readme.txt

h.248协议源码

Also, make sure that both the Greenhills compiler and the OSE bin
directory are in the path. This completes the build environment for
Solaris.

On Windows, there are two choices. The first is to download native versions of
all of the individual tools required (see list given in the Build Requirements
section of this readme file). Unfortunatly finding recent versions of
Windows GNU tools is not easy. Once they are acquired place them in a directory
and put that directory in the path along with the Greenhills compiler and
the OSE executable directory. For example:
SET Path=C:\Green;C:\OSE\OSE4.2.2\WIN32\BIN;C:\TOOLS;%PATH%

The second, and probably preferable option for Windows, is to download the
Cygwin GNU package from http://www.cygwin.com. Instructions for setting up and
installing this package can be found on the Cywin web site but here is a set
of "quick-start" instructions:
     1. Go to http://www.cygwin.com
     2. Go to the download page.
     3. Select an FTP mirror site near you.
     4. Download the SETUP.EXE program
     5. run the SETUP.EXE program
     6. Tell Setup that you want to install from the Internet
     7. Select the FTP site you downloaded SETUP.EXE from
     8. By default it will install all Cygwin packages. Unless disk space
        is limited or you have a slow internet connection, just install
        all of them.
     9. Make sure to tell setup to install into the C:\cygwin directory (you
        can use any drive letter). Note: make sure it doesn't try to install
        into C:\, or things could get messy.
     10. Tell Setup to use DOS text by default
     11. Wait for all packages to install and exit Setup. You can put
         the icons on the desktop and the Start Menu if you like. It's
         a good idea to have it install the icons on one or the other so you 
         see how to start a bourne command shell, even though building this 
         toolkit will not be done from that shell (although make will use the 
         shell itself).
     12. Place the cygwin bin directory (ie C:\cygwin\bin) at the front of
         the Windows path. Putting it at the front is important so
         that the proper commands (like sort, for example) are called.
	 This can be done only when building this toolkit or on a system
         wide basis since the Cygwin tools can be very useful for a number
         of things.
Then, put the Greenhills and OSE executable directories at the beginning
of your path. For example:
SET Path=C:\Green;C:\OSE\OSE4.2.2\WIN32\BIN;%PATH%

Finally, set the OSE_ROOT environment variable to the OSE root directory.
Once that is done the build environment is ready.


Build Configuration
-------------------
For OSE, the TARGET_OS should be set to ose and the COMPILER should be
set to ghs. Set HOST_OS to solaris or win32 as approriate. The
TARGET_FAMILY, TARGET_CPU, and COMPILER_CPU need to be set specifically
based on the target precessor:
     TARGET_FAMILY	This is used to find the proper compiler tools
			provided by ghs. One example is that for the
			Power PC, the compiler is called ccppc, thus
			TARGET_FAMILY needs to be set to ppc.

     TARGET_CPU		This is used to find the proper kernel information
			for the target CPU. It should be set to the path
			of the kernel directory to be used relative to the
			OSE root directory. Make sure to use a "/" as a
			directory seperator even on a windows host. One
			example is that for a Power PC 8260 CPU, TARGET_CPU
			should be set to powerpc/krn-82xx.

     COMPILER_CPU	This is used to tell the compiler what CPU to
			compile for. This is based on the allowed settings
			for the GHS compiler (via the -cpu option) and is
			described in the GHS documentation. One example is
			that for a Power PC 8260, it should be set to ppc8260.

Also note that the compiler or OSE may require additional switches for a
paraticular CPU which must be defined on the CFLAGS line of the default.mak
file.

Here is an example of the Build Parameters and CFLAGS settings for a Power PC
8260 being built in debug mode on a Windows host for a OSE target. In this
example, the -DBIG_ENDIAN is required because the OSE needs to know what mode
the CPU is operating in. This example could be copied directly into the
default.mak for this particular environment:

# Build Parameters
TYPE ?= debug
TARGET_OS ?= ose
HOST_OS ?= win32
TARGET_FAMILY ?= ppc
TARGET_CPU ?= powerpc/krn-82xx
COMPILER_CPU ?= ppc8260
COMPILER ?= ghs

# Additional special compiler flags
CFLAGS := -DBIG_ENDIAN


Building the Toolkit
-------------------------
Follow the standard command line build instructions as described in
this readme file.


Linking with the Toolkit Libraries
---------------------------------------
There is one board specific piece of information that the Toolkit
requires for operation that must be provided by the implementation.
It is the hostname of the target system and should be set in the
environment variable HOSTNAME. The hostname must be set in the process
block of any process making a call to any of the libraries. For
example, for a process to set the hostname to "EST8260" for the current
process block, the HOSTNAME environment variable would set as follows:

set_env(get_bid(current_process()), "HOSTNAME", "EST8260");

In order to start the megacomg application, simply call the following function:
void appMain();

There are currently no parameters to appMain. Configuration
parameters must be set in the  megaco/client/megacomg/rvmegaco.c file itself via a
number of #define parameters. Once called the application will start and
this function never returns.

This application can easily be run from the OSE shell. A good place to start
is the OSE inetutil demo. You will need to add pthreads to the OSE
configuration and add the megacomg command to the shell command list. The command
can simply call a function such as this:

COMMAND(megacomg)
{
	set_env(get_bid(current_process()), "HOSTNAME", "EST8260"); /* set host name */
	appMain();
	return RET_SUCCESS;
}

The inetutil makefile will also have to be modified to link with the megaco
toolkit libraries.

From that application, to start the megacomgc application, simply call
the following function:

void appMain(void) 

The mgc address is taken from RV_MGC_IP, and the terminations information 
is taken from the configtable structure.



