[C674x DSPLIB/MATHLIB] How to track bug fixes among the releases ?

Kawada,

I went ahead and moved this thread over to the device forum in hopes that a DSPLib expert can help answer this more quickly.

Hi David,

As you know, the release note of DSPLIB encourages us to ask BIOS Forum (TI-RTOS) regarding the technical questions. Hope you could answer the question here anyway. 

Best Regards,
Kawada

Hi Kawada,

Thanks for your post.

Actually, there is a Global SDOCM database which is an TI's internal bug tracking tool to evaluate for the IR's filed for each bug and its status, but unfortunately, this cannot be shared to external customers. 

Please see the wiki below for the current status and workaround of each known issue of updated C674x DSP libaray release. You can see few issues are active still & some are fixed. Kindly check the below one:

http://processors.wiki.ti.com/index.php/C674x_DSPLIB_Known_Issues

Thanks & regards,

Sivaraj K

-------------------------------------------------------------------------------------------------------

Please click the Verify Answer button on this post if it answers your question

-------------------------------------------------------------------------------------------------------

Hi Sivaraj,

Thanks for your reply. 

As for the internal bug tracking system, I understood. Let me confirm more things.

The wiki you suggested introduces the status of some known issues for v12 release.
v12 is for dsplib_1_03_00_01, right ? As I mentioned in the previous post, DSPF_sp_fir_r2 is not interruptive and it should be regarded as a known issue. But actually, this fact is NOT described in this wiki. Further more, this issue has not been described in the latest (version 3.x) release notes yet. I would like you to consider about the maintenance for these docs. 

The latest version of dsplib is v3_2_0_1 at this time frame. Are these issues (marked as "active" in this wiki) are still applicable to the latest release ?

Also, as I mentioned, c67x dsplib (NOT c674x) document is available at http://www.ti.com/lit/ug/spru657c/spru657c.pdf
and this introduces some known issues at the section "B.3 Known Issues". Are these issues also applicable to dsplib_1_03_00_01 and the recent versions of dsplib (3.x) ?

Best Regards,
Kawada

Hi,

Yes you are right. The wiki i shared in my previous post is only applicable to v12 for dsplib_1_03_00_01. I also apologise that, it is not applicable to the latest DSP lib v3_2_0_1 and as per the known issues at the section "B.3" mentioned in  c67x dsplib doc. 

But anyway, i would take up your consideration in maintaining the DSP LIB document relases and will try to forward your request to right TI team experts to fulfill the same

Thanks for your inputs and suggestion.

Thanks & regards,

Sivaraj K

-------------------------------------------------------------------------------------------------------

Please click the Verify Answer button on this post if it answers your question

-------------------------------------------------------------------------------------------------------

Hi Sivarj

Thanks for your understanding. Hope the documents would be revised in the near future.

This is the final question -  as I mentioned previous post, c67x dsplib (NOT c674x) document is available at http://www.ti.com/lit/ug/spru657c/spru657c.pdf

and this introduces some known issues at the section "B.3 Known Issues". 
Here is just a reference for you.

==============

There are some issues with how interrupts are handled in a few of the functions.
So when calling the following functions you should disable interrupts
before the call and restore interrupts after:
DSPF_sp_minval
DSPF_sp_maxval
DSPF_sp_ifftSPxSP
DSPF_sp_dotprod
DSPF_sp_fir_gen
DSPF_sp_fir_r2
Failure to disable interrupts before calling these specific functions can
cause corruption of the stack or sometimes a register which ultimately can
lead to the application crashing.
The DSPF_dp_vecrecip function gives incorrect results for every two
alternate values of the input array.
The DSPF_dp_biquad function results differ from the Matlab results.
The DSPF_dp_mat_mul function in v2 of DSBLIB does not work for odd
rows and columns.
The DSPF_sp_blk_move function corrupts the next two returning lines of
code.

==============

I asked you if these issues are also applicable to dsplib_1_03_00_01 and the recent versions of dsplib (3.x).
And I got the answers for the latest version of dsplib (i.e., the all of known issues described in the wiki you introduced and the above known issues described in c67x dsplib have been resolved in the latest release!, If my understanding is not correct, please let me know), but how about for dsplib_1_03_00_01 ?

Actually, one of my customer is using dsplib_1_03_00_01 and found that there was a trouble in DSPF_sp_fir_r2 (it should not be interruptive otherwise system can crash).  He has solved this issue by gating the interrupts during the call. So, he is wondering dsplib_1_03_00_01 could have same issues described the above. What do you think of this ? If using the latest version of DSPLIB (v3_2_0_1) is a recommendation, I'll ask the same to him, but again I have to ask you the same thing ... you should maintain the release notes for the more detailed bug/know issue tracking if we (TI/Distributors) ask the users to use the latest version of libraries. The same thing is for MATHLIB.

Best Regards,
Kawada

Hi Sivaraj

Thanks for the answers.
My customer has already tried the latest version (v3_2_0_1) of dsplib but the interruptibility issue of DSPF_sp_fir_r2 was still there. I think it may be better to re-confirm whether the all of the issues reported in the previous posts are really fixed or not.

Anyway, I suggested him to use the latest version of dsplib. We really hope the release notes would be kept to maintain correctly for the bug fixes.

Thank you.
Kawada 

Hi,

Thanks for your update.

Yes, you are right, the interruptibility issue is still there and all issues are not fixed only few are fixed and many are in progress which are active yet to fix (developers are working on).

Thanks for your note.

Regards,

Sivaraj K

-------------------------------------------------------------------------------------------------------

Please click the Verify Answer button on this post if it answers your question

-------------------------------------------------------------------------------------------------------

Hi Sivaraj,

>Yes, you are right, the interruptibility issue is still there and all issues are not fixed only few are fixed and many are in >progress which are active yet to fix (developers are working on).

Ok...  This means users may not be able to fix the interruptibility issues even though they update the current lib to the latest one. If you have already identified the functions having the interruptibility issue for each (dsplib v1_03_00_01 and the latest one), could you please list them ? We can apply the workaround to them.

Sorry to ask you so much.

Best Regards,
Kawada