value, please read the
description of <seealso marker="#info/2">info/2</seealso> below.</p>
<p>The function throws a <c>badarg</c> exception if the driver
is not present in the system.</p>
</desc>
</func>
<func>
<name>info(Name, Tag) -> Value</name>
<fsummary>Retrieve specific information about one driver</fsummary>
<type>
<v>Name = string() | atom()</v>
<v>Tag = processes | driver_options | port_count | linked_in_driver | permanent | awaiting_load | awaiting_unload</v>
<v>Value = term()</v>
</type>
<desc>
<p>This function returns specific information about one aspect
of a driver. The <c>Tag</c> parameter specifies which aspect
to get information about. The <c>Value</c> return differs
between different tags:</p>
<taglist>
<tag><em>processes</em></tag>
<item>
<p>Return all processes containing <seealso marker="#users">users</seealso> of the specific drivers
as a list of tuples <c>{pid(),int()}</c>, where the
<c>int()</c> denotes the number of users in the process
<c>pid()</c>.</p>
</item>
<tag><em>driver_options</em></tag>
<item>
<p>Return a list of the driver options provided when
loading, as well as any options set by the driver itself
during initialization. The currently only valid option
being <c>kill_ports</c>.</p>
</item>
<tag><em>port_count</em></tag>
<item>
<p>Return the number of ports (an <c>int()</c>) using the driver.</p>
</item>
<tag><em>linked_in_driver</em></tag>
<item>
<p>Return a <c>bool()</c>, being <c>true</c> if the driver is a
statically linked in one and <c>false</c> otherwise.</p>
</item>
<tag><em>permanent</em></tag>
<item>
<p>Return a <c>bool()</c>, being <c>true</c> if the driver has made
itself permanent (and is <em>not</em> a statically
linked in driver). <c>false</c> otherwise.</p>
</item>
<tag><em>awaiting_load</em></tag>
<item>
<p>Return a list of all processes having monitors for
<c>loading</c> active, each process returned as
<c>{pid(),int()}</c>, where the <c>int()</c> is the
number of monitors held by the process <c>pid()</c>.</p>
</item>
<tag><em>awaiting_unload</em></tag>
<item>
<p>Return a list of all processes having monitors for
<c>unloading</c> active, each process returned as
<c>{pid(),int()}</c>, where the <c>int()</c> is the
number of monitors held by the process <c>pid()</c>.</p>
</item>
</taglist>
<p>If the options <c>linked_in_driver</c> or <c>permanent</c>
return true, all other options will return the value
<c>linked_in_driver</c> or <c>permanent</c> respectively.</p>
<p>The function throws a <c>badarg</c> exception if the driver
is not present in the system or the tag is not supported.</p>
</desc>
</func>
<func>
<name>load(Path, Name) -> ok | {error, ErrorDesc}</name>
<fsummary>Load a driver</fsummary>
<type>
<v>Path = Name = string() | atom()</v>
<v>ErrorDesc = term()</v>
</type>
<desc>
<p>Loads and links the dynamic driver <c>Name</c>. <c>Path</c>
is a file path to the directory containing the driver.
<c>Name</c> must be a sharable object/dynamic library. Two
drivers with different <c>Path</c> parameters cannot be
loaded under the same name. The <c>Name</c> is a string or
atom containing at least one character.</p>
<p>The <c>Name</c> given should correspond to the filename
of the actual dynamically loadable object file residing in
the directory given as <c>Path</c>, but <em>without</em> the
extension (i.e. <c>.so</c>). The driver name provided in
the driver initialization routine must correspond with the
filename, in much the same way as erlang module names
correspond to the names of the <c>.beam</c> files.</p>
<p>If the driver has been previously unloaded, but is still
present due to open ports against it, a call to
<c>load/2</c> will stop the unloading and keep the driver
(as long as the <c>Path</c> is the same) and <c>ok</c> is
returned. If one actually wants the object code to be
reloaded, one uses <seealso marker="#reload/2">reload/2</seealso> or the low-level
interface <seealso marker="#try_load/3">try_load/3</seealso>
instead. Please refer to the description of <seealso marker="#scenarios">different scenarios</seealso> for
loading/unloading in the introduction.</p>
<p>If more than one process tries to load an already loaded
driver withe the same <c>Path</c>, or if the same process
tries to load it several times, the function will return
<c>ok</c>. The emulator will keep track of the
<c>load/2</c> calls, so that a corresponding number of
