Where to Obtain Latest Tool Releases?

Hi Darwin,

Our DVSDK releases (what you get thru our software update site) is tested to ensure software component compatibility.  This is why I tend to suggests customers work with DVSDK (as opposed to updating individual software components) as it avoids many head-aches.

I believe you are using the latest DVSDK version but you are substituting the Linux kernel that comes in the DVSDK with the one that you can get from MV Zone; is this correct?  If so, than you need to make sure the the kernel that you are putting together (your own customized kernel with the patches you choose) from MV Zone have all the necessary support that the rest of the DVSDK software (e.g. encodedecode demos) require.  In other words, when you start updating individual components, it is the responsibility of the software integrator to ensure they maintain software compatibility (not for the faint of heart). 

When it comes to the Linux kernel, compatibility is normally defined a the linux kernel version level (e.g. requires kernel 2.6.10); at the patch level, it is mostly a trial and error situation and integrator must know the patches well they are working with.  Often, to play it safe, this means getting more patches than you really need.  When it comes to other software components, compatibility is defined in the release notes.  For example, a codec data-sheet will tell your it needs a certain version of codec engine; codec engine release notes will in turn tell you it needs certain version of xdais, framework components, dsplink... and so on. 

In conclusion
1) We do test for software compatibility at the DVSDK level
2) We do provide software compatibility information via release notes documents.
3) Fine grain software compatibility information (e.g. Linux kernel patch level) is the responsibility of the software integrator; this is dependant on what services the applications need underneath and each product is different, hence when customizing your software at this level, it requires good understanding of the overall system and software.

Juan,

You sell the MV4 Pro release, and now you are basically saying that you do not support it. (Actually, that was obvious from the complete lack of documentation related to the pro release -- i.e., all of the documentation assumes the demo release.) To support the pro release in the case that we are dealing with, all that you would need to do is provide the kernel patch level, which you clearly state that you are not willing to do, even though someone within your organization clearly knows what the patch level is (since she or he had to bring the kernel up to that patch level).

This is kind of poor customer service, to say the least. (If you think about it, it would kind of make sense to give your customers what they need to be successful, especially when it should be completely trivial to do so, as in this case...)

Have a great day!

Take care,

Darwin

Darwin,

I believe there is miscommunication here; I have posted the path level you asked for ( pro-1641 ) a couple of times on this thead.

I am patiently trying to explain things to you and would appreciate it if you would ask again if you do not understand the answers rather than criticize.  I am here to help you and would rather engage in providing solutions and explaining our support model rather than responding to unproductive commentary.

That said, we do resell MV License on MontaVista's behalf and do support it, but please understand the intricacies of working with Linux patches as described in my previous post is general to the industry and not specific to TI.  It is very difficult for any Linux distro company (not the TI is one) to tell you explicitly what patches you should apply for your project, but we can tell you what patch addresses a certain fix you are looking for (what I have done).  The patches are made available to the customer and it is general knowledge that you should apply them in sequence to avoid head-aches.  I am not sure what your Linux background is, but your expectations are somewhat unreasonable.  We provide a free distro, we provide a path for customers to get a newer distro via MV License.  MV takes our patches, merges them and tests their distro and makes patches available to you.  The MV Zone Kernel is not flawed and with intermediate Linux knowledge, you should be able to take patches and use them to meet your needs.

If there is something that is not clear, please ask; if you want us to improve certain aspects, your 'constructive' feedback is always welcome.  But please, no more unproductive commentary. 

Juan,

I am trying to be constructive. If your team would just put the requisite patch level in the documentation (e.g., on the Codec Engine Roadmap page), it would save a lot of trouble for people, like me. I am not sure how stating that it being destructive, but if it is, then I apologize. BTW, I must have missed the patch level in your previous message, and I do really appreciate your assistance...

Anyway, I went ahead and installed the current LSP for the current DVSDK over the one that came with our MV4 Pro installation, and this did allow us to both build the kernel and the DVSDK and demos. However, now we are getting errors that we have never seen before:

Encodedecode demo started.
DSP-side configuration mismatch/failure
0              -> success
Positive value -> DSP-side failure code.
(Uint32) -1    -> DSP-side component was not initialized.

