I agree with Chuck's assessment.

I think the databuffer SendData, because it is sent from the MIPS, has a
pointer that was not allocated from the TM.

Associating that databuffer with a TSSA Packet, and subsequently tearing
down the graph, probably means the pointer is being freed and this is not
allowed.

I think the memcpy is probably the best way to handle this situation, where
data has been passed across the MIPS-TM boundary.
It may be an assumption that the copying is critically wasteful but I would
check this to make sure, it may be fine.

Chris
========================
Chris Bore
BORES Signal Processing
chris@...
www.bores.com
Tel:    +44 1483 740138
Mobile: +44 7921 153 219

-----Original Message-----
From: trimedia@yahoogroups.com [mailto:trimedia@yahoogroups.com] On Behalf
Of Chuck Peplinski
Sent: 20 June 2007 14:16
To: trimedia@yahoogroups.com
Subject: Re: [trimedia] memcpy in TSSA

On TCSHelp, there is an FAQ about how to debug an RSE.  Because the
exception seems to happen at tear down, I'm guessing that the TM is trying
to free a pointer that it did not allocate.  You can look up the address at
which the RSE occurred and check this.  Alternatively, it may be that the
address used by the MIPS is not located inside of the TM's aperture.  If
that were the case, the RSE would happen when the TM tried to write into
that buffer.
In either case, the RSE tells you the PC where the exception occurred.
Look up that address.

     Chuck

Chuck Peplinski
TriMedia Tech Support for NXP Semiconductors  www.TCSHelp.com
TriMedian at MDS 			 www.mds.com




nastooh_avesta wrote:
>
>
> Hello All,
> I have used phStbDspTransfer as a model to pass data, received from
> MIPS, via RPC, to the rest of the TSSA chain. The problem is that when
> I try to prune the code, I get an exception. Specifically, within the
> original code there is a memcpy statement that copies the content of
> the received data into the outgoing packet, i.e., ....
> err = tmPacket_GetBufferAddr(pPacket, buffersUsed,
> (Pointer*)&pPacketData); ....
> memcpy(pPacketData,(unsigned char *) SendData, sizeToWrite); ....
> where SendData is the char buffer received from MIPS, which is to be
> sent to the rest of the TSSA chain. The above looked a bit redundant
> to me, so I modified it as follows:
> err = tmPacket_SetBufferAddr(pPacket, buffersUsed, (Pointer)SendData).
>
> The main body of the client program, running on the MIPS, works fine.
> That is, data is sent to TM0 and passed through the TSSA chain, and
> eventually is written into a file, via phStbFileWrite. The exception
> occurs, when one attempts to tear down the graph. Here is my code
> fragment:
> ....
> DBG_PRINT0(DBG_ROOT,DBG_ALL_FLAGS,"Tearing down stream graph\n");
> phStbStreamingSystem_Server_Stop(graph);
> DBG_PRINT0(DBG_ROOT,DBG_ALL_FLAGS,"Graph Stopped\n");
> phStbStreamingSystem_Server_Disconnect(graph,Connection01);
> DBG_PRINT0(DBG_ROOT,DBG_ALL_FLAGS,"Disconnected Graph\n");
> /* Commit the new graph */
> phStbStreamingSystem_Server_Commit(graph);
> DBG_PRINT0(DBG_ROOT,DBG_ALL_FLAGS,"Graph Committed\n");
> /* Remove all the resources */
> phStbStreamingSystem_Server_RemoveResource(graph,MipsReadResource);
> DBG_PRINT0(DBG_ROOT,DBG_ALL_FLAGS,"Removed MipsReadResource\n");
> !!!!!!!!!!!!!!!! Exception happens next !!!!!!!!!!!!!!!!!!!!!!!!!!
> phStbStreamingSystem_Server_RemoveResource(graph,FileWriteResource);
> DBG_PRINT0(DBG_ROOT,DBG_ALL_FLAGS,"Removed FileWriteResource\n");
>
> I receive the following from TM0 DP:
>
> TIME:1-12:02:50.840;TASK:0x00BC0000;TICK:0x2A9FDB9A;
> [tmbslPnx8550_RseExceptionHandler] CPU RSE reserved store exception !
> ERROR: CPU RSE exception cause: ILSTO - Illegal data store/write
> address RSE_PC Program Counter address: 0x03D3061A RSE_CAUSE status
> register value: 0x000F0002 Interrupt context (!0 = In ISR): 0x00000000
>
> pSOS Task/Stack Information:
> Task ID: 0x00BC0000
> Task Name: [T010]
> Stack high addr: 0x040F4F68
> Stack pointer addr: 0x03E50552
> Stack low addr: 0x040EEEEC
> Stack max usage: 0x000009F4
>
> pSOS Task Stack Data Dump (Count: 64 UInt32s) ___Addr____ ___0x0____
> ___0x4____ ___0x8____ ___0xC____
> 0x03E50552: 0x76250063 0x69737265 0x203A6E6F 0x00252037
> 0x03E50562: 0x61654843 0x6C415F70 0x00636F6C 0x61654843
> 0x03E50572: 0x73250070 0x002B000A 0x30097325 0x50582578
> 0x03E50582: 0x4863000A 0x00706165 0x61654863 0x3D212070
> 0x03E50592: 0x70003020 0x21207274 0x0030203D 0x657A6973
> 0x03E505A2: 0x203D2120 0x73250030 0x0A642509 0x7A697300
> 0x03E505B2: 0x72650065 0x002D0072 0x61654843 0x61435F70
> 0x03E505C2: 0x636F6C6C 0x65484300 0x465F7061 0x00656572
> 0x03E505D2: 0x48436D74 0x5F706165 0x74736544 0x00796F72
> 0x03E505E2: 0x48436D74 0x5F706165 0x61657243 0x63006574
> 0x03E505F2: 0x74616572 0x3D212065 0x2A003020 0x61657263
> 0x03E50602: 0x43006574 0x6B636F6C 0x30203220 0x00353420
> 0x03E50612: 0x03E50603 0x00000001 0x00000002 0x00000000
> 0x03E50622: 0x636F6C43 0x54535F6B 0x5F455441 0x55544553
> 0x03E50632: 0x6C430050 0x5F6B636F 0x54415453 0x55525F45
> 0x03E50642: 0x4E494E4E 0x6D740047 0x636F6C43 0x00632E6B
>
> 1-12:02:50.000;ROOT;tmMain: Fatal error in
> src/tm_psos/tmbslPnx8550_OsSpecific.c(1353)
>
> Do note that the exception did NOT occur in my routine, but rather
> when removing phStbFileWrite resource. I did change the order of the
> following statements, ie.,
> phStbStreamingSystem_Server_RemoveResource(graph,FileWriteResource);
> phStbStreamingSystem_Server_RemoveResource(graph,MipsReadResource);
> to no avail. (Thinking there might be a memory allocation issue, while
> freeing the resource.) However, phalReadMipsRPCClose(Int
> instance) behaves nicely, execution proceeds, without a hitch, through :
> ...
> defCap.numCurrentInstances--;
> tmosalSystemMutexExit();
> tmDefault_Free (ivp->publicSetup->defaultSetup); tmDefault_Free
> (ivp->publicSetup); tmDefault_Free (ivp); tmTr_K (-); return rval;
> Once again graph is dismantled nicely if I follow the original code,
> using memcpy.
> Any thought, as how to avoid the memcpy overhead?
>
>