<c>unload/2</c> calls will have to be done from the same
process before the driver will actually get unloaded. It is
therefore safe for an application to load a driver that is
shared between processes or applications when needed. It can
safely be unloaded without causing trouble for other
parts of the system. </p>
<p>It is not allowed to load
several drivers with the same name but with different
<c>Path</c> parameters.</p>
<note>
<p>Note especially that the <c>Path</c> is interpreted
literally, so that all loaders of the same driver needs to
give the same <em>literal</em><c>Path</c> string, even
though different paths might point out the same directory
in the filesystem (due to use of relative paths and
links).</p>
</note>
<p>On success, the function returns <c>ok</c>. On
failure, the return value is <c>{error,ErrorDesc}</c>,
where <c>ErrorDesc</c> is an opaque term to be
translated into human readable form by the <seealso marker="#format_error/1">format_error/1</seealso>
function.</p>
<p>For more control over the error handling, again use the
<seealso marker="#try_load/3">try_load/3</seealso>
interface instead.</p>
<p>The function throws a <c>badarg</c> exception if the
parameters are not given as described above. </p>
</desc>
</func>
<func>
<name>load_driver(Path, Name) -> ok | {error, ErrorDesc}</name>
<fsummary>Load a driver</fsummary>
<type>
<v>Path = Name = string() | atom()</v>
<v>ErrorDesc = term()</v>
</type>
<desc>
<p>Works essentially as <c>load/2</c>, but will load the driver
with options other options. All ports that are using the
driver will get killed with the reason
<c>driver_unloaded</c> when the driver is to be unloaded.</p>
<p>The number of loads and unloads by different <seealso marker="#users">users</seealso> influence the actual loading
and unloading of a driver file. The port killing will
therefore only happen when the <em>last</em><seealso marker="#users">user</seealso> unloads the driver, or the
last process having loaded the driver exits.</p>
<p>This interface (or at least the name of the functions) is
kept for backward compatibility. Using <seealso marker="#try_load/3">try_load/3</seealso> with
<c>{driver_options,[kill_ports]} </c> in the option list will
give the same effect regarding the port killing.</p>
<p>The function throws a <c>badarg</c> exception if the
parameters are not given as described above. </p>
</desc>
</func>
<func>
<name>monitor(Tag, Item) -> MonitorRef</name>
<fsummary>Create a monitor for a driver</fsummary>
<type>
<v>Tag = driver </v>
<v>Item = {Name, When}</v>
<v>Name = atom() | string()</v>
<v>When = loaded | unloaded | unloaded_only</v>
<v>MonitorRef = ref()</v>
</type>
<desc>
<p>This function creates a driver monitor and works in many
ways as the function <seealso marker="erts:erlang#erlang:monitor/2">erlang:monitor/2</seealso>,
does for processes. When a driver changes state, the monitor
results in a monitor-message being sent to the calling
process. The <c>MonitorRef</c> returned by this function is
included in the message sent.</p>
<p>As with process monitors, each driver monitor set will only
generate <em>one single message</em>. The monitor is
"destroyed" after the message is sent and there is then no
need to call <seealso marker="#demonitor/1">demonitor/1</seealso>.</p>
<p>The <c>MonitorRef</c> can also be used in subsequent calls
to <seealso marker="#demonitor/1">demonitor/1</seealso> to
remove a monitor.</p>
<p>The function accepts the following parameters:</p>
<taglist>
<tag><em>Tag</em></tag>
<item>
<p>The monitor tag is always <c>driver</c> as this function
can only be used to create driver monitors. In the future,
driver monitors will be integrated with process monitors,
why this parameter has to be given for consistence.</p>
</item>
<tag><em>Item</em></tag>
<item>
<p>The <c>Item</c> parameter specifies which driver one
wants to monitor (the name of the driver) as well as
which state change one wants to monitor. The parameter
is a tuple of arity two whose first element is the
driver name and second element is either of:</p>
<taglist>
<tag><em>loaded</em></tag>
<item>
<p>Notify me when the driver is reloaded (or loaded if
loading is underway). It only makes sense to monitor
drivers that are in the process of being loaded or
reloaded. One cannot monitor a future-to-be driver
name for loading, that will only result in a
<c>'DOWN'</c> message being immediately sent.