    DRV configuration status    [0xffffffff]
    IPS configuration status    [0xffffffff]
    POOL configuration status   [0xffffffff]
    MPCS configuration status   [0xffffffff]
    MPLIST configuration status [0xffffffff]
    MQT configuration status    [0xffffffff]
    RINGIO configuration status [0xffffffff]
TraceUtil> Error: Failed to open codec engine "encodedecode"
TraceUtil> Error: Aborting TraceUtil_start
DSP-side configuration mismatch/failure
0              -> success
Positive value -> DSP-side failure code.
(Uint32) -1    -> DSP-side component was not initialized.

    DRV configuration status    [0xffffffff]
    IPS configuration status    [0xffffffff]
    POOL configuration status   [0xffffffff]
    MPCS configuration status   [0xffffffff]
    MPLIST configuration status [0xffffffff]
    MQT configuration status    [0xffffffff]
    RINGIO configuration status [0xffffffff]
Encodedecode Error: Failed to open codec engine encodedecode
DSP-side configuration mismatch/failure
0              -> success
Positive value -> DSP-side failure code.
(Uint32) -1    -> DSP-side component was not initialized.

    DRV configuration status    [0xffffffff]
    IPS configuration status    [0xffffffff]
    POOL configuration status   [0xffffffff]
    MPCS configuration status   [0xffffffff]
    MPLIST configuration status [0xffffffff]
    MQT configuration status    [0xffffffff]
    RINGIO configuration status [0xffffffff]
Capturing 720x480 video (cropped to 720x480)
Encodedecode Error: Failed to open codec engine encodedecode

Do you have any idea what the cause of this could be? (It looks like the codec engine is not starting or improperly configured, but why would that be; I am just building off of an unmodified DVSDK?)

Have a great day!

Take care,

Darwin

Juan,

Please disregard the issue in my last post, since I resolved it on my own.

BTW, I apologize profusely if any of my comments were negative. While I do believe that you (T.I.) could and definitely should do a much, much better job with your documentation, there is no excuse for being rude. Life is difficult enough as it is.

Have a great week!

Take care,

Darwin

Darwin,

It is great news that you got things working.

I do agree we could certainly improve our software and documentation.  BTW, our team keeps a running list of the areas that are customers have the most difficult time with and are constantly feeding this information back to our product teams to help us improve our offering; the DVSDK is much improved from what was provided a couple of years ago but we still have ways to go.  This is where your constructive feedback is important to us and believe me, I take customer feedback very serious, so please keep it coming.  Capturing information such as number of customers reporting an issue, business opportunity, frequency of issue... helps us prioritize our work accordingly.  FYI, from having worked with you in the past thru our Seibel system, I have the information I need for you already. 

Let us continue to work together; asl always, I we are here to help you.

Darwin,

In an attempt to clear up some the confusion regarding TI Vs MV Patches and how they are associated with prticular LSP and/or DVSDK versions, we have created the following wiki article.  Hopefully you and other will find this helpful.

http://tiexpressdsp.com/wiki/index.php?title=PSP_Introduction_and_Product_FAQ

I would make one overall comment as I'm dealing with a lot of this upgrade to MVL5 headache right now...

TI needs to centralize documentation and patches. Many of the links right now are via myTI. Some are off the Wiki, others are off of these forums, others still come from MV. The spiderweb of packages and downloads, documentation etc. is just daunting to anyone trying to traverse it. Google is almost required to find anything.

It must be clear what uPs are supported by which packages.

One question: Do the DM355 mpeg4 SP and jpeg codecs come with CE 2.21?

To be clear, we do not fully support MVL5 at this moment.  If you're looking for a "one stop shop" then I do not recommend using MVL5 (yet).

Most of our components have been adding support for MVL5 in anticipation of full support.  That's why right now you have to go grab all the individual components.  In late January (maybe early Feb) we will release DVSDK 2.0.  At that time then MVL5 will be considered "fully supported" in terms of multimedia support.  You will be able to download it from the DVEVM updates site.  Everything you need will be available there.  That is, you will not have to go to a bunch of different web sites to get the stuff (CE, LSP, etc).