[Non-text portions of this message have been removed]



Sent via the TriMedia mailing list
trimedia@yahoogroups.com - archive at http://groups.yahoo.com/group/trimedia
Yahoo! Groups Links

Hello All,
I have a working program that reads a customized file format and
writes the extracted data into a file via phStbFileWrite. I've been
trying to modify this program to, instead, pass the data to a video
render, for display. I am not sure if I am going about it correctly,
though. Here are some excerpts of the program, which is loosely based
on phStbDFBVideoProviderMpeg4.c:

phStbReadMipsRPC_Init(); /*Initialize the file reader */
phStbStreamingSystem_Server_CreateStreamGraph(&common.graph);
DFBVideoProviderCommonInitialiseVideoLayer(&common,&syncInfo,&clockID,resourceNa\
me
) ;
DFBVideoProviderCommonInitialiseEventListener(&common,resourceName,0xFFFFFFFF);
phStbReadMipsRPC_Create(&ReadMipsRPC,resourceName );
phStbStreamingSystem_Server_AddResource(common.graph, resourceName,
&common.resources[VIDEO_DECODER]);
phStbStreamingSystem_Server_Connect(common.graph,
                                         common.resources[VIDEO_DECODER],0,
                                         common.resources[VIDEO_LAYER],0,
&common.connections[VIDEO_DECODER_TO_VIDEO_RENDERER]);

DFBVideoProviderCommonCreateCommitStreamGraph(&common,PH_STBVIDEOPROVIDER_VIDEO_\
READY_TIMEOUT);
….
    Parse the data and pass it to TSSA chain
….
phStbStreamingSystem_Server_Stop(common.graph);
phStbVideoRenderer_GetVideoLayerResource(common.pRendInst,common.layerID,&pVidLa\
yer,resourceName);
phStbStreamingSystem_Server_DumpResource(resourceName);
phStbStreamingSystem_Server_Commit(common.graph);
DFBVideoProviderCommonRemoveCommonResources(&common);
phStbReadMipsRPC_Destroy(ReadMipsRPC);
phStbStreamingSystem_Server_DestroyStreamGraph(common.graph);
….

And I am getting the following compilation errors:
[root@localhost stb810-SP6]# cbuild :mt
[root@localhost stb810-SP6]# cbuild -m exStbFXVDisplay

## Selecting CI exStbFXVDisplay:
"/STB810/MDS_STB810/SP6_RC4_Base/stb810-SP6/src/project/apps/exStbFXVDisplay"
make --no-print-directory
CXX
/STB810/MDS_STB810/SP6_RC4_Base/stb810-SP6/build_128M_NOWMV/apps/exStbFXVDisplay\
/src/exStbFXVDisplay.o
In file included from src/exStbFXVDisplay.cpp:13:
"/STB810/MDS_STB810/SP6_RC4_Base/stb810-SP6/build_128M_NOWMV/staging/opt/nxp/inc\
lude/phStbEvent.h",
line 37:24: error:
tmNxCompId.h: No such file or directory

/STB810/MDS_STB810/SP6_RC4_Base/stb810-SP6/build_128M_NOWMV/staging/usr/local/in\
clude/directfb-internal/core/surfaces.h:
In function 'void* dfb_surface_data_offset(const CoreSurface*, void*,
int, int, int)':
"/STB810/MDS_STB810/SP6_RC4_Base/stb810-SP6/build_128M_NOWMV/staging/usr/local/i\
nclude/directfb-internal/core/surfaces.h",
line 209: error: pointer of type 'void *' used in arithmetic
"/STB810/MDS_STB810/SP6_RC4_Base/stb810-SP6/build_128M_NOWMV/staging/usr/local/i\
nclude/directfb-internal/core/surfaces.h",
line 209: error: pointer of type 'void *' used in arithmetic
make: ***
[/STB810/MDS_STB810/SP6_RC4_Base/stb810-SP6/build_128M_NOWMV/apps/exStbFXVDispla\
y/src/exStbFXVDisplay.o]
Error 1

**ERROR: Returned 512: make --no-print-directory
Directory was:
/STB810/MDS_STB810/SP6_RC4_Base/stb810-SP6/src/project/apps/exStbFXVDisplay

**FATAL: make returned error code 512
Errors: 2
/*************************************

Evidently, tmNxCompId.h, included in phStbDFBVideoProviderCommon.h, is
a Trimedia component and my test program is a Linux program residing
in … stb810-SP6/src/project/apps/. However, I do need
phStbDFBVideoProviderCommon.h for #defs and function prototypes. As
for the second error, … .
Is this a correct approach or should I be using the following, as in
exStbDFBVideo.
     DFBCHECK( DirectFBInit(&argc, &argv) );
     /* Create the DirectFB root interface. */
     DFBCHECK( DirectFBCreate(&pDfb) );
     /* Check command line parameters, get video layer number */
     videoLayer = atoi(argv[argFileParam + 1]);
…
err = pDfb->CreateVideoProvider(pDfb, argv[argFileParam], &gpProvider);
….
If so, how would I create a new video provider for
pDfb->CreateVideoProvider?

My mds pci card often make my PC break down!

I use NDK5.5 and mds's latest version driver which download from
mds's site.Now everytime I run tmdbg,the PC will break down just
when the TMDBG's  GUI appear or when I start to download program to
my board.