Monitoring for loading is therefore most useful when
triggered by the <seealso marker="#try_load/3">try_load/3</seealso> function,
where the monitor is created <em>because</em> the
driver is in such a pending state.</p>
<p>Setting a driver monitor for <c>loading</c> will
eventually lead to one of the following messages
being sent:</p>
<taglist>
<tag><em>{'UP', ref(), driver, Name, loaded}</em></tag>
<item>
<p>This message is sent, either immediately if the
driver is already loaded and no reloading is
pending, or when reloading is executed if
reloading is pending. </p>
<p>The <seealso marker="#users">user</seealso> is
expected to know if reloading is demanded prior
to creating a monitor for loading.</p>
</item>
<tag><em>{'UP', ref(), driver, Name, permanent}</em></tag>
<item>
<p>This message will be sent if reloading was
expected, but the (old) driver made itself
permanent prior to reloading. It will also be
sent if the driver was permanent or statically
linked in when trying to create the monitor.</p>
</item>
<tag><em>{'DOWN', ref(), driver, Name, load_cancelled}</em></tag>
<item>
<p>This message will arrive if reloading was
underway, but the <seealso marker="#users">user</seealso> having requested
reload cancelled it by either dying or calling
<seealso marker="#try_unload/2">try_unload/2</seealso>
(or <c>unload/1</c>/<c>unload_driver/1</c>)
again before it was reloaded.</p>
</item>
<tag><em>{'DOWN', ref(), driver, Name, {load_failure, Failure}}</em></tag>
<item>
<p>This message will arrive if reloading was
underway but the loading for some reason
failed. The <c>Failure</c> term is one of the
errors that can be returned from <seealso marker="#try_load/3">try_load/3</seealso>. The
error term can be passed to <seealso marker="#format_error/1">format_error/1</seealso>
for translation into human readable form. Note
that the translation has to be done in the same
running erlang virtual machine as the error
was detected in.</p>
</item>
</taglist>
</item>
<tag><em>unloaded</em></tag>
<item>
<p>Monitor when a driver gets unloaded. If one
monitors a driver that is not present in the system,
one will immediately get notified that the driver got
unloaded. There is no guarantee that the driver was
actually ever loaded.</p>
<p>A driver monitor for unload will eventually result
in one of the following messages being sent:</p>
<taglist>
<tag><em>{'DOWN', ref(), driver, Name, unloaded}</em></tag>
<item>
<p>The driver instance monitored is now
unloaded. As the unload might have been due to a
<c>reload/2</c> request, the driver might once
again have been loaded when this message
arrives.</p>
</item>
<tag><em>{'UP', ref(), driver, Name, unload_cancelled}</em></tag>
<item>
<p>This message will be sent if unloading was
expected, but while the driver was waiting for
all ports to get closed, a new <seealso marker="#users">user</seealso> of the driver
appeared and the unloading was cancelled.</p>
<p>This message appears when an <c>{ok, pending_driver}</c>) was returned from <seealso marker="#try_unload/2">try_unload/2</seealso>)
for the last <seealso marker="#users">user</seealso> of the driver and
then a <c>{ok, already_loaded}</c> is returned
from a call to <seealso marker="#try_load/3">try_load/3</seealso>.</p>
<p>If one wants to <em>really</em> monitor when the
driver gets unloaded, this message will distort
the picture, no unloading was really done.
The <c>unloaded_only</c> option creates a monitor
similar to an <c>unloaded</c> monitor, but does
never result in this message.</p>
</item>
<tag><em>{'UP', ref(), driver, Name, permanent}</em></tag>
<item>
<p>This message will be sent if unloading was
expected, but the driver made itself
permanent prior to unloading. It will also be
sent if trying to monitor a permanent or
statically linked in driver.</p>
</item>
</taglist>
</item>
<tag><em>unloaded_only</em></tag>
<item>
<p>A monitor created as <c>unloaded_only</c> behaves
exactly as one created as <c>unloaded</c> with the
exception that the <c>{'UP', ref(), driver, Name, unload_cancelled}</c> message will never be
sent, but the monitor instead persists until the
driver <em>really</em> gets unloaded.</p>
</item>
</taglist>
</item>
</taglist>
<p>The function throws a <c>badarg</c> exception if the
parameters are not given as described above. </p>
</desc>
</func>
<func>
<name>reload(Path, Name) -> ok | {error, ErrorDesc}</name>
<fsummary>Replace a driver</fsummary>
<type>
<v>Path = Name = string() | atom()</v>
<v>ErrorDesc = pending_process | OpaqueError</v>
<v>OpaqueError = term()</v>
</type>
<desc>
<p>Reloads the driver named <c>Name</c> from a possibly
different <c>Path</c> than was previously used. This
function is used in the code change <seealso marker="#scenarios">scenario</seealso> described in the
introduction.</p>
<p>If there are other <seealso marker="#users">users</seealso>
of this driver, the function will return <c>{error, pending_process}</c>, but if there are no more users, the
function call will hang until all open ports are closed.</p>
<note>
<p>Avoid mixing
several <seealso marker="#users">users</seealso>
with driver reload requests.</p>
</note>
<p>If one wants to avoid hanging on open ports, one should use
the <seealso marker="#try_load/3">try_load/3</seealso>
function instead.</p>
<p>The <c>Name</c> and <c>Path</c> parameters have exactly the
