dsp56725 jtag scan error

sec_FBT · ‎04-15-2011

Hi all,

I'm fighting with a DSP56725 and the associated Symphony Studio configuration for on-chip debugging via openOCD.

I've tried every configuration found in Symphony, but with no success; I receive an error message like:

Error: jtag.c:1156 jtag_validate_chain(): Error validating JTAG scan chain, IR mismatch, scan returned 0x03ff

I connected the oscilloscope and I see that DSP react to the requests of Jtag, so this is not an Hardware problem.

Perhaps it could be a software problem linked to ID identifier for DSP56725; so, have anyone of you tried working with DSP56725?

Have you got any problem with the openOCD with Symphony Studio?

clangen · ‎04-18-2011

Hi,

please describe the exact hardware configuration you use for debgging the DSP56725. I assume you have the DSPAUDIOEVMMB motherboard and the DSPB725DB daughterboard but this is just a gess. Do you use the parallel port connection to the host PC or an additional JTAG adapter. Please see if the motherboard is jumpered properly. You can send me a list of all jumper settings or take a high resolution photo as well.

Attached you can find a small tutorial how to debug the DSP56725.

Best regards

Christian

sec_FBT · ‎04-18-2011

Hi,

we use a self-made board that cointain, among other components, a DSP56725 and a Jtag port.

To debug i use a wiggler, connected to the board through an adapter OnCe - Jtag with a schematic identical to that conitained in the motherboard of the EVM, that i attach.

The schematic of this board in the same that we have used with the older DSP56364, with which we have no problem in debug. So, because the DSP56725 is composed of two 56300 cores, i've believed that i could use the same debug circuit.

Are there some differences in the debug section that i've missed to consider?

I've analyzed the behavior of the DSP in debugging when connected with Symphony Studio IDE with an oscilloscope: i've noticed that the signals on TCK, TDI and TDO pins seem to be good. In fact, i've detected a response of the DSP similar to that of the 56364, that is correctly recognized from the IDE.

why does the software not recognize this response, giving me the jtag scan error?

could it be an incompatibility with the older wiggler device?

rocco · ‎04-19-2011

Hi Sec,

First, a disclaimer: I am using the DSP56721, not the 25, but they should be the same, according to the data sheet.

I have attached the JTAG circuit that I am using. I don't see any reason why your's doesn't work. It is nearly identical to mine, except that I have an additional pullup resistor on pin-9, "DSP_ONLY_RESET". Notice that I have nothing attached to pins 13 and 14.

This circuit has been working for me, on both the Freescale parallel -port wiggler (Macraigor Systems) and Xverve's USB Signalyzer. However, we could not get Symphony-Studio to work at all, so we never determined how reliably the Symphony-Studio / Signalyzer combination would work with both cores. When Freescale dropped support for Symphony-Studio, we switched back to the old, but not trustworthy, Suite-56 tools, and used the parallel-port wiggler. We were able to debug both cores with the Suite-56 / Wiggler combination, but not without difficulties.

With Suite-56, I had to run a script to get the debugger to access both cores, but I don't recall if Symphony-Studio had a similar requirement.

mark

sec_FBT · ‎04-19-2011

Dear Mark,

the problems rise when i try to use Symphony Studio, because it doesn't recognize my DSP, even if, when i analyzes it with an oscilloscope i can clearly see that it recognizes the debug request and response correctly.

You are telling me that Freescale has dropped support for Symphony Studio... so for debugging we are forced to use the old 56kSuite??? It gives me problems even for debugging the older 56364, with this new dual-core DSP it will be a nightmare to debug

I'd like to say what kind of debugger other developers use with the 567xx family, because i can't believe that everybody use the 56ksuite...

Do you think that a Signalyzer could solve my problems? Wiggler is an old device, and perhaps the parallel port could create some incompatibility problem...

I'm getting mad with this DSP

rocco · ‎04-19-2011

sec@FBT wrote:

Do you think that a Signalyzer could solve my problems? Wiggler is an old device, and perhaps the parallel port could create some incompatibility problem...

It could be a parallel-port problem. When Tech-support told us to switch to Suite56, we responded that it had no USB support, and all of our workstations had USB and no parallel ports. They suggested that we try a USB->parallel-port adapter for the old wiggler (which we then had to buy). We tried every USB->Parallel-port adapter we could find, and NONE worked with the wiggler. We ended up retrieving a ten-year-old workstation from storage that had a parallel port on the motherboard.