CE 2.21 does not come with any codecs.  Codec Engine is the "middle ware", the codecs are separate.

Brad

I'm having the same problem. I downloaded all of the dm6446 files from the softwareupdates site except for mvl_ files. I used the mvl pro CDs that we purchased. I was told by the local FAE that he was able to run make in the dvsdk_1_30_01_41 successfully. I haven't been successful yet with this dvsdk, but I was able to successfully build dvevm1_10 that came with the CDs. I downloaded the mvl_ demp files and tried again, but get the same error

Mike

MRI said:

I'm having the same problem. I downloaded all of the dm6446 files from the softwareupdates site except for mvl_ files. I used the mvl pro CDs that we purchased. I was told by the local FAE that he was able to run make in the dvsdk_1_30_01_41 successfully. I haven't been successful yet with this dvsdk, but I was able to successfully build dvevm1_10 that came with the CDs. I downloaded the mvl_ demp files and tried again, but get the same error

What issue are you having?

I would think that you would need at least the mvl_4_0_1_demo_lsp_setuplinux_01_20_00_014.bin.  This has an update Linux LSP for the DM6446.  Plus, I would suggest updating to the other packages on the DM6446 Extranet which includes the appropriate DSP/BIOS version, XDC tools version and C64x code generation tools.

Okay, I obtained the CE 2.21 and all of the tools that are supposed to be compatible with it per the roadmap (http://wiki.davincidsp.com/index.php?title=Codec_Engine_Roadmap). However, building DSP Link 1.60 is a real bear.

I configured it via the following command:

perl dsplinkcfg.pl --platform=DAVINCI --nodsp=1 --dspcfg_0=DM6446GEMSHMEM --dspos_0=DSPBIOS5XX --gppos=MVL5G --comps=ponslrmc

Then I set the DSPLINK environment variable to point to the main DSP Link 1.60 directory. Then I went to gpp/src directory and tried to build it via make -s.

This produced an error that /home/darwin/montavista5_glibc/pro/devkit/arm/v5t_le/bin/arm_v5t_le-gcc was not found. So, I created a link from montavista5_glibc to our montavista5 install directory in my home directory.

Then I tried to rebuild DSP Link via make -s.

This produced the following error:

Compiling dm6446gem_phy_shmem.c...
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/timex.h:187,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/sched.h:48,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:9,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:34:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/timex.h:15:28: error: asm/arch/timex.h: No such file or directory

So, I created a link from the arch-omap directory to arch in our /opt/montavista5/pro/devkit/arm/v5t_le/target/usr/include/asm-arm directory.

The I tried to rebuild DSP Link via the make -s command. This produced the following errors, which is where I am stuck:

Compiling dm6446gem_phy_shmem.c...
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/sched.h:49,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:9,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:34:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:33:3: error: #error You lose.
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/jiffies.h:210:31: error: division by zero in #if
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/semaphore.h:15,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/sched.h:57,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:9,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:34:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/rwsem.h:39:66: error: asm/rwsem.h: No such file or directory
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/page.h:179,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/sched.h:58,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:9,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:34:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h: In function ‘virt_to_phys’:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h:178: error: ‘CONFIG_DRAM_BASE’ undeclared (first use in this function)
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h:178: error: (Each undeclared identifier is reported only once
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h:178: error: for each function it appears in.)
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h: In function ‘phys_to_virt’:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h:183: error: ‘CONFIG_DRAM_BASE’ undeclared (first use in this function)
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h: In function ‘virt_to_bus’:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h:201: error: ‘CONFIG_DRAM_BASE’ undeclared (first use in this function)
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h: In function ‘bus_to_virt’:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/memory.h:206: error: ‘CONFIG_DRAM_BASE’ undeclared (first use in this function)
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/memory_hotplug.h:7,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/mmzone.h:339,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/gfp.h:4,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/slab.h:14,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/percpu.h:4,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/rcupdate.h:41,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/pid.h:4,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/sched.h:70,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:9,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:34:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/notifier.h: At top level:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/notifier.h:47: error: field ‘rwsem’ has incomplete type
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:9,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:34:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/sched.h:460: error: field ‘mmap_sem’ has incomplete type
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/sysdev.h:24,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/sched.h:1793,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:9,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:34:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/kobject.h:171: error: field ‘rwsem’ has incomplete type
In file included from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:34:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:611:2: error: #error Modules should never use the kernel headers found in /usr/include
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:612:2: error: #error but rather the headers found in your LSP kernel sources.
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/device.h:19,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/pci.h:53,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:37:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:611:2: error: #error Modules should never use the kernel headers found in /usr/include
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/module.h:612:2: error: #error but rather the headers found in your LSP kernel sources.
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/fs.h:314,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/mm.h:15,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h:6,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/dma-mapping.h:34,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm-generic/pci-dma-compat.h:7,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/pci.h:5,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/pci.h:759,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:37:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/quota.h:290: error: field ‘dqptr_sem’ has incomplete type
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/mm.h:15,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h:6,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/dma-mapping.h:34,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm-generic/pci-dma-compat.h:7,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/pci.h:5,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/pci.h:759,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:37:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/fs.h:520: error: field ‘i_alloc_sem’ has incomplete type
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/mm.h:15,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h:6,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/dma-mapping.h:34,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm-generic/pci-dma-compat.h:7,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/pci.h:5,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/pci.h:759,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:37:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/fs.h:875: error: field ‘s_umount’ has incomplete type
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h:6,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/dma-mapping.h:34,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm-generic/pci-dma-compat.h:7,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/pci.h:5,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/pci.h:759,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:37:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/mm.h: In function ‘lowmem_page_address’:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/mm.h:531: error: ‘CONFIG_DRAM_BASE’ undeclared (first use in this function)
In file included from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/dma-mapping.h:34,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm-generic/pci-dma-compat.h:7,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/pci.h:5,
                 from /opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/linux/pci.h:759,
                 from DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:37:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h: In function ‘dma_map_single’:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h:151: error: ‘CONFIG_DRAM_BASE’ undeclared (first use in this function)
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h: In function ‘dma_map_sg’:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h:258: error: ‘CONFIG_DRAM_BASE’ undeclared (first use in this function)
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h: In function ‘dma_sync_single_for_cpu’:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h:318: error: ‘CONFIG_DRAM_BASE’ undeclared (first use in this function)
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h: In function ‘dma_sync_single_for_device’:
/opt/montavista5/pro/devkit/arm/v5t_le/bin/../target/usr/include/asm/dma-mapping.h:326: error: ‘CONFIG_DRAM_BASE’ undeclared (first use in this function)
DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c: In function ‘DM6446GEM_phyShmemInit’:
DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:139: warning: implicit declaration of function ‘IO_ADDRESS’
DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c:139: error: ‘IO_PHYS’ undeclared (first use in this function)
make[2]: *** [DM6446GEM/shmem/Linux/dm6446gem_phy_shmem.c.deb] Error 1
make[1]: *** [objdeb] Error 2
make: *** [arch.objdeb] Error 2

What is the problem, here? Is there an easier way for me to go about this? (For example, is there a prebuilt version of DSP Link 1.60 that works with MVL5?)

Have a great day!

Take care,

Darwin

Are you using GNU make 3.81 as specified in the DSPLINK 1.61 release notes: https://www-a.ti.com/downloads/sds_support/targetcontent/link/link_1_60/link_1_61_00/LinuxLinkReleaseNotes.pdf ?

Juan,

Yes, we have been using GNU make V3.81 for quite some time, now. I was pretty clear in my post about the issue that we are facing. Can you please respond to the issue in my previous post. (I think that my post was quite explicit, but if you do not agree, then please ask for clarification, so that we can get this resolved within a reasonable period of time.)

Have a great day!

Take care,

Darwin

Darwin,

I aplogize if you had already mentioned which GNU version you were using; I could not find it anywhere in this thread.  The GNU make version is one of the more common problems we encounter; therefore it was worth mentioning it for the benefit of all the community.  Especially since I could not readily find references to GNU make v3.81 in this thread.

With regards to your other questions, I am afraid I do not have an answer for them at this time; otherwise I would gladly have posted them. 