same meaning as when calling the plain <seealso marker="#load/2">load/2</seealso> function.</p>
<note>
<p>Avoid mixing
several <seealso marker="#users">users</seealso>
with driver reload requests.</p>
</note>
<p>On success, the function returns <c>ok</c>. On
failure, the function returns an opaque error, with the
exception of the <c>pending_process</c> error described
above. The opaque errors are to be translated into human
readable form by the <seealso marker="#format_error/1">format_error/1</seealso> function.</p>
<p>For more control over the error handling, again use the
<seealso marker="#try_load/3">try_load/3</seealso>
interface instead.</p>
<p>The function throws a <c>badarg</c> exception if the
parameters are not given as described above. </p>
</desc>
</func>
<func>
<name>reload_driver(Path, Name) -> ok | {error, ErrorDesc}</name>
<fsummary>Replace a driver</fsummary>
<type>
<v>Path = Name = string() | atom()</v>
<v>ErrorDesc = pending_process | OpaqueError</v>
<v>OpaqueError = term()</v>
</type>
<desc>
<p>Works exactly as <seealso marker="#reload/2">reload/2</seealso>, but for drivers
loaded with the <seealso marker="#load_driver/2">load_driver/2</seealso> interface. </p>
<p>As this interface implies that ports are being killed when
the last user disappears, the function wont hang waiting for
ports to get closed.</p>
<p>For further details, see the <seealso marker="#scenarios">scenarios</seealso> in the module
description and refer to the <seealso marker="#reload/2">reload/2</seealso> function description.</p>
<p>The function throws a <c>badarg</c> exception if the
parameters are not given as described above. </p>
</desc>
</func>
<func>
<name>try_load(Path, Name, OptionList) -> {ok,Status} | {ok, PendingStatus, Ref} | {error, ErrorDesc}</name>
<fsummary>Load a driver</fsummary>
<type>
<v>Path = Name = string() | atom()</v>
<v>OptionList = [ Option ]</v>
<v>Option = {driver_options, DriverOptionList} | {monitor, MonitorOption} | {reload, ReloadOption}</v>
<v>DriverOptionList = [ DriverOption ]</v>
<v>DriverOption = kill_ports</v>
<v>MonitorOption = pending_driver | pending</v>
<v>ReloadOption = pending_driver | pending</v>
<v>Status = loaded | already_loaded | PendingStatus </v>
<v>PendingStatus = pending_driver | pending_process</v>
<v>Ref = ref()</v>
<v>ErrorDesc = ErrorAtom | OpaqueError</v>
<v>ErrorAtom = linked_in_driver | inconsistent | permanent | not_loaded_by_this_process | not_loaded | pending_reload | pending_process</v>
</type>
<desc>
<p>This function provides more control than the
<c>load/2</c>/<c>reload/2</c> and
<c>load_driver/2</c>/<c>reload_driver/2</c> interfaces. It
will never wait for completion of other operations related
to the driver, but immediately return the status of the
driver as either:</p>
<taglist>
<tag><em>{ok, loaded}</em></tag>
<item>
<p>The driver was actually loaded and is immediately
usable.</p>
</item>
<tag><em>{ok, already_loaded}</em></tag>
<item>
<p>The driver was already loaded by another process
and/or is in use by a living port. The load by you is
registered and a corresponding <c>try_unload</c> is
expected sometime in the future.</p>
</item>
<tag><em>{ok, pending_driver}</em>or <em>{ok, pending_driver, ref()}</em></tag>
<item>
<p>The load request is registered, but the loading is
delayed due to the fact that an earlier instance of the
driver is still waiting to get unloaded (there are open
ports using it). Still, unload is expected when you are
done with the driver. This return value will
<em>mostly</em> happen when the
<c>{reload,pending_driver}</c> or
<c>{reload,pending}</c> options are used, but
<em>can</em> happen when another <seealso marker="#users">user</seealso> is unloading a driver in
parallel and the <c>kill_ports</c> driver option is
set. In other words, this return value will always need
to be handled!</p>
</item>
<tag><em>{ok, pending_process}</em>or <em>{ok, pending_process, ref()}</em></tag>
<item>
<p>The load request is registered, but the loading is
delayed due to the fact that an earlier instance of the
driver is still waiting to get unloaded by another
<seealso marker="#users">user</seealso> (not only by a
port, in which case <c>{ok,pending_driver}</c> would
have been returned). Still, unload is expected when you
are done with the driver. This return value will
<em>only</em> happen when the <c>{reload,pending}</c>
option is used.</p>
</item>
</taglist>
<p>When the function returns <c>{ok, pending_driver}</c> or
<c>{ok, pending_process}</c>, one might want to get information
about when the driver is <em>actually</em> loaded. This can
be achieved by using the <c>{monitor, PendingOption}</c> option.</p>
<p>When monitoring is requested, and a corresponding <c>{ok, pending_driver}</c> or <c>{ok, pending_process}</c> would be
returned, the function will instead return a tuple <c>{ok, PendingStatus, ref()}</c> and the process will, at a later
time when the driver actually gets loaded, get a monitor
message. The monitor message one can expect is described in
the <seealso marker="#monitor/2">monitor/2</seealso>
function description. </p>
<note>
<p>Note that in case of loading, monitoring can
<em>not</em> only get triggered by using the <c>{reload, ReloadOption}</c> option, but also in special cases where
the load-error is transient, why <c>{monitor, pending_driver}</c> should be used under basically