8. Nucleus Target Environment
==========================

Supported Target OS's
---------------------
Nucleus Plus 1.13.2 or higher with Net 4.4.2 or higher
IMPORTANT: Net requires these two patches:
        spr 51 (year 2001) : fix for name conflict of INT_MAX and LONG_MAX macros
        spr 182 (year 2001) : fix for preventing bind of duplicate port numbers
    Both of these are being incorprated into future version of Net. Check
    the history.txt file to determine if they are present in your version,
    if not, contact ATI for these fixes.

Supported Host OS's
-------------------
Solaris 7**
Solaris 8**
Windows NT 4.0
Windows 2000

**Note: ATI currently supports only Windows so a functional Nucleus
environment would have to be built first.


Supported compiler's
--------------------
Diab 4.3g


Build Environment
------------------
The first thing that must be done to set up the environment is
to install and build Nucleus Plus and Net. Make sure the
LM_LICENSE variable is set up properly for the diab compiler.

On Solaris, simply set the following environment variable:
NUCLEUS_ROOT	- NUCLEUS root directory
Use a forward slash ("/") for directory paths, even on win32 hosts.

Also, make sure that both the diab compiler directory is in the path.
This completes the build environment for Solaris.

On Windows, there are two choices. The first is to download native versions of
all of the individual tools required (see list given in the Build Requirements
section of this readme file). Unfortunatly finding recent versions of
Windows GNU tools is not easy. Once they are acquired place them in a directory
and put that directory in the path along with the Diab compiler directory.
For example:
SET Path=C:\diab\4.3g\WIN32\bin;%PATH%

The second, and probably preferable option for Windows, is to download the
Cygwin GNU package from http://www.cygwin.com. Instructions for setting up and
installing this package can be found on the Cywin web site but here is a set
of "quick-start" instructions:
     1. Go to http://www.cygwin.com
     2. Go to the download page.
     3. Select an FTP mirror site near you.
     4. Download the SETUP.EXE program
     5. run the SETUP.EXE program
     6. Tell Setup that you want to install from the Internet
     7. Select the FTP site you downloaded SETUP.EXE from
     8. By default it will install all Cygwin packages. Unless disk space
        is limited or you have a slow internet connection, just install
        all of them.
     9. Make sure to tell setup to install into the C:\cygwin directory (you
        can use any drive letter). Note: make sure it doesn't try to install
        into C:\, or things could get messy.
     10. Tell Setup to use DOS text by default
     11. Wait for all packages to install and exit Setup. You can put
         the icons on the desktop and the Start Menu if you like. It's
         a good idea to have it install the icons on one or the other so you 
         see how to start a bourne command shell, even though building this 
         toolkit will not be done from that shell (although make will use the 
         shell itself).
     12. Place the cygwin bin directory (ie C:\cygwin\bin) at the front of
         the Windows path. Putting it at the front is important so
         that the proper commands (like sort, for example) are called.
	 This can be done only when building this toolkit or on a system
         wide basis since the Cygwin tools can be very useful for a number
         of things.
Then put the Diab executable directories at the beginning of you path.
For example:
SET Path=C:\diab\4.3g\WIN32\bin;%PATH%

Finally, set the NUCLEUS_ROOT environment variable to the NUCLEUS root directory.
Remember to use a forward slash ("/") for directory paths, even on win32 hosts.
Once that is done the build environment is ready.


Build Configuration
-------------------
For Nucleus, the TARGET_OS should be set to nucleus and the COMPILER
should be set to diab. Set HOST_OS to solaris or win32 as approriate.
The COMPILER_CPU needs to be set based on the target CPU as per the Diab
compiler -t option. Thus, for the a Embedded Power PC 603 CPU, it would
be set to PPC603ES (E = elf format and S = software floating point).
Set TARGET_FAMILY and TARGET_CPU to be the same as COMPILER_CPU since
thay are not used by the Diab compiler (and are actually ignored).

Also note that the compiler or Nucleus may require additional switches for a
paraticular CPU which must be defined on the CFLAGS line of the default.mak
file.

Here is an example of the Build Parameters and CFLAGS settings for a Power PC
8260 being built in debug mode on a Windows host for a Nucleus target. 
This example could be copied directly into the default.mak for this
particular environment:

# Build Parameters -- Diab 4.3g (now supports 8260)
TYPE ?= debug
TARGET_OS ?= nucleus
HOST_OS ?= win32
TARGET_FAMILY ?= PPC8260ES
TARGET_CPU ?= PPC8260ES
COMPILER_CPU ?= PPC8260ES
COMPILER ?= diab

Building the Toolkit
-------------------------
Follow the standard command line build instructions as described in
this readme file.


Linking with the Toolkit Libraries
---------------------------------------
There are some additional requirements that the Toolkit requires for
operation. First, the following global variable must be defined:
NU_MEMORY_POOL  System_Memory;

This must be set to the system memory pool or other memory pool with
the NU_Create_Memory_Pool command. This memory pool will be used
by the stacks are to allocate memory from. This variable is normally
set by Nucleus apps so will probably already exist.

No calls to any function within the library can be made until  after
the OS is fully initialized and tasking is enabled. Also, netwoking
(devices and stack) must be initialized along with stdio (and any
related devices). If the default logging facilities will not be used,
then initializing sdtio is optional.

In order to start the megacomg application, simply call the following function:
void appMain();

There are currently no parameters to appMain. Configuration
parameters must be set in the megaco/client/megacomg/rvmegacomg.c file itself via a
number of #define parameters. Once called the application will start and
this function never returns. Remember, do NOT call appMain from the
Application_Initialize call but after everthing (OS, NET, stdio) is
initialized and operating.

In order to start the mgc application, simply call the following function:
void appMain(void) 

The mgc address is taken from RV_MGC_IP, and the terminations information 
is taken from the configtable structure.