Hi Darwin,

This thread is getting a bit long, so forgive me if I missed this.  May I ask what your end goal is in moving to MV Pro 5.0 and DSPLINK 1.61?  Do you just want to run codec examples (e.g. video_copy??) in this new environment ?  If so, you do not need to rebuild DSPLINK at all, simply download MV Pro 5 LSP, Codec Egine 2.21, and BIOS 5.33.  I believe you have all of these already. 

Also, I am told DVSDK 2.0 should be releasing in early Feb, and it should have all these components integrated in a full software release.

Thanks, Juan.

Juan> May I ask what your end goal is in moving to MV Pro 5.0 and DSPLINK 1.61

Because the prebuilt version of DSP Link that comes with CE 2.21 does load into MV Pro 5. (It is compiled for MV 4. Thus, when we try to load the module into the kernel, the kernel balks.) So, we need to build DSP Link 1.61 from scratch, unless you have a prebuilt version. (The issues that we are faced with are discussed in my previous, long post, in this thread.)

Can you please look at the problem mentioned in that previous long post (in this thread) and let me know how to circumvent that?

Have a great night!

Take care,

Darwin

Quick correction, the pre-built version of DSP Link that comes with CE 2.21 was built against an MVPro 5 kernel:

> strings codec_engine_2_21/examples/apps/system_files/DM6446/dsplinkk.ko | grep vermagic
vermagic=2.6.18_pro500-davinci_evm-arm_v5t_le preempt mod_unload ARMv5 gcc-4.2

It's true that if your MV5 distro has been reconfigured or patched, you should rebuild DSP Link's .ko modules (as well as CMEM and all other modules), so the pre-built modules may not load into your kernel.  But it _is_ built against an MV5 Pro distro.

Chris

Darwin,

You shouldn't have to manually make up the links to arch-omap in your kernel distribution. When you are building DSPLink (since it is a kernel-module), you need to make sure that you kernel distribution has been setup for the target device. i.e. you need to ensure that all the soft links are already in place in your kernel distribution.

You can do a build of your Linux kernel distribution for the target device before building DSPLink to get rid of this error.

Regards,
Mugdha

Chris,

The signature that it gives when we try to load the module is an MV4 signature. So, I wonder what is going on...

Have a great day!

Take care,

Darwin

ChrisRing said:

Quick correction, the pre-built version of DSP Link that comes with CE 2.21 was built against an MVPro 5 kernel:

> strings codec_engine_2_21/examples/apps/system_files/DM6446/dsplinkk.ko | grep vermagic
vermagic=2.6.18_pro500-davinci_evm-arm_v5t_le preempt mod_unload ARMv5 gcc-4.2

It's true that if your MV5 distro has been reconfigured or patched, you should rebuild DSP Link's .ko modules (as well as CMEM and all other modules), so the pre-built modules may not load into your kernel.  But it _is_ built against an MV5 Pro distro.

Chris

Thanks, Mugdha.

Yes, we have no problems building the kernel, and that is not even the point at which we are stuck. Can you please read the entire post and then help us to resolve the issue that we are stuck at?

Have a great day!

Take care,

Darwin

MugdhaK said:

Darwin,

You shouldn't have to manually make up the links to arch-omap in your kernel distribution. When you are building DSPLink (since it is a kernel-module), you need to make sure that you kernel distribution has been setup for the target device. i.e. you need to ensure that all the soft links are already in place in your kernel distribution.

You can do a build of your Linux kernel distribution for the target device before building DSPLink to get rid of this error.

Regards,
Mugdha

HyperGalaxy said:

Yes, we have no problems building the kernel, and that is not even the point at which we are stuck. Can you please read the entire post and then help us to resolve the issue that we are stuck at?

Darwin,

If you're manually creating links then you have not built your kernel correctly.  You need to back up a step and figure out why that link is not already created.  Did you correctly configure your kernel?  Perhaps there's a permissions error causing the config step to not be able to create the links?

Brad, Juan, etc.,

Okay, I finally finish the client task that I swapped in while waiting for your assistance with this issue and am back on this.