<em>all</em> real world circumstances!</p>
</note>
<p>The function accepts the following parameters:</p>
<taglist>
<tag><em>Path</em></tag>
<item>
<p>The filesystem path to the directory where the driver
object file is situated. The filename of the object file
(minus extension) must correspond to the driver name
(used in the name parameter) and the driver must
identify itself with the very same name. The
<c>Path</c> might be provided as an <em>io_list</em>,
meaning it can be a list of other io_lists, characters
(eight bit integers) or binaries, all to be flattened
into a sequence of characters.</p>
<p>The (possibly flattened) <c>Path</c> parameter must be
consistent throughout the system, a driver should, by
all <seealso marker="#users">users</seealso>, be loaded
using the same <em>literal</em><c>Path</c>. The
exception is when <em>reloading</em> is requested, in
which case the <c>Path</c> may be specified
differently. Note that all <seealso marker="#users">users</seealso> trying to load the
driver at a later time will need to use the <em>new</em><c>Path</c> if the <c>Path</c> is changed using a
<c>reload</c> option. This is yet another reason
to have <em>only one loader</em> of a driver one wants to
upgrade in a running system! </p>
</item>
<tag><em>Name</em></tag>
<item>
<p>The name parameter is the name of the driver to be used
in subsequent calls to <seealso marker="erts:erlang#open_port/2">open_port</seealso>. The
name can be specified either as an <c>io_list()</c> or
as an <c>atom()</c>. The name given when loading is used
to find the actual object file (with the
help of the <c>Path</c> and the system implied
extension suffix, i.e. <c>.so</c>). The name by which
the driver identifies itself must also be consistent
with this <c>Name</c> parameter, much as a beam-file's
module name much correspond to it's filename.</p>
</item>
<tag><em>OptionList</em></tag>
<item>
<p>A number of options can be specified to control the
loading operation. The options are given as a list of
two-tuples, the tuples having the following values and
meanings:</p>
<taglist>
<tag><em>{driver_options, DriverOptionsList}</em></tag>
<item>
<p>This option is to provide options that will change
it's general behavior and will "stick" to the driver
throughout it's lifespan.</p>
<p>The driver options for a given driver name need
always to be consistent, <em>even when the driver is reloaded</em>, meaning that they are as much a part
of the driver as the actual name.</p>
<p>Currently the only allowed driver option is
<c>kill_ports</c>, which means that all ports opened
towards the driver are killed with the exit-reason
<c>driver_unloaded</c> when no process any longer
has the driver loaded. This situation arises either
when the last <seealso marker="#users">user</seealso> calls <seealso marker="#try_unload/2">try_unload/2</seealso>, or
the last process having loaded the driver exits.</p>
</item>
<tag><em>{monitor, MonitorOption}</em></tag>
<item>
<p>A <c>MonitorOption</c> tells <c>try_load/3</c> to
trigger a driver monitor under certain
conditions. When the monitor is triggered, the
function will return a three-tuple <c>{ok, PendingStatus, ref()}</c>, where the <c>ref()</c> is
the monitor ref for the driver monitor.</p>
<p>Only one <c>MonitorOption</c> can be specified and
it is either the atom <c>pending</c>, which means
that a monitor should be created whenever a load
operation is delayed, and the atom
<c>pending_driver</c>, in which a monitor is
created whenever the operation is delayed due to
open ports towards an otherwise unused driver. The
<c>pending_driver</c> option is of little use, but
is present for completeness, it is very well defined
which reload-options might give rise to which
delays. It might, however, be a good idea to use the
same <c>MonitorOption</c> as the <c>ReloadOption</c>
if present.</p>
<p>If reloading is not requested, it might still be
useful to specify the <c>monitor</c> option, as
forced unloads (<c>kill_ports</c> driver option or
the <c>kill_ports</c> option to <seealso marker="#try_unload/2">try_unload/2</seealso>) will
trigger a transient state where driver loading
cannot be performed until all closing ports are
actually closed. So, as <c>try_unload</c> can, in
almost all situations, return <c>{ok, pending_driver}</c>, one should always specify at least
<c>{monitor, pending_driver}</c> in production
code (see the monitor discussion above). </p>
</item>
<tag><em>{reload,RealoadOption}</em></tag>
<item>
<p>This option is used when one wants to
<em>reload</em> a driver from disk, most often in a
code upgrade scenario. Having a <c>reload</c> option
also implies that the <c>Path</c> parameter need
<em>not</em> be consistent with earlier loads of
the driver.</p>
<p>To reload a driver, the process needs to have previously
loaded the driver, i.e there has to be an active <seealso marker="#users">user</seealso> of the driver in the process. </p>
<p>The <c>reload</c> option can be either the atom
<c>pending</c>, in which reloading is requested for
any driver and will be effectuated when <em>all</em>
ports opened against the driver are closed. The
replacement of the driver will in this case take
place regardless of if there are still
pending <seealso marker="#users">users</seealso>
having the driver loaded!