Several months ago my MDS jtag card didn't work and was sent to
repair .MDS seemed to give me a new card instead of reparing because
it's quite new.
This card seemed not to be the same version as my older one,it make
my PC breakdown once and then.Everytime I must plug the card in
another pci slot and reboot the PC .If I did not plug the card in
another pci slot ,the PC will still breakdown when the TMDBG's gui
appear or start to download program to board.
But yesterday this unplug and replug did not work when my PC
breakdown  again.I even reinstall windows Xp ,ndk5.5 and jtag card
driver,the tmdbg still can't work.

Who can tell me why ?Is this card not a stable one  ?
My old card never make my PC breakdown.
This new card's version is C1.

I wonder if the wrong driver is being used for the newer card.

Please contact directly MDS for assistance.


Best Regards,
Alfred Lee


----- Original Message -----
From: "gongwjh" <GONGXF@...>
To: <trimedia@yahoogroups.com>
Sent: Thursday, June 21, 2007 9:06 AM
Subject: [trimedia] Help about my MDS pci Jtag card


> My mds pci card often make my PC break down!
>
> I use NDK5.5 and mds's latest version driver which download from
> mds's site.Now everytime I run tmdbg,the PC will break down just
> when the TMDBG's  GUI appear or when I start to download program to
> my board.
>
> Several months ago my MDS jtag card didn't work and was sent to
> repair .MDS seemed to give me a new card instead of reparing because
> it's quite new.
> This card seemed not to be the same version as my older one,it make
> my PC breakdown once and then.Everytime I must plug the card in
> another pci slot and reboot the PC .If I did not plug the card in
> another pci slot ,the PC will still breakdown when the TMDBG's gui
> appear or start to download program to board.
> But yesterday this unplug and replug did not work when my PC
> breakdown  again.I even reinstall windows Xp ,ndk5.5 and jtag card
> driver,the tmdbg still can't work.
>
> Who can tell me why ?Is this card not a stable one  ?
> My old card never make my PC breakdown.
> This new card's version is C1.
>
>
>
> Sent via the TriMedia mailing list
> trimedia@yahoogroups.com - archive at http://groups.yahoo.com/group/trimedia
> Yahoo! Groups Links
>
>
>
>

Hello
Does anyone know of any attempt at porting OpenCV
(http://opencvlibrary.sourceforge.net/)  to TriMedias? It might be an
interesting and perhaps useful exercise.

--
Dr Sergio A. Velastin
Reader, Digital Imaging Research Centre
Faculty of Computing, Information Systems & Mathematics
Kingston University
Penrhyn Road
Kingston-upon-Thames
Surrey KT1 2EE
United Kingdom
Tel:  +44 (20) 8547 7719
Fax:  +44 (20) 8547 7972
Cell: +44 (701) 701 9380

sergio.velastin@...
sergio.velastin@...
sergio.velastin@...
http://dircweb.king.ac.uk

Thanks Afred.I try the older version driver and the latest download
from http://www.mds.com/support/extra_support.asp?prod=Nexperia%
5Cpci_jtag_support.html.
The card still can't work.

Sometimes the TMDBG may work successfully but the PC will breakdown
soon after this,and it will never work again.

--- In trimedia@yahoogroups.com, "Alfred Lee" <alfred@...> wrote:
>
> I wonder if the wrong driver is being used for the newer card.
>
> Please contact directly MDS for assistance.
>
>
> Best Regards,
> Alfred Lee
>
>
> ----- Original Message -----
> From: "gongwjh" <GONGXF@...>
> To: <trimedia@yahoogroups.com>
> Sent: Thursday, June 21, 2007 9:06 AM
> Subject: [trimedia] Help about my MDS pci Jtag card
>
>
> > My mds pci card often make my PC break down!
> >
> > I use NDK5.5 and mds's latest version driver which download from
> > mds's site.Now everytime I run tmdbg,the PC will break down just
> > when the TMDBG's  GUI appear or when I start to download program
to
> > my board.
> >
> > Several months ago my MDS jtag card didn't work and was sent to
> > repair .MDS seemed to give me a new card instead of reparing
because
> > it's quite new.
> > This card seemed not to be the same version as my older one,it
make
> > my PC breakdown once and then.Everytime I must plug the card in
> > another pci slot and reboot the PC .If I did not plug the card
in
> > another pci slot ,the PC will still breakdown when the TMDBG's
gui
> > appear or start to download program to board.
> > But yesterday this unplug and replug did not work when my PC
> > breakdown  again.I even reinstall windows Xp ,ndk5.5 and jtag
card
> > driver,the tmdbg still can't work.
> >
> > Who can tell me why ?Is this card not a stable one  ?
> > My old card never make my PC breakdown.
> > This new card's version is C1.
> >
> >
> >
> > Sent via the TriMedia mailing list
> > trimedia@yahoogroups.com - archive at
http://groups.yahoo.com/group/trimedia
> > Yahoo! Groups Links
> >
> >
> >
> >
>

Hi Sergio,

We have do a fast attempt to port it, but it was before TCS5.01, and we have
meet some C++ compilation problems (in cxcore from memory). Perhaps we will
resume this task in next months...



====================================
Frédéric BERNON
HYMATOM SA
Chef de projet informatique
Microsoft Certified Professional
Tél. : +33 (0)4-67-87-61-10
Fax. : +33 (0)4-67-70-85-44
Email : frederic.bernon@...
Web Site : http://www.hymatom.fr <http://www.hymatom.fr/>
====================================
P Avant d'imprimer, penser à l'environnement


	 -----Message d'origine-----
	 De : trimedia@yahoogroups.com [mailto:trimedia@yahoogroups.com] De la part de
Sergio A Velastin
	 Envoyé : jeudi 21 juin 2007 22:12
	 À : trimedia@yahoogroups.com
	 Objet : [trimedia] OpenCV



	 Hello
	 Does anyone know of any attempt at porting OpenCV
	 (http://opencvlibrary.sourceforge.net/ <http://opencvlibrary.sourceforge.net/>
) to TriMedias? It might be an
	 interesting and perhaps useful exercise.