I took a step back and tried to rebuild DVSDK 1.30.01.41 with CE 2.21, etc. This time, I noticed the following error when trying to make install:

install: cannot stat `/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/BUILD/EXPORT/RELEASE/dsplinkk.ko': No such file or directory

Do you have any idea what is causing this? (I am just trying to use the pre-built version of DSP Link that came with CE 2.21.)

Here are the different dsplinkk.ko kernel modules that I found in CE 2.21:

/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/pack
ages/dsplink/gpp/export/BIN/Linux/DAVINCI/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/pack
ages/dsplink/gpp/export/BIN/Linux/DAVINCI/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/Linux/DAVINCIHD/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/Linux/DAVINCIHD/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/Linux/DM357/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/Linux/DM357/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/Linux/OMAP2530/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/Linux/OMAP2530/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/Linux/OMAP3530/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/Linux/OMAP3530/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/DAVINCI/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/DAVINCI/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/DAVINCIHD/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/DAVINCIHD/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/DM357/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/DM357/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/OMAP2530/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/OMAP2530/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/OMAP3530/DEBUG/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/dsplink/gpp/export/BIN/ucLinux/OMAP3530/RELEASE/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/ti/bios/power/modules/dm357/dsplink/dsplinkk.ko

/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/ti/bios/power/modules/dm6446/dsplink/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/ti/bios/power/modules/omap2530/dsplink/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/cetools/packages/ti/bios/power/modules/omap3530/dsplink/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/examples/apps/system_files/DM6446/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/examples/apps/system_files/DM6467/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/examples/apps/system_files/OMAP2530/dsplinkk.ko
/home/darwin/projects/LightSpeed/dvsdk_1_30_01_41/codec_engine_2_21/examples/apps/system_files/OMAP3530/dsplinkk.ko

Thus, it appears that the pre-built version of DSP Link that is include with CE 2.21 does not have the version of the kernel module that the build system expects.

Have a great day!

Take care,

Darwin

The DVSDK 1.30 Makefile has this:

    @install -m 666 $(LINK_INSTALL_DIR)/packages/dsplink/gpp/BUILD/EXPORT/RELEASE/dsplinkk.ko $(EXEC_DIR)

Better change that to:

    @install -m 666 $(CE_INSTALL_DIR)/cetools/packages/dsplink/gpp/export/BIN/Linux/DAVINCI/RELEASE/dsplinkk.ko $(EXEC_DIR)

Thanks, Brad.

That solved the installation issue. However, now none of the example applications are working.

For example, here is what happens when I execute ./loadmodules.sh:

ioremap_nocache(0x87800000, 8388608)=0xc9000000
allocated heap buffer 0xc9000000 of size 0xf7000
cmem initialized 4 pools between 0x87800000 and 0x88000000
dsplinkk: no version for "struct_module" found: kernel tainted.
DSPLINK Module (1.60) created on Date: Oct 24 2008 Time: 12:59:07

That looks fine, but here is what happens when I exectue ./encode -v out.mpeg4:

Encode demo started.
TraceUtil> Error: Failed to open codec engine "encode"
TraceUtil> Error: Aborting TraceUtil_start
i2c_davinci i2c_davinci.1: controller timed out
Encode Error: Failed to open codec engine encode
Capturing 720x480 video (cropped to 720x480)
Encode Error: Failed to open codec engine encode

Here are the files that the DVSDK installs:

-rw-rw-rw- 1 500 500   23438 Feb  2  2009 cmemk.ko
drwxr-xr-x 4 500 500    4096 Feb  2  2009 data
-rwxr-xr-x 1 500 500  319978 Feb  2  2009 decode
-r--r--r-- 1 500 500    4898 Feb  2  2009 decode.txt
-rw-rw-rw- 1 500 500 3645541 Feb  2  2009 decodeCombo.x64P
-rw-rw-rw- 1 500 500  123356 Feb  2  2009 dsplinkk.ko
-rwxrwxr-x 1 500 500     384 Mar 20  2008 dvevmdemo
-rwxr-xr-x 1 500 500  317524 Feb  2  2009 encode
-r--r--r-- 1 500 500    5283 Feb  2  2009 encode.txt
-rw-rw-rw- 1 500 500 1863279 Feb  2  2009 encodeCombo.x64P
-rwxr-xr-x 1 500 500  312143 Feb  2  2009 encodedecode
-r--r--r-- 1 500 500    4504 Feb  2  2009 encodedecode.txt
-rwxr-xr-x 1 500 500   60693 Feb  2  2009 interface
-rwxrwxrwx 1 500 500     372 Feb  2  2009 loadmodules.sh
-rw-rw-rw- 1 500 500 2550902 Feb  2  2009 loopbackCombo.x64P
drwxr-xr-x 2 500 500    4096 Feb  2  2009 thirdpartydemos
drwxrwxr-x 3 500 500    4096 Nov  5  2008 web

Can you please tell me how to resolve this?