The option also triggers port-killing (if the
<c>kill_ports</c> driver option is used) even though
there are pending users, making it usable for forced
driver replacement, but laying a lot of
responsibility on the driver <seealso marker="#users">users</seealso>. The pending option is
seldom used as one does not want other <seealso marker="#users">users</seealso> to have loaded the
driver when code change is underway. </p>
<p>The more useful option is <c>pending_driver</c>,
which means that reloading will be queued if the
driver is <em>not</em> loaded by any other <seealso marker="#users">users</seealso>, but the driver has
opened ports, in which case <c>{ok, pending_driver}</c> will be returned (a
<c>monitor</c> option is of course recommended).</p>
<p>If the driver is unloaded (not present in the
system), the error code
<c>not_loaded</c> will be returned. The
<c>reload</c> option is intended for when the user
has already loaded the driver in advance.</p>
</item>
</taglist>
</item>
</taglist>
<p>The function might return numerous errors, of which some
only can be returned given a certain combination of options.</p>
<p>A number of errors are opaque and can only be interpreted by
passing them to the <seealso marker="#format_error/1">format_error/1</seealso> function,
but some can be interpreted directly:</p>
<taglist>
<tag><em>{error,linked_in_driver}</em></tag>
<item>
<p>The driver with the specified name is an erlang
statically linked in driver, which cannot be manipulated
with this API.</p>
</item>
<tag><em>{error,inconsistent}</em></tag>
<item>
<p>The driver has already been loaded with either other
<c>DriverOptions</c> or a different <em>literal</em><c>Path</c> argument.</p>
<p>This can happen even if a <c>reload</c> option is given,
if the <c>DriverOptions</c> differ from the current.</p>
</item>
<tag><em>{error, permanent}</em></tag>
<item>
<p>The driver has requested itself to be permanent, making
it behave like an erlang linked in driver and it can no
longer be manipulated with this API.</p>
</item>
<tag><em>{error, pending_process}</em></tag>
<item>
<p>The driver is loaded by other <seealso marker="#users">users</seealso> when the <c>{reload, pending_driver}</c> option was given.</p>
</item>
<tag><em>{error, pending_reload}</em></tag>
<item>
<p>Driver reload is already requested by another <seealso marker="#users">user</seealso> when the <c>{reload, ReloadOption}</c> option was given.</p>
</item>
<tag><em>{error, not_loaded_by_this_process}</em></tag>
<item>
<p>Appears when the <c>reload</c> option is given. The
driver <c>Name</c> is present in the system, but there is no
<seealso marker="#users">user</seealso> of it in this
process.</p>
</item>
<tag><em>{error, not_loaded}</em></tag>
<item>
<p>Appears when the <c>reload</c> option is given. The
driver <c>Name</c> is not in the system. Only drivers
loaded by this process can be reloaded.</p>
</item>
</taglist>
<p>All other error codes are to be translated by the <seealso marker="#format_error/1">format_error/1</seealso>
function. Note that calls to <c>format_error</c> should be
performed from the same running instance of the erlang
virtual machine as the error was detected in, due to system
dependent behavior concerning error values.</p>
<p>If the arguments or options are malformed, the function will
throw a <c>badarg</c> exception.</p>
</desc>
</func>
<func>
<name>try_unload(Name, OptionList) -> {ok,Status} | {ok, PendingStatus, Ref} | {error, ErrorAtom}</name>
<fsummary>Unload a driver</fsummary>
<type>
<v>Name = string() | atom()</v>
<v>OptionList = [ Option ]</v>
<v>Option = {monitor, MonitorOption} | kill_ports</v>
<v>MonitorOption = pending_driver | pending</v>
<v>Status = unloaded | PendingStatus </v>
<v>PendingStatus = pending_driver | pending_process</v>
<v>Ref = ref()</v>
<v>ErrorAtom = linked_in_driver | not_loaded | not_loaded_by_this_process | permanent</v>
</type>
<desc>
<p>This is the low level function to unload (or decrement
reference counts of) a driver. It can be used to force port
killing, in much the same way as the driver option
<c>kill_ports</c> implicitly does, and it can trigger a
monitor either due to other <seealso marker="#users">users</seealso> still having the driver
loaded or that there are open ports using the driver.</p>
<p>Unloading can be described as the process of telling the
emulator that this particular part of the code in this
particular process (i.e. this <seealso marker="#users">user</seealso>) no longer needs the
driver. That can, if there are no other users, trigger
actual unloading of the driver, in which case the driver
name disappears from the system and (if possible) the memory
occupied by the driver executable code is reclaimed. If the
driver has the <c>kill_ports</c> option set, or if
<c>kill_ports</c> was specified as an option to this
function, all pending ports using this driver will get
killed when unloading is done by the last <seealso marker="#users">user</seealso>. If no port-killing is
involved and there are open ports, the actual unloading
