Linux/AM5728: EMIF tool generates data for only one EMIF

Wilson Perez

Part Number: AM5728

Tool/software: Linux

Using version 1.0.0 from March 20, 2016, 2 EMIF interfaces are selected from the "Step 1-System Details" sheet. However, only one emif_regs struct is generated in the "Register Values" sheet. How can the second emif_regs struct be generated so that it can be used in fucntion emif_get_reg_dump (board.c)? Using the same struct for EMIF1_BASE (case 1) and EMIF2_BASE (case 2) appears to cause the board to hang.

Thanks.

Wilson

over 7 years ago

0 Biser Gatchev-XID over 7 years ago

TI__Guru**** 393215 points

Did you fill in correctly parameter 4 in section 1A?

0 Wilson Perez over 7 years ago in reply to Biser Gatchev-XID

Intellectual 355 points

Please find attached screenshots of our configuration. It appears parameter 4 in section 1A is selected correctly. Our system consists of 4 MT41K512M16HA-125IT:A DDR3 SDRAM for a total of 4GB.

Thanks!

Step 1

Step 2

Step 3

0 Biser Gatchev-XID over 7 years ago in reply to Wilson Perez

TI__Guru**** 393215 points

I have asked the DDR experts to help on this. They will respond here.

0 Wilson Perez over 7 years ago in reply to Biser Gatchev-XID

Intellectual 355 points

Biser,

Any word on this matter?

Thanks.

0 Biser Gatchev-XID over 7 years ago in reply to Wilson Perez

TI__Guru**** 393215 points

Sorry about the delay. I have escalated the request.

0 Brad Griffis over 7 years ago in reply to Biser Gatchev-XID

TI__Guru*** 125430 points

You are observing the expected behavior. There should be only a single structure of timing data given that both EMIF's use the exact same timing.

0 Wilson Perez over 7 years ago in reply to Brad Griffis

Intellectual 355 points

Thanks, Brad.

So, in function emif_get_reg_dump (board/ti/am57xx/board.c) the same emif_regs struct should be used for both case 1 and case 2. Correct?

Wilson

0 Brad Griffis over 7 years ago in reply to Wilson Perez

TI__Guru*** 125430 points

Correct.

Also, thank you for this question. I hadn't noticed that u-boot actually is using two separate structures (one per EMIF). It looks like that has already led to issues, e.g. at one point the timings were updated, but only for EMIF2! I'm following up with the software team to see about straightening this out.

Processors

Processors forum

Linux/AM5728: EMIF tool generates data for only one EMIF