Known DSP bug?

[DrTypo]

I'm going back to my 3D engine and I'm moving part of it to the DSP.

I encountered a strange behaviour:

  movei #_nTriAdd,r0
  store rNAdd,(r0)

This randomly craps out. The correct value is not always stored at _nTriAdd.

I found a fix:

  movei #_nTriAdd,r0
  or  r0,r0
  store rNAdd,(r0)

Does it ring a bell to anyone?

Thanks.

 

[SCPCD]

I don't have read or uncounter this sort of bug (from what I remind, but I haven't play much with the DSP)

 

Does _nTriAdd in external memory ?

Is there a load somewhere before ?

Does rNAdd comes from a long instruction (as div?)

 

I would suspect the #1 or  #7 in the "Hardware Bugs" reference manual.

Edited March 27, 2021 by SCPCD

[DrTypo]

Yes, _nTriAdd is in external memory.

 

About the load instruction, this was my first suspicion but any load instruction is complete before the store.

 

rNAdd comes from a short instruction (addq #1,rNAdd).

 

Thank you for your feedback! It indeed looks a bit like one of the hardware bugs in the manual but it doesn't match exactly one.