Volatile Properties

In addition to PICL_READ and PICL_WRITE property access modes, the plug-in modules specify whether a property is volatile or not by setting the bit PICL_VOLATILE.

#define   PICL_VOLATILE   0x4

For a volatile property, the plug-in module provides the access functions to read and/or write the property in the ptree_propinfo_t argument passed when creating the property.

The daemon invokes the access functions of volatile properties when clients access their values. Two arguments are passed to the read access functions. The first argument is a pointer to ptree_rarg_t, which contains the handle of the node, the handle of the accessed property and the credentials of the caller. The second argument is a pointer to the buffer where the value is to be copied.

typedef struct {


         picl_nodehdl_t nodeh;


         picl_prophdl_t proph;


         door_cred_t    cred;
} ptree_rarg_t;

The prototype of the read access function for volatile property is:

int read(ptree_rarg_t *rarg, void *buf);

The read function returns PICL_SUCCESS to indicate successful completion.

Similarly, when a write access is performed on a volatile property, the daemon invokes the write access function provided by the plug-in for that property and passes it two arguments. The first argument is a pointer to ptree_warg_t, which contains the handle to the node, the handle of the accessed property and the credentials of the caller. The second argument is a pointer to the buffer containing the value to be written.

typedef struct {


        picl_nodehdl_t  nodeh;


        picl_prophdl_t  proph;


        door_cred_t     cred;
} ptree_warg_t;

The prototype of the write access function for volatile property is:

int write(ptree_warg_t *warg, const void *buf);

The write function returns PICL_SUCCESS to indicate successful completion.

For all volatile properties, the 'size' of the property must be specified to be the maximum possible size of the value. The maximum size of the value cannot exceed PICL_PROPSIZE_MAX. This allows a client to allocate a sufficiently large buffer before retrieving a volatile property's value.

Plug-in Modules

Plug-in modules are shared objects that are located in well-known directories for the daemon to locate and load them. Plug-in modules are located in one of the following plug-in directories depending on the platform-specific nature of the data they collect and publish.

/usr/platform/picl/plugins/`uname -i`/
/usr/platform/picl/plugins/`uname -m`/
/usr/lib/picl/plugins/

A plug-in module may specify its dependency on another plug-in module using the -l linker option. The plug-ins are loaded by the PICL daemon using dlopen(3C) according to the specified dependencies. Each plug-in module must define a .init section, which is executed when the plug-in module is loaded, to register themselves with the daemon. See picld_plugin_register(3PICLTREE) for more information on plug-in registration.

The plug-in modules may use the picld_log(3PICLTREE) function to log their messages to the system log file.