is delayed until there are no more open ports using the
driver. If, in this case, another <seealso marker="#users">user</seealso> (or even this user) loads the
driver again before the driver is actually unloaded, the
unloading will never take place.</p>
<p>To allow the <seealso marker="#users">user</seealso> that
<em>requests unloading</em> to wait for <em>actual unloading</em> to
take place, <c>monitor</c> triggers can be specified in much
the same way as when loading. As <seealso marker="#users">users</seealso> of this function however
seldom are interested in more than decrementing the
reference counts, monitoring is more seldom needed. If the
<c>kill_ports</c> option is used however, monitor trigging is
crucial, as the ports are not guaranteed to have been killed
until the driver is unloaded, why a monitor should be
triggered for at least the <c>pending_driver</c> case.</p>
<p>The possible monitor messages that can be expected are the
same as when using the <c>unloaded</c> option to the
<seealso marker="#monitor/2">monitor/2</seealso> function.</p>
<p>The function will return one of the following statuses upon
success:</p>
<taglist>
<tag><em>{ok, unloaded}</em></tag>
<item>
<p>The driver was immediately unloaded, meaning that the
driver name is now free to use by other drivers and, if
the underlying OS permits it, the memory occupied by the
driver object code is now reclaimed.</p>
<p>The driver can only be unloaded when there are no open
ports using it and there are no more <seealso marker="#users">users</seealso> requiring it to be
loaded.</p>
</item>
<tag><em>{ok, pending_driver}</em>or <em>{ok, pending_driver, ref()}</em></tag>
<item>
<p>This return value indicates that this call removed the
last <seealso marker="#users">user</seealso> from the
driver, but there are still open ports using it.
When all ports are closed and no new <seealso marker="#users">users</seealso> have arrived, the driver
will actually be reloaded and the name and memory
reclaimed.</p>
<p>This return value is valid even when the option
<c>kill_ports</c> was used, as killing ports may not be
a process that completes immediately. The condition is,
in that case, however transient. Monitors are as always
useful to detect when the driver is really unloaded.</p>
</item>
<tag><em>{ok, pending_process}</em>or <em>{ok, pending_process, ref()}</em></tag>
<item>
<p>The unload request is registered, but there are still
other <seealso marker="#users">users</seealso> holding
the driver. Note that the term <c>pending_process</c>
might refer to the running process, there might be more
than one <seealso marker="#users">user</seealso> in the
same process.</p>
<p>This is a normal, healthy return value if the call was
just placed to inform the emulator that you have no
further use of the driver. It is actually the most
common return value in the most common <seealso marker="#scenarios">scenario</seealso>
described in the introduction.</p>
</item>
</taglist>
<p>The function accepts the following parameters:</p>
<taglist>
<tag><em>Name</em></tag>
<item>
<p>The name parameter is the name of the driver to be
unloaded. The name can be specified either as an
<c>io_list()</c> or as an <c>atom()</c>. </p>
</item>
<tag><em>OptionList</em></tag>
<item>
<p>The <c>OptionList</c> argument can be used to specify
certain behavior regarding ports as well as triggering
monitors under certain conditions:</p>
<taglist>
<tag><em>kill_ports</em></tag>
<item>
<p>Force killing of all ports opened using this driver,
with the exit reason <c>driver_unloaded</c>, if you are
the <em>last</em><seealso marker="#users">user</seealso> of the driver.</p>
<p>If there are other <seealso marker="#users">users</seealso> having the driver
loaded, this option will have no effect.</p>
<p>If one wants the consistent behavior of killing ports
when the last <seealso marker="#users">user</seealso>
unloads, one should use the driver option
<c>kill_ports</c> when loading the driver instead.</p>
</item>
<tag><em>{monitor, MonitorOption}</em></tag>
<item>
<p>This option creates a driver monitor if the condition
given in <c>MonitorOptions</c> is true. The valid
options are:</p>
<taglist>
<tag><em>pending_driver</em></tag>
<item>
<p>Create a driver monitor if the return value is to
be <c>{ok, pending_driver}</c>.</p>
</item>
<tag><em>pending</em></tag>
<item>
<p>Create a monitor if the return value will be either
<c>{ok, pending_driver}</c> or <c>{ok, pending_process}</c>.</p>
</item>
</taglist>
<p>The <c>pending_driver</c><c>MonitorOption</c> is by far
the most useful and it has to be used to ensure that the
driver has really been unloaded and the ports closed
whenever the <c>kill_ports</c> option is used or the
driver may have been loaded with the <c>kill_ports</c>
driver option.</p>
<p>By using the monitor-triggers in the call to
<c>try_unload</c> one can be sure that the monitor is
actually added before the unloading is executed, meaning
that the monitor will always get properly triggered,
which would not be the case if one called