	 --
	 Dr Sergio A. Velastin
	 Reader, Digital Imaging Research Centre
	 Faculty of Computing, Information Systems & Mathematics
	 Kingston University
	 Penrhyn Road
	 Kingston-upon-Thames
	 Surrey KT1 2EE
	 United Kingdom
	 Tel: +44 (20) 8547 7719
	 Fax: +44 (20) 8547 7972
	 Cell: +44 (701) 701 9380

	 sergio.velastin@... <mailto:sergio.velastin%40kingston.ac.uk>
	 sergio.velastin@... <mailto:sergio.velastin%40iee.org>
	 sergio.velastin@... <mailto:sergio.velastin%40ieee.org>
	 http://dircweb.king.ac.uk <http://dircweb.king.ac.uk>







[Non-text portions of this message have been removed]

hi,

   I am trying to designed decoder component giving YUV outputs, when i
give this to renderer it sends error like ERR_BAD_PACKET is this because
of headers requared by renderer. and what kind of header information
needed by renderer,
and how can i set it during packet formation.
   Please reply....

___________Admin________________
This email has been ClamScanned !
           www.clamav.net

I suggest you look into the FbRead component and copy what it does.  The
FbRead component contains a frame buffer and when the frame buffer is
updated it sends the packets to the renderer.  Do look at the latest
NDK, as there are recent updates.

     Chuck

Chuck Peplinski
TriMedia Tech Support for NXP Semiconductors  www.TCSHelp.com
TriMedian at MDS 			 www.mds.com



amol tukaram chaudhari wrote:
>
> hi,
>
> I am trying to designed decoder component giving YUV outputs, when i
> give this to renderer it sends error like ERR_BAD_PACKET is this because
> of headers requared by renderer. and what kind of header information
> needed by renderer,
> and how can i set it during packet formation.
> Please reply....
>
> ___________Admin________________
> This email has been ClamScanned !
> www.clamav.net
>
>


[Non-text portions of this message have been removed]

Just in case anyone here is not registered on TCSHelp, I want to share
the announcement of NDK 5.6, and encourage you to be registered.

Dear TriMedia User,

The MPTK team at NXP Semiconductors is proud to announce a new update to
the MPTK and NDK packages.  NDK 5.6 and MPTK 2.3 deliver bug fixes and
additional features.  A complete release note is available to registered
users with a brief description of improvements, new features and known
issues at http://www.tcshelp.com/pnx1500_tools_downloads.html.  The
packages are now available for download through the NXP Electronic
Software Distribution system at TCSHelp.

All users of record should now have received a notice that their new
downloads are available.  If you have not received notification, please
submit a new support issue via TCSHelp - do not reply to this message.

NXP strongly encourages all programmers of Nexperia Media Processors to
update to the latest release.  It becomes increasingly difficult to
provide effective support for old releases.  The newest releases contain
bug fixes and improvements that are valuable to all projects.

*HIGHLIGHTS*

     * Building under Linux has been tested and is now supported.  For
       users comfortable with Linux, this can reduce build times.
     * The use of the Dinkumware C++ libraries provided by TCS 5.x has
       been enabled in the SDE2 build environment.
     * The memCheck tool previously posted on TCSHelp is now included in
       this NDK release.
     * In addition to Blunk's TargetTCP, Unicoi's Fusion network stack
       has been tested in the context of MPTK.
     * The USB stack has been updated to fix many problems noted in
       extensive tests.
       -  The isp156x and isp1761 stacks have been unified behind a
       single interface, tmUsbIsp.  One or the other is chosen by diversity.
       -  HID (Human Interface Devices) class for keyboards and mice is
       now added to USB support.
     * The FAT32 library has been updated.
       -  Performance improvements/optimization of FAT32 file-system.
       -  Forty-eight bit LBA support is provided for IDE to support the
       larger disk size i.e. disk sizes greater than 137GB.
     * The MPTK components that support MPEG4 decoding have been enhanced
       to support multiple instances.



The TriMedia Support Team




[Non-text portions of this message have been removed]

hi,
     i ma succesfully able to solve Rendering Problem using FbRead
component reference. i am getting Video Out but problem in croma
component. i am seding three Buffers YUV in single packet to renderer
with required videoFormat like

static tmVideoFormat_t videoformat =
{
     sizeof(tmVideoFormat_t),  /* size                               */
     0,                        /* hash                               */
     0,                        /* referenceCount                     */
     avdcVideo,                /* dataClass                          */
     vtfYUV,                   /* dataType                           */
     vdfYUV420Planar,          /* dataSubtype                        */
     vdfProgressive,           /* description                        */
     512,                      /* imageWidth                         */
     384,                      /* imageHeight                        */
     512*2,                    /* imageStride                        */
     0,                        /* activeVideoStartX                  */
     0,                        /* activeVideoStartY                  */
     512,                      /* activeVideoEndX                    */
     384,                      /* activeVideoEndY                    */
     vasNone,                  /* tmVideoAnalogStandard_t            */
     0,                        /* imageUVStride                      */
};

but poblem in color out which are no comming properly , i think some chroma
value problem,

Any suggestions please inform me.

> I suggest you look into the FbRead component and copy what it does.  The
> FbRead component contains a frame buffer and when the frame buffer is
> updated it sends the packets to the renderer.  Do look at the latest
> NDK, as there are recent updates.
>
>     Chuck
>
> Chuck Peplinski
> TriMedia Tech Support for NXP Semiconductors  www.TCSHelp.com
> TriMedian at MDS 			 www.mds.com
>
>
>
> amol tukaram chaudhari wrote:
>>
>> hi,
>>
>> I am trying to designed decoder component giving YUV outputs, when i
>> give this to renderer it sends error like ERR_BAD_PACKET is this because
>> of headers requared by renderer. and what kind of header information
>> needed by renderer,
>> and how can i set it during packet formation.
>> Please reply....
>>
>> ___________Admin________________
>> This email has been ClamScanned !
>> www.clamav.net
>>
>>
>
>
> [Non-text portions of this message have been removed]
>
>

___________Admin________________
This email has been ClamScanned !
           www.clamav.net

Are you setting "needBuffer" = True for the renderer?  Running in assert
or trace mode?  The QVCP needs semiplanar data.  The MBS normally
converts planar to that format in VrendGfxVo.  To do this, needBuffer
needs to be true.  Maybe if you ran in retail mode this error would not
be flagged?  I'm guessing.  I never use retail mode, and I can't see
what you see on the screen.

