Flash erase not working

[cvu1]

Hi All,

I have a problem when erase the flash.

Gpu_Hal_WaitCmdfifo_empty(phost); // Ensure it's attached
Timing(0);
Gpu_CoCmd_FlashErase(phost);
Gpu_Hal_WaitCmdfifo_empty(phost); // Wait for command to finish
Timing(1);

From Time 0 to 1 take only 268.5 us. And of course, this is not working. I expect it should longer.

This not seem right to me but I have no idea how to trouble this. I have a custom EVE4 display with 64MB Flash from Riverdi. The Display working fine with EVE Asset Builder that can erase, update flash, etc. But when I try to erase and write blob from MCU. It seems like not working. The Erase command finished too fast. Any idea?

[RudolphRiedel]

Yes, it usually takes several seconds to over a minute to erase the flash, this executes the chip-erase command of the chip.
But the display needs to be initialized with the flash having reached status BASIC.

And while I have seen the API, this is not my code, there is no way to actually tell which implementation you actually used, if these functions are actually any good and if the display got initialized before this.

[cvu1]

// Wait for the command FIFO to be empty
Gpu_Hal_WaitCmdfifo_empty(phost);

// Attach the flash if necessary
Gpu_CoCmd_FlashDetach(phost);
App_Flush_Co_Buffer(phost);
Gpu_Hal_WaitCmdfifo_empty(phost);

Gpu_CoCmd_FlashAttach(phost);
App_Flush_Co_Buffer(phost);
Gpu_Hal_WaitCmdfifo_empty(phost);

UARTCharPut(UART3_BASE, 0xDD);
// Send the Flash Erase command
Gpu_CoCmd_FlashErase(phost);
Gpu_Hal_WaitCmdfifo_empty(phost);  // Wait for command to finish

UARTCharPut(UART3_BASE, 0xCC);
delay_ms(100);

the API is from bridgetek. I also make sure that the flash is in basic mode by detach and attach but seem no effect

[RudolphRiedel]

Have you also checked in REG_FLASH_STATUS that the flash actually is in BASIC mode?
I do not know how long it takes from issuing CMD_FLASHATTACH to REG_FLASH_STATUS returning 2 for FLASH_STATUS_BASIC.

And yes, the library code is from Bridgetek, but this repository is not. :-)

[cvu1]

`TimingIO(0);

App_WrCoCmd_Buffer(phost, CMD_FLASHERASE);
App_Flush_Co_Buffer(phost);
Gpu_Cmd_waitFlush(phost);

TimingIO(1);`

This one working fine with EVE3 (BT815) (https://riverdi.com/product/ritft70capux?srsltid=AfmBOoo215CQFJRcskMi_vYH2GSwCWtxgBKrS2zioNMJ-eET51zE_s7S)

When replaced with EVE4 (BT817Q). Doesn't works. (get correct CHIP ID and able display some dummy screen on the LCD Screen)

One thing I note with EVE4 (BT817Q) that when using EVE Asset Builder 2.12.2. The Erase or Program option won't works if it missing blob file (mean can't switch to full state). The only option works using EAB is Update flash.

So I will work around to get the blob install first. What I think is write the blob to RAM_G first and then use CMD_FLASHUPDATE to write blob to flash and switch the flash to full state for future update.

[RudolphRiedel]

Hmm, ok, I just tried this myself with my RVT101HVBNWC00-B.
And I can confirm that CMD_FLASHERASE is not working as expected.

I added reading of REG_FLASH_STATUS directly before the command, it returns 0x02.
I also checked the sequence with the logic analyzer and the command was sent correctly.
This did nothing at all, the flash remained intact.

When I am however using my EVE_init_flash(); function which checks the status and issues CMD_FLASHFAST when status BASIC is reached, reading REG_FLASH_STATUS returns 0x03 and the flash gets erased.
Or at least the content I stored in the flash before is gone.
The erase command takes very little time though.

Odd.

[RudolphRiedel]

I am trying to find a fitting Micron 512Mb datasheet, turns out Micron does not provide these to the public.
The Flash chip on my RVT101H is definitive a Micron, I only do not know which one.
This might be a MT25QL512ABB.
The bulk erase time for this device is 153 to 460 seconds.

Ok, this reads Device ID Data from the flash chip thru the BT817:

EVE_cmd_flashdetach();
EVE_cmd_flashspidesel();
uint8_t tx_data[] = {0x90, 0, 0, 0, 0, 0};
EVE_cmd_flashspitx(4, tx_data);
EVE_cmd_flashspirx(0x000f5000, 2);
EVE_cmd_flashspidesel();

tx_data[0] = 0x9f