<c>erl_ddll:monitor/2</c> separately.</p>
</item>
</taglist>
</item>
</taglist>
<p>The function may return several error conditions, of which
all are well specified (no opaque values):</p>
<taglist>
<tag><em>{error, linked_in_driver}</em></tag>
<item>
<p>You were trying to unload an erlang statically linked in
driver, which cannot be manipulated with this interface
(and cannot be unloaded at all).</p>
</item>
<tag><em>{error, not_loaded}</em></tag>
<item>
<p>The driver <c>Name</c> is not present in the system.</p>
</item>
<tag><em>{error, not_loaded_by_this_process}</em></tag>
<item>
<p>The driver <c>Name</c> is present in the system, but
there is no <seealso marker="#users">user</seealso> of
it in this process. </p>
<p>As a special case, drivers can be unloaded from
processes that has done no corresponding call to
<c>try_load/3</c> if, and only if, there are <em>no users of the driver at all</em>, which may happen if the
process containing the last user dies.</p>
</item>
<tag><em>{error, permanent}</em></tag>
<item>
<p>The driver has made itself permanent, in which case it
can no longer be manipulated by this interface (much
like a statically linked in driver).</p>
</item>
</taglist>
<p>The function throws a <c>badarg</c> exception if the
parameters are not given as described above. </p>
</desc>
</func>
<func>
<name>unload(Name) -> ok | {error, ErrorDesc}</name>
<fsummary>Unload a driver</fsummary>
<type>
<v>Name = string() | atom()</v>
<v>ErrorDesc = term()</v>
</type>
<desc>
<p>Unloads, or at least dereferences the driver named
<c>Name</c>. If the caller is the last <seealso marker="#users">user</seealso> of the driver, and there
are no more open ports using the driver, the driver will
actually get unloaded. In all other cases, actual unloading
will be delayed until all ports are closed and there are no
remaining <seealso marker="#users">users</seealso>.</p>
<p>If there are other <seealso marker="#users">users</seealso> of the driver, the reference
counts of the driver is merely decreased, so that the caller
is no longer considered a user of the driver. For usage
scenarios, see the <seealso marker="#scenarios">description</seealso> in the beginning
of this document. </p>
<p>The <c>ErrorDesc</c> returned is an opaque value to be
passed further on to the <seealso marker="#format_error/1">format_error/1</seealso>
function. For more control over the operation, use the
<seealso marker="#try_unload/2">try_unload/2</seealso>
interface.</p>
<p>The function throws a <c>badarg</c> exception if the
parameters are not given as described above. </p>
</desc>
</func>
<func>
<name>unload_driver(Name) -> ok | {error, ErrorDesc}</name>
<fsummary>Unload a driver</fsummary>
<type>
<v>Name = string() | atom()</v>
<v>ErrorDesc = term()</v>
</type>
<desc>
<p>Unloads, or at least dereferences the driver named
<c>Name</c>. If the caller is the last <seealso marker="#users">user</seealso> of the driver, all
remaining open ports using the driver will get killed with
the reason <c>driver_unloaded</c> and the driver will
eventually get unloaded.</p>
<p>If there are other <seealso marker="#users">users</seealso>
of the driver, the reference counts of the driver is merely
decreased, so that the caller is no longer considered a
<seealso marker="#users">user</seealso>. For
usage scenarios, see the <seealso marker="#scenarios">description</seealso> in the beginning
of this document.</p>
<p>The <c>ErrorDesc</c> returned is an opaque value to be
passed further on to the <seealso marker="#format_error/1">format_error/1</seealso>
function. For more control over the operation, use the
<seealso marker="#try_unload/2">try_unload/2</seealso>
interface.</p>
<p>The function throws a <c>badarg</c> exception if the
parameters are not given as described above. </p>
</desc>
</func>
<func>
<name>loaded_drivers() -> {ok, Drivers}</name>
<fsummary>List loaded drivers</fsummary>
<type>
<v>Drivers = [Driver()]</v>
<v>Driver = string()</v>
</type>
<desc>
<p>Returns a list of all the available drivers, both
(statically) linked-in and dynamically loaded ones.</p>
<p>The driver names are returned as a list of strings rather
than a list of atoms for historical reasons.</p>
<p>More information about drivers can be obtained using one of
the <seealso marker="#info/0">info</seealso> functions.</p>
</desc>
</func>
<func>
<name>format_error(ErrorDesc) -> string()</name>
<fsummary>Format an error descriptor</fsummary>
<type>
<v>ErrorDesc -- see below</v>
</type>
<desc>
<p>Takes an <c>ErrorDesc</c> returned by load, unload or
reload functions and returns a string which
describes the error or warning.</p>
<note>
<p>Due to peculiarities in the dynamic loading interfaces on
different platform, the returned string is only guaranteed
to describe the correct error <em>if format_error/1 is called in the same instance of the erlang virtual machine as the error appeared in</em> (meaning the same operating
system process)!</p>
</note>
</desc>
</func>
</funcs>
<section>
<title>SEE ALSO</title>
<p>erl_driver(4), driver_entry(4)</p>
</section>
</erlref>