We never tried the wiggler on Symphony-Studio. It might be worth checking with Macraigor to see if they support the 725 on Symphony-Studio.

I felt that the Signalyzer was a good device, but I could not do enough debugging with it to determine whether it has flaws.

sec_FBT · ‎04-19-2011

I'm near to believe that Symphony Studio is not updated to work wit the new 56725, because it seems not to recognize the product ID requested at the start of debugging with this IDE.

To solve the parallel port compatibility problem, I will try to install the Symphony Studio in an older PC and try to use it with wiggler.

rocco wrote:

When Tech-support told us to switch to Suite56...

___________________________________________________________________________________________

It's incredible that support has suggested to use a very old platform with the new dual core processors.... In this way it seems that Symphony Studio IDE is practically unusable...
Another information: what kind of script have you used to work with the dual-core DSP on 56ksuite?

rocco · ‎04-19-2011

It's incredible that support has suggested to use a very old platform with the new dual core processors.... In this way it seems that Symphony Studio IDE is practically unusable...

Maybe not unusable, but it was unsupportable.

Their explanation was that the proprietary DSP portions of Symphony-Studio were written by a third party, and tech-support did not have access to that code, and therefore could not support it.

Another information: what kind of script have you used to work with the dual-core DSP on 56ksuite?

The purpose of the script was to tell the debugger where on the JTAG chain the two cores were. Here it is attached, straight .from the service request response.

sec_FBT · ‎04-19-2011

At this point, I don't know if trying with this DSP it's better than changing with a new supported one.

The project is at the first step, then perhaps changing DSP it's not a bad idea.

I will try to connect it with with the old 56K-suite, but i don't believe it could be a good solution to develop code with all the debugging problems this HW configuration will create.

Thx mark for these (bad for me...) informations...

clangen · ‎04-19-2011

Hi all,

it's all not that bad as it looks like:

The parallel port interface (identical for the Wiggler and the DSPAUDIOEVMMB) is supported both by Suite 56 and Symphony Studio. See the attached tutorial how to use it. We've used Suite56 with the parallel port interface with the DSP56725 sucessfully as well as Symphony Studio with parallel port and the Xverve Signalizer that is obsolete now unfortunately and can no longer be purchased that way. An alternative can be the Macraigor USB Wiggler thar doesn't work with the DSP56371 but all other derivatives see: http://macraigor.com/usbWiggler.htm

I will make an alternaive for the Signalizer available soon that works with Symphony Studio for all DSP56xxx.

Tomorrow we will evaluate different test programs for the DSP56725/Symphony Studio combination. Basically debugging is quite easy since dual core DSP are supported. After invoking the debug interface two different debug sessions can be invoked independantly by chosing core 0 and 1 respectively. Both sessions can be started and stopped individually.I've checked this today without experiencing any problems. The problem seems to be lack of documentation and experience of the user community. The behaviour is identically to other development toolchains using Eclipse and GDB (gnu debugger) in combination with OpenOCD.

A detailled report with screenshots will follow soon (hopefully this or next week).

Best regards

Christian

clangen · ‎04-20-2011

Hi all,

today I've checked different demonstration programs for the DSP5672x using Symphony Studio and a parallel port connection. Since Symphony Studio projects always generates one target file only the individual codes run on each code have to be kept in different projects as well.

Once a running external tool configuration is established multiple debug sessions can be started with individual codes uploaded to the cores 0 and 1 individually. The debug sessions on both cores can be started, paused, stopped and debugged individually as well.

I can resume that Symphony Studio is better suited for dual core programming and debugging than Suite56.

Unfortunaltely I can no longer find the documentation on the Symphony Studio website http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=SYMPH_STUDIO&fpsp=1&tab=Documentation... so I attach it to this post.

A colleague of mine will summarize the steps how to debug the 56725 as I demonstrated to him today.

Best regards

Christian

clangen · ‎04-19-2011

Hi all,

the intended tutorial was too large for the last post. I've stripped most screenshots out of it to reduce the size. The remaining sceleton is attached here.

Best regards

Christian