     Chuck

Chuck Peplinski
TriMedia Tech Support for NXP Semiconductors  www.TCSHelp.com
TriMedian at MDS 			 www.mds.com



amol tukaram chaudhari wrote:
>
>
> hi,
> i ma succesfully able to solve Rendering Problem using FbRead
> component reference. i am getting Video Out but problem in croma
> component. i am seding three Buffers YUV in single packet to renderer
> with required videoFormat like
>
> static tmVideoFormat_t videoformat =
> {
> sizeof(tmVideoFormat_t), /* size */
> 0, /* hash */
> 0, /* referenceCount */
> avdcVideo, /* dataClass */
> vtfYUV, /* dataType */
> vdfYUV420Planar, /* dataSubtype */
> vdfProgressive, /* description */
> 512, /* imageWidth */
> 384, /* imageHeight */
> 512*2, /* imageStride */
> 0, /* activeVideoStartX */
> 0, /* activeVideoStartY */
> 512, /* activeVideoEndX */
> 384, /* activeVideoEndY */
> vasNone, /* tmVideoAnalogStandard_t */
> 0, /* imageUVStride */
> };
>
> but poblem in color out which are no comming properly , i think some
> chroma
> value problem,
>
> Any suggestions please inform me.
>
> > I suggest you look into the FbRead component and copy what it does. The
> > FbRead component contains a frame buffer and when the frame buffer is
> > updated it sends the packets to the renderer. Do look at the latest
> > NDK, as there are recent updates.
> >
> > Chuck
> >
> > Chuck Peplinski
> > TriMedia Tech Support for NXP Semiconductors www.TCSHelp.com
> > TriMedian at MDS www.mds.com
> >
> >
> >
> > amol tukaram chaudhari wrote:
> >>
> >> hi,
> >>
> >> I am trying to designed decoder component giving YUV outputs, when i
> >> give this to renderer it sends error like ERR_BAD_PACKET is this
> because
> >> of headers requared by renderer. and what kind of header information
> >> needed by renderer,
> >> and how can i set it during packet formation.
> >> Please reply....
> >>
> >> ___________Admin________________
> >> This email has been ClamScanned !
> >> www.clamav.net
> >>
> >>
> >
> >
> > [Non-text portions of this message have been removed]
> >
> >
>
> ___________Admin________________
> This email has been ClamScanned !
> www.clamav.net
>
>


[Non-text portions of this message have been removed]

Hi, good afternoon everybody! There are two problems that I
faced with recently.

      The first problem, why is program dead easily when the *.out
file is loaded in dvpMon5.3 with debug edition? That is to say, the
running program is dead easily when the *.out file is loaded in
dvpMon5.3 with ¡°set _TMTGTREL = debug¡± in dvpMon_Win1700.bat file,
but it isn¡¯t dead for ever with ¡°set _TMTGTREL = retail (or
assert)¡± in dvpMon_Win1700.bat file. Why?? How to solve it? The
following error appears once in a while in the debug print
information.

     ¡°1-00:29:27.820;thread 0x360000;ASSERTION FAILED IN LINE 755,
       FILE tmFormat.c: FAILED (t £½tmFormat_Check (format)) == 0U:
       MESSAGE

      Function: tmFormat_Claim

      Version: %version: PE9#13 %

      Error 0x6003005CU (CID_CLASS_INFRASTR | CID_TYPE_NONE |
      CID_COMP_TSSA_DEFAULTS | CID_LAYER_NONE | TM_ERR_BAD_FORMAT) ¡±

      The second problem, the program is running after the *.out file
is loaded in dvpMon5.3 with ¡°set _TMTGTREL = retail (or assert)¡± in
dvpMon_Win1700.bat file. The above program runs well and isn¡¯t dead
for ever when ¡°dvpMon5.3¡± debug tool is open all the while, but it
is dead fast when ¡°dvpMon5.3¡± debug tool is close. Is the above
phenomenon right? However, It seems that some programs loaded in
dvpMon5.3 still run well and isn¡¯t dead after ¡°dvpMon5.3¡± debug
tool is close. Why??

      In my above project programs, tmPFconvert component is evolved
from tmCopyIO component, the exolVidPip application and tmPFconvert
component program will realize the following function. Firstly,
720x576 vision data¡ª>VIP¡ª>PFCONVERT_MAIN_INPUT of tmPFconvert
component¡ª>1024x768 buffer¡ª>MBS¡ª>Layer0 of Qvcp in VrendGfxVo.
Secondly, audio data¡ª>PFCONVERT_SECOND_INPUT of tmPFconvert
component¡ª>some algorithm processes (Here may be omitted
temporarily)¡ª>the above 1024x768 buffer¡ª>MBS
¡ª> Layer0 of Qvcp in VrendGfxVo. In result, the scaled vision data
was displayed on the center of CRT, and the scaled variable audio
pole data was displayed on two sides of CRT.

     Best regards! Wait for your reply sincerely!!

dvpMon does not work with debug build.  In fact, it is pointless since
dvpMon is not a debugger.  To fully taking making use of debug build,
you should use the debugger, tmdbg.exe.  It also support JTAG operation.

Why do you need to keep the target program running after closing dvpMon?

Best Regards,
Alfred Lee


----- Original Message -----
From: "gshzhong2005" <gshzhong2005@...>
To: <trimedia@yahoogroups.com>
Sent: Tuesday, July 03, 2007 1:47 AM
Subject: [trimedia] *Why is it dead easily when *.out file loaded in dvpMon5.3
with debug edition?


      Hi, good afternoon everybody! There are two problems that I
faced with recently.

      The first problem, why is program dead easily when the *.out
file is loaded in dvpMon5.3 with debug edition? That is to say, the
running program is dead easily when the *.out file is loaded in
dvpMon5.3 with ¡°set _TMTGTREL = debug¡± in dvpMon_Win1700.bat file,
but it isn¡¯t dead for ever with ¡°set _TMTGTREL = retail (or
assert)¡± in dvpMon_Win1700.bat file. Why?? How to solve it? The
following error appears once in a while in the debug print
information.

     ¡°1-00:29:27.820;thread 0x360000;ASSERTION FAILED IN LINE 755,
       FILE tmFormat.c: FAILED (t £½tmFormat_Check (format)) == 0U:
       MESSAGE

      Function: tmFormat_Claim

      Version: %version: PE9#13 %

      Error 0x6003005CU (CID_CLASS_INFRASTR | CID_TYPE_NONE |
      CID_COMP_TSSA_DEFAULTS | CID_LAYER_NONE | TM_ERR_BAD_FORMAT) ¡±

      The second problem, the program is running after the *.out file
is loaded in dvpMon5.3 with ¡°set _TMTGTREL = retail (or assert)¡± in
dvpMon_Win1700.bat file. The above program runs well and isn¡¯t dead
for ever when ¡°dvpMon5.3¡± debug tool is open all the while, but it
is dead fast when ¡°dvpMon5.3¡± debug tool is close. Is the above
phenomenon right? However, It seems that some programs loaded in
dvpMon5.3 still run well and isn¡¯t dead after ¡°dvpMon5.3¡± debug
tool is close. Why??

