Can some one explain how manage single bit error fix on SDM of eTPU?
I don't know very well this next sentence. So I'm searching for sample code example.
SDM fixes can be done using the CDC (see Coherent Dual-parameter Controller (CDC)),
limited to the lower 8 Kbytes of SDM. The Host application simply has to program a
transfer with both origin and destination positions pointing to the address of the data to be
fixed. Because the PBBASE field has an address granularity of 8 bytes, a word adjacent
to the one to be fixed must also be rewritten. The CDC transfer is atomic, therefore the
fix is done coherently and does not interfere with the application code or data flows.
I don't understand If single error present into SDM are automatically corrected when host access to SDM (read information performed by Host) but it's not true the "viceversa" , i.e. if there is an error into SDM area reading access from eTPU (i.e the engines works using wrong data?) is not automatically corrected.
It in the RM is suggested to use CDC I thing we are talking about "re-write" correct data into SDM. But the usage of CDC shall be used when Host normally write data into SDM or it shall be used after having "found" problem and the MESR regiser shall be used to trap error and address error?