RM48L952: Which Flash Controller Register allows FAPI access to Banks 0 and 1

Hi QJ,

 It is the with 5.2.12 version of the TI bootloader and one of your colleagues named Jean Marc helped in having the bootloader load a new application image over SPI from the SDCard using the FAPI drivers (we do not use the CAN, UART of Ethernet interfaces).

So with Jean Marc how it worked was the old 5.2.12 TI bootloader would check a bit in flash that the application would set or clear depending on whether there was an upgrade to new software. If the flag is clear, it would just jump to the sector of flash with the start of the application that was built on HalCoGen drivers. If you remember, the problem was a few of our customers experienced brownouts with the TI platform when the bootloader is running which would corrupt flash but if the brownout occured when it got to the HalCoGen based application no problems. So while we are waiting on TI's official new release, we made a bootloader from the HalCoGen we are using and of course because flash is locked to erases/writes we do not have the problem with brownouts. But now, when we need the bootloader to replace the application in the other sectors of flash, we need to be able to unlock the flash via the registers to replace the application with the upgrade. This at least limits the risk to only when there are upgrades which is a couple times a year. All the other boots will be default like HalCoGen does it locking the flash so a brownout will not corrupt it.  Let me know if you want me to upload the bootloaders. So just wanted to know which registers we need to use to allow erase/writes like in the 5.2.12 version that we can change at runtime to allow for the scenerio of the upgrade.  Thank you.

Hi QJ,

  Please see attached. First zip is bootloader project based on the TI bootloader with the problem of corruption of flash with brownout when bootloader is running but we have access to flash for an upgrade, and 2nd bootloader based on halcogen (I included both HalCoGen 4.3.00 and 4.7.00 for you to see, just exclude the halcogen driver directory you do not care about), where now it is protected from brownout but we do not have access to flash anymore for upgrade.  Thank you. Tammy

2818.OriginalTIBootloaderBased.zip

4578.HalcoGenBootloaderBased.zip

Hi QJ,

Great thank you. I am using the august 2013 version of RM48 reference manual. So

FFF8 7030h FBPROT Flash Bank Protection Register
Bit Field Value Description
31-1 Reserved 0 Read returns 0. Writes have no effect.
0 PROTL1DIS PROTL1DIS: Level 1 Protection Disabled
Level 1 Protection Disable bit. Setting this bit disables protection from writing to the
OTPPROTDIS bits as well as the Sector Enable registers FBSE for all banks. Clearing this
bit enables protection and disables write access to the OTPPROTDIS register bits and
FBSE register.
0 Level 1 protection is disabled.
1 Level 1 protection is enabled.

So I have to set FBPROT bit 0 = 0?

and FFF8 7034h FBSE Flash Bank Sector Enable Register
Bit Field Value Description
31-16 Reserved 0 Read returns 0. Writes have no effect.
15-0 BSE Bank Sector Enable
Each bit corresponds to a flash sector in the bank specified by the FMAC register. Bit 0
corresponds to sector 0, bit 1 corresponds to sector 1, etc. These bits can be set only when
PROTL1DIS = 1 and in privilege mode.
0 The corresponding numbered sector is disabled for program or erase access.
1 The corresponding numbered sector is enabled for program or erase access.


and for FBPROT for the corresponding sectors in bank 0/1 have to set those bits to 1?

Is it still necessarry for me to do the _coredisableFlashEcc_() also? Thank you again, Tammy

Hi QJ,

So if the version of the technical manual is correct we have to do below, but where is PROTL1DIS to set that to 1, etc.? The manual stated that to modify the FBSE, we have to set the FMAC to the bank we want to erase/program and "...These bits can be set only when PROTL1DIS = 1 and in privilege mode...":

...
_coreDisableFlashEcc_();
flashWREG->FBPROT = 0; //0 Level 1 protection is disabled.
?? //These bits can be set only when PROTL1DIS = 1 and in privilege mode.
flashWREG->FMAC = 0; //bank specified by the FMAC register bank 0
flashWREG->FBSE = 0x7FF0; //bank 0 excl 1st four sectors with bootloader
flashWREG->FMAC = 0; //bank specified by the FMAC register bank 1
flashWREG->FBSE = 0x7FFF;//bank 1

// Erase Application Flash Region
FapiStatus = bootloader_FlashEraseApplication();

also when we set the pair FMAC and then FBSE, can we do both banks up front or do I have to integrate into where we use FAPI for the flash erase to go from bank to bank for the relative sectors on that bank? The TI based bootloader code is the same I used in HalCoGen, and everything was done before entering the flash access code, so it seems odd that would be necessary to do now. Thank you.

Hi QJ,

Thank you I understand. Did you see the zip files I uploaded (see above a few comments back)? The Fapi functions do not work over the either the HalCoGen 4.3.00 or 4.7.00 generated bootloader. The good news is the browwnout problem goes away, but then same Fapi functions no longer are able to access the flash. Thank you again, Tammy

Hi QJ,

I will attach one more time in case uploads did not get through. Please see attached. Second zip is bootloader project based on the TI bootloader with the problem of corruption of flash with brownout when bootloader is running but we have access to flash for an upgrade, and 1st zip file is bootloader based on halcogen (I included both HalCoGen 4.3.00 and 4.7.00 for you to see that both HalCoGen based bootloader have registers set so that the FAPI functions do not have access, where now it is protected from brownout but we do not have access to flash anymore for upgrade. 

Some of the differences in the sys_core.asm register for example, in the HalCoGen generated sys_core:

 ; Switch to System Mode ( Shares User Mode registers ) (M = 11111)
  cps #31

whereas the other TI bootloader sys_core.asm

; Switch back to Supervisor Mode (M = 10011)
 cps #19

So, is the problem we need to to switch from system mode to supervisor mode for the FAPI libs to have access?

Thank you. Tammy

1882.HalcoGenBootloaderBased.zip

8787.OriginalTIBootloaderBased.zip

Hi QJ, We are on the RM48. When I clicked the link I did not see the bootloader you did. Can you attached again so I can do a diff between what HalCoGen generates for the RM48 versus the MS570? Did you use a newer version of the FAPI libraries then the one TI provides with the RM48?  Thanks. Tammy

Hello Tammy,

1. Both the HALCoGen generated HW driver and the previous driver I used in the original bootloader work the same way.

2. I generated a new uart bootloader using HalCoGen generated HW driver, it works fine. This is for LS3137 device (I didn't find a RM48 HDK). You can modify it for RM48 device (change the device part number, select the little endian instead of big32), and do a test on your board. 

https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/312/2376.TMS570LS3137HDK_5F00_UART_5F00_Bootlader.7z

BTW, you need to check you HW design to figure out the power supply issue which causes the brownout. The voltage drop caused the reset is out of device spec.

The boot loader downloaded from ti.com is an example so it is not entirely vetted for use in a safety system.