      In my above project programs, tmPFconvert component is evolved
from tmCopyIO component, the exolVidPip application and tmPFconvert
component program will realize the following function. Firstly,
720x576 vision data¡ª>VIP¡ª>PFCONVERT_MAIN_INPUT of tmPFconvert
component¡ª>1024x768 buffer¡ª>MBS¡ª>Layer0 of Qvcp in VrendGfxVo.
Secondly, audio data¡ª>PFCONVERT_SECOND_INPUT of tmPFconvert
component¡ª>some algorithm processes (Here may be omitted
temporarily)¡ª>the above 1024x768 buffer¡ª>MBS
¡ª> Layer0 of Qvcp in VrendGfxVo. In result, the scaled vision data
was displayed on the center of CRT, and the scaled variable audio
pole data was displayed on two sides of CRT.

     Best regards! Wait for your reply sincerely!!




Sent via the TriMedia mailing list
trimedia@yahoogroups.com - archive at http://groups.yahoo.com/group/trimedia
Yahoo! Groups Links

Hello All,
On a MDS810 I am getting the following message, which happened after
pointing to a wrong rootfs.

MicroBTM, Compiled Aug 22 2006 10:50:29
Using Uart 1
Initialising NAND driver...OK
Loading Bootloader...Nand Address:00004000, Page load Failed! (0x10F01803)

Any ideas?
  Cheers,

I can give a simple and generic answer, but it may not answer your
questions.  We rely heavily on assertions in the code base for
debugging.  If you get an assertion, it tells you that the code has
encountered a condition that the programmers deemed dangerous.  You
should never use the retail version of the code for debugging as all
assertions are removed in the retail version.  So a condition that
results in an assertion will stop the program in assert, trace or debug
mode.  But in retail mode, the program may pass this point only to crash
mysteriously at some later time.

So if you get an assertion, you need to look at the cause and debug it.

I think I understand you to be asking another question about what
happens when you close dvpMon.  When a program is built with the
_jtagmon diversity, it expects the dvpMon host to be present.  If you
close dvpMon, the behavior of the target is not defined.  If you want a
program to run without dvpMon, you simply must build it this way.  I see
that you are also talking to an NXP application engineer.  He can
probably help you come to a question we can understand better.

     Chuck

Chuck Peplinski
TriMedia Tech Support for NXP Semiconductors  www.TCSHelp.com
TriMedian at MDS 			 www.mds.com



gshzhong2005 wrote:
>
> Hi, good afternoon everybody! There are two problems that I
> faced with recently.
>
> The first problem, why is program dead easily when the *.out
> file is loaded in dvpMon5.3 with debug edition? That is to say, the
> running program is dead easily when the *.out file is loaded in
> dvpMon5.3 with ¡°set _TMTGTREL = debug¡± in dvpMon_Win1700.bat file,
> but it isn¡¯t dead for ever with ¡°set _TMTGTREL = retail (or
> assert)¡± in dvpMon_Win1700.bat file. Why?? How to solve it? The
> following error appears once in a while in the debug print
> information.
>
> ¡°1-00:29:27.820;thread 0x360000;ASSERTION FAILED IN LINE 755,
> FILE tmFormat.c: FAILED (t £½tmFormat_Check (format)) == 0U:
> MESSAGE
>
> Function: tmFormat_Claim
>
> Version: %version: PE9#13 %
>
> Error 0x6003005CU (CID_CLASS_INFRASTR | CID_TYPE_NONE |
> CID_COMP_TSSA_DEFAULTS | CID_LAYER_NONE | TM_ERR_BAD_FORMAT) ¡±
>
> The second problem, the program is running after the *.out file
> is loaded in dvpMon5.3 with ¡°set _TMTGTREL = retail (or assert)¡± in
> dvpMon_Win1700.bat file. The above program runs well and isn¡¯t dead
> for ever when ¡°dvpMon5.3¡± debug tool is open all the while, but it
> is dead fast when ¡°dvpMon5.3¡± debug tool is close. Is the above
> phenomenon right? However, It seems that some programs loaded in
> dvpMon5.3 still run well and isn¡¯t dead after ¡°dvpMon5.3¡± debug
> tool is close. Why??
>
> In my above project programs, tmPFconvert component is evolved
> from tmCopyIO component, the exolVidPip application and tmPFconvert
> component program will realize the following function. Firstly,
> 720x576 vision data¡ª>VIP¡ª>PFCONVERT_MAIN_INPUT of tmPFconvert
> component¡ª>1024x768 buffer¡ª>MBS¡ª>Layer0 of Qvcp in VrendGfxVo.
> Secondly, audio data¡ª>PFCONVERT_SECOND_INPUT of tmPFconvert
> component¡ª>some algorithm processes (Here may be omitted
> temporarily)¡ª>the above 1024x768 buffer¡ª>MBS
> ¡ª> Layer0 of Qvcp in VrendGfxVo. In result, the scaled vision data
> was displayed on the center of CRT, and the scaled variable audio
> pole data was displayed on two sides of CRT.
>
> Best regards! Wait for your reply sincerely!!
>
>


[Non-text portions of this message have been removed]

Dear Dr Sergio,
   I am Arun Pradeep from Bangalore, India. I am
interested in porting to TriMedia. Do let me know some
pointers and starting point .

Looking forward your support,

Best Regards
Arun
--- Sergio A Velastin <sergio.velastin@...> wrote:

> Hello
> Does anyone know of any attempt at porting OpenCV
> (http://opencvlibrary.sourceforge.net/)  to
> TriMedias? It might be an
> interesting and perhaps useful exercise.
>
> --
> Dr Sergio A. Velastin
> Reader, Digital Imaging Research Centre
> Faculty of Computing, Information Systems &
> Mathematics
> Kingston University
> Penrhyn Road
> Kingston-upon-Thames
> Surrey KT1 2EE
> United Kingdom
> Tel:  +44 (20) 8547 7719
> Fax:  +44 (20) 8547 7972
> Cell: +44 (701) 701 9380
>
> sergio.velastin@...
> sergio.velastin@...
> sergio.velastin@...
> http://dircweb.king.ac.uk
>
>




________________________________________________________________________________\
____
Food fight? Enjoy some healthy debate
in the Yahoo! Answers Food & Drink Q&A.
http://answers.yahoo.com/dir/?link=list&sid=396545367

Has there been any update on the Pthreads port or Pthreads mapping to
trimedia's pSOS. The last post I found was dated nov 2003.