Have a great day!

Take care,

Darwin

Brad Griffis said:

The DVSDK 1.30 Makefile has this:

    @install -m 666 $(LINK_INSTALL_DIR)/packages/dsplink/gpp/BUILD/EXPORT/RELEASE/dsplinkk.ko $(EXEC_DIR)

Better change that to:

    @install -m 666 $(CE_INSTALL_DIR)/cetools/packages/dsplink/gpp/export/BIN/Linux/DAVINCI/RELEASE/dsplinkk.ko $(EXEC_DIR)

 

HyperGalaxy said:

That looks fine, but here is what happens when I exectue ./encode -v out.mpeg4:

Encode demo started.
TraceUtil> Error: Failed to open codec engine "encode"
TraceUtil> Error: Aborting TraceUtil_start
i2c_davinci i2c_davinci.1: controller timed out
Encode Error: Failed to open codec engine encode
Capturing 720x480 video (cropped to 720x480)
Encode Error: Failed to open codec engine encode

Not sure what this is, but often you can gain more insight into what's failing by setting CE_DEBUG=1 (or 2?) in the environment (http://wiki.davincidsp.com/index.php?title=CE_DEBUG) before running the ./encode app.

Chris

Chris,

Here is the output with CE_DEBUG set to 1 (minimum):

Encode demo started.
@0,989,420us: [+6 T:0x4001ee50] CE - Engine_init> CE debugging on (CE_DEBUG=1; allowed CE_DEBUG levels: 1=min, 2=good, 3
=max)
TraceUtil_start> note: CE_DEBUG env. var is set, so TraceUtil is not active (unset CE_DEBUG if you need TraceUtil)
i2c_davinci i2c_davinci.1: controller timed out
@21,517,306us: [+7 T:0x40aeb490] OP - Processor_create_d> Loading and starting DSP server './encodeCombo.x64P' FAILED, s
tatus=[0x80008017] (look for error code 'DSP_EBASE + 0x17' in dsplink*/packages/dsplink/gpp/inc/usr/errbase.h)
@21,517,672us: [+6 T:0x40aeb490] OP - Processor_delete_d> Closing remote transport FAILED, status=0x80008000.
@21,517,885us: [+6 T:0x40aeb490] OP - Processor_delete_d> Stopping DSP FAILED, status=0x80008000
VIDIOC_QUERYSTD failed on /dev/video0. Video input connected?
@21,535,388us: [+6 T:0x42ce6490] CE - rserverOpen: can't start './encodeCombo.x64P'; Processor_create failed
Encode Error: Failed to open codec engine encode
@42,007,771us: [+7 T:0x40aeb490] OP - Processor_create_d> Loading and starting DSP server './encodeCombo.x64P' FAILED, s
tatus=[0x80008017] (look for error code 'DSP_EBASE + 0x17' in dsplink*/packages/dsplink/gpp/inc/usr/errbase.h)
@42,008,101us: [+6 T:0x40aeb490] OP - Processor_delete_d> Closing remote transport FAILED, status=0x80008000.
@42,008,314us: [+6 T:0x40aeb490] OP - Processor_delete_d> Stopping DSP FAILED, status=0x80008000
@42,018,726us: [+6 T:0x4001ee50] CE - rserverOpen: can't start './encodeCombo.x64P'; Processor_create failed
Encode Error: Failed to open codec engine encode

I still have no clue about how to circumvent this, but I would guess that it is DSP Link related from the output.

Do you have any ideas about what is causing this and how to resolve it?

Have a great day!

Take care,

Darwin