The OpenCV librery will compile fine only a few warnings, but fails
at the linking stage as it requires the cxcore libs, they inturn
require pthreads support, and it doesn't seem able to support
pthreads directly.

--- In trimedia@yahoogroups.com, Arun Pradeep <mapradeep@...> wrote:
>
> Dear Dr Sergio,
>   I am Arun Pradeep from Bangalore, India. I am
> interested in porting to TriMedia. Do let me know some
> pointers and starting point .
>
> Looking forward your support,
>
> Best Regards
> Arun
> --- Sergio A Velastin <sergio.velastin@...> wrote:
>
> > Hello
> > Does anyone know of any attempt at porting OpenCV
> > (http://opencvlibrary.sourceforge.net/)  to
> > TriMedias? It might be an
> > interesting and perhaps useful exercise.
> >
> > --
> > Dr Sergio A. Velastin
> > Reader, Digital Imaging Research Centre
> > Faculty of Computing, Information Systems &
> > Mathematics
> > Kingston University
> > Penrhyn Road
> > Kingston-upon-Thames
> > Surrey KT1 2EE
> > United Kingdom
> > Tel:  +44 (20) 8547 7719
> > Fax:  +44 (20) 8547 7972
> > Cell: +44 (701) 701 9380
> >
> > sergio.velastin@...
> > sergio.velastin@...
> > sergio.velastin@...
> > http://dircweb.king.ac.uk
> >
> >
>
>
>
>
>
______________________________________________________________________
______________
> Food fight? Enjoy some healthy debate
> in the Yahoo! Answers Food & Drink Q&A.
> http://answers.yahoo.com/dir/?link=list&sid=396545367
>

If you assume pthreads starts another task, you can create a
wraper to call tmosalTaskCreate ().

Best Regards,
Alfred Lee


----- Original Message -----
From: "d_szlamp" <d_szlamp@...>
To: <trimedia@yahoogroups.com>
Sent: Thursday, July 05, 2007 3:29 AM
Subject: [trimedia] Re: OpenCV


> The OpenCV librery will compile fine only a few warnings, but fails
> at the linking stage as it requires the cxcore libs, they inturn
> require pthreads support, and it doesn't seem able to support
> pthreads directly.
>
> --- In trimedia@yahoogroups.com, Arun Pradeep <mapradeep@...> wrote:
>>
>> Dear Dr Sergio,
>>   I am Arun Pradeep from Bangalore, India. I am
>> interested in porting to TriMedia. Do let me know some
>> pointers and starting point .
>>
>> Looking forward your support,
>>
>> Best Regards
>> Arun
>> --- Sergio A Velastin <sergio.velastin@...> wrote:
>>
>> > Hello
>> > Does anyone know of any attempt at porting OpenCV
>> > (http://opencvlibrary.sourceforge.net/)  to
>> > TriMedias? It might be an
>> > interesting and perhaps useful exercise.
>> >
>> > --
>> > Dr Sergio A. Velastin
>> > Reader, Digital Imaging Research Centre
>> > Faculty of Computing, Information Systems &
>> > Mathematics
>> > Kingston University
>> > Penrhyn Road
>> > Kingston-upon-Thames
>> > Surrey KT1 2EE
>> > United Kingdom
>> > Tel:  +44 (20) 8547 7719
>> > Fax:  +44 (20) 8547 7972
>> > Cell: +44 (701) 701 9380
>> >
>> > sergio.velastin@...
>> > sergio.velastin@...
>> > sergio.velastin@...
>> > http://dircweb.king.ac.uk
>> >
>> >
>>
>>
>>
>>
>>
> ______________________________________________________________________
> ______________
>> Food fight? Enjoy some healthy debate
>> in the Yahoo! Answers Food & Drink Q&A.
>> http://answers.yahoo.com/dir/?link=list&sid=396545367
>>
>
>
>
>
> Sent via the TriMedia mailing list
> trimedia@yahoogroups.com - archive at http://groups.yahoo.com/group/trimedia
> Yahoo! Groups Links
>
>
>
>

Hello All,
I am trying to create a video provider, which is a mixture of c++ and
c files. However, the Probe function fails on dlopen, complaining that
_Znwj symbol is not defined. c++filt resolves _Znwj to new(unsigned
int), indicating that c++ library is not linked in the final .so
image. Indeed, the following is observed:
[IDirectFBVideoProvider]$ mips_4KCle-nm -C -l
libphStbDFBVideoProvider_t.so | grep -w -i U | grep new
          U operator new[](unsigned int)
          U operator new(unsigned int)
I have added the following to my local makefile to no avail:
LOCAL_LDFLAGS += -lgcc -lstdc++ -lsupc++ -lstdc++.so.6.0.4
Any suggestions?
  Cheers,

Dear Sir,
    I am working  on TM3260 and want to optimise my code.
Will you provide any sample code for data cache optimization using
custom operation like pref, alloc, dinvalid and  dcb. Beacause i try to
use thease instruction but didn't get good result.

Regards,
Mital Pawar

Dear Mital,

     According to my experience, the data cache optimization instructions are not
useful really. However, data re-arrangement is. If you search the archieve, you
should be able to find relevant mails.

Basically, the idea is to re-arrange the data such that you access data which is
in spatial and temporal proximity. For example, if you are numbers in even
location, store the even numbers separately. This would roughly reduce data
cache stalls by a factor of 2. Similarly, while working on video, store the 8x8
or 16x16 blocks in spatial proximity. You may interleave chrominance UV samples
(one, two, or 4 samples of U with corresponding samples of V).

Hope this gives you and idea.

Regards,
-Ahmad

   ----- Original Message -----
   From: pmits_11
   To: trimedia@yahoogroups.com
   Sent: Friday, July 06, 2007 6:31 PM
   Subject: [trimedia] TM3260 Data cache optimization {SN_Disarmed}


   Dear Sir,
   I am working on TM3260 and want to optimise my code.
   Will you provide any sample code for data cache optimization using
   custom operation like pref, alloc, dinvalid and dcb. Beacause i try to
   use thease instruction but didn't get good result.

   Regards,
   Mital Pawar





[Non-text portions of this message have been removed]

Is it possible to adjust the video brightness using the MBS?





Thanks



Simon Gauthier

BSP/Core Group Team Leader

Video Intelligence Solutions

Verint System Inc.

450.686.9000 x 232

450.686.0198

simon.gauthier@...

www.verint.com





________________________________________________________________________________\
__________
This electronic message may contain proprietary and confidential information of
Verint Systems Inc., its affiliates and/or subsidiaries.
The information is intended to be for the use of the individual(s) or
entity(ies) named above.  If you are not the intended recipient (or authorized
to receive this e-mail for the intended recipient), you may not use, copy,
disclose or distribute to anyone this message or any information contained in
this message.  If you have received this electronic message in error, please
notify us by replying to this e-mail. (1)

[Non-text portions of this message have been removed]

Brightness just adds to luminance (Y). So if you are in YUV then I don't
think the MBS helps.

If you are converting from RGB then you can adjust the color space
conversion matrix to apply brightness.

And you can apply contrast and saturation using the MBS. You do this also
through color space conversion.

For example multiply the coefficients of the first row by a contrast factor
to apply contrast to the Y component of YUV when converting from RGB to YUV.

You can convert from one color space to the same color space - this is how
the BCS unit (VCBM) in QVCP works in fact.

Chris
========================
Chris Bore
BORES Signal Processing
chris@...
www.bores.com
Tel:    +44 1483 740138
Mobile: +44 7921 153 219

-----Original Message-----
From: trimedia@yahoogroups.com [mailto:trimedia@yahoogroups.com] On Behalf
Of Gauthier, Simon
Sent: 09 July 2007 19:06
To: trimedia@yahoogroups.com
Subject: [trimedia] Adjusting video brightness via MBS.


Is it possible to adjust the video brightness using the MBS?





Thanks



Simon Gauthier

BSP/Core Group Team Leader

Video Intelligence Solutions

Verint System Inc.

450.686.9000 x 232

450.686.0198

simon.gauthier@...

www.verint.com





____________________________________________________________________________
______________
This electronic message may contain proprietary and confidential information
of Verint Systems Inc., its affiliates and/or subsidiaries.
The information is intended to be for the use of the individual(s) or
entity(ies) named above.  If you are not the intended recipient (or
authorized to receive this e-mail for the intended recipient), you may not
use, copy, disclose or distribute to anyone this message or any information
contained in this message.  If you have received this electronic message in
error, please notify us by replying to this e-mail. (1)

[Non-text portions of this message have been removed]



Sent via the TriMedia mailing list
trimedia@yahoogroups.com - archive at http://groups.yahoo.com/group/trimedia
Yahoo! Groups Links

Please refer to the training slides that come with NDK 5.6 (the latest
version).  Same with NDK 5.5.  Data cache optimization is the hardest
challenge in TriMedia optimizations.  As Ahmad says, data rearrangement
is critical.  You must understand the working set of your algorithm and
you must understand how this working set maps into the cache.  As one
speaker at the recent TMUG said, "programmers' unfortunate obsession
with modularization and data hiding" works against data cache optimzation.

Chuck Peplinski
TriMedia Tech Support for NXP Semiconductors  www.TCSHelp.com
TriMedian at MDS 			 www.mds.com




pmits_11 wrote:
>
> Dear Sir,
> I am working on TM3260 and want to optimise my code.
> Will you provide any sample code for data cache optimization using
> custom operation like pref, alloc, dinvalid and dcb. Beacause i try to
> use thease instruction but didn't get good result.
>
> Regards,
> Mital Pawar
>
>


[Non-text portions of this message have been removed]

Hi,

Do we have any performance numbers for H.264/WMV encode on TM3260?

TIA

Ramakrishnan

Hi, good morning everybody! There is one problems that I faced with
recently.

Why does image appear jitter with exolVidPip in NDK5.5 when
¡°needBuffer = False¡± is set up? The following is the set-up of the
parameters of exolVidPip in NDK5.5.

In the function of ¡°tmErrorCode_t exolVidPipSetupVcapVip()¡±,
pVcapVipSetup->bufferLayout = tmVideoFieldBuffer;

In the function of ¡°void RunVip(exolVidPipInstance_t* ivp)¡±,
     pInputFormat->description = vdfFieldInField;

In the function of ¡°void SelectInputFormat(exolVidPipInstance_t*
ivp)¡±,
ivp->InputFormat = defFormatPAL;
ivp->OutputFormat=defFormatPAL;
ivp->InputVideoAdapter=vaaCVBS;
ivp->OutputVideoAdapter=tmVencAna_Cvbs;
ivp->pipWindow = False;

In the function of ¡°tmErrorCode_t  exolVidPipSetupVrendGfxVo( ),
pVrendGfxVoSetup->maxLayerSetup[0].needBuffer = False;
pVrendGfxVoSetup->layerSetup[0].operatingMode =
                 tmolVrendGfxVo_OperatingModeFieldBase;

That is to say, why does the Layer0 vision image not scaling appear
jitter with exolVidPip in NDK5.5 when the parameters are set up
according to the above? Here the input and output format of vision
image must be ¡°vdfFieldInField¡± and ¡°needBuffer = False¡±. How to
solve it?

     Best regards! Wait for your reply sincerely!!

Is it posable to link components linked with 64bit doubles to
components linked with 32bit doubles? It seems there is no 64 bit
librery for pSOS, drv_conf, sysinit & bsp. these are all required to
buld under NDK.

TCSHelp knowledge base item 82 answers this question.  I went to
TCSHelp, chose knowlege base, search, and then I entered fp64.  There I
found the answer I wrote in the past.
To answer your specific question briefly, all of the things you listed
are recompiled for every application.  If they are compiled and linked
with -fp64, then the right stuff will happen.  But knowledge base issue
82 has more details.

     Chuck

Chuck Peplinski
TriMedia Tech Support for NXP Semiconductors  www.TCSHelp.com
TriMedian at MDS 			 www.mds.com




d_szlamp wrote:
>
> Is it posable to link components linked with 64bit doubles to
> components linked with 32bit doubles? It seems there is no 64 bit
> librery for pSOS, drv_conf, sysinit & bsp. these are all required to
> buld under NDK.
>
>


[Non-text portions of this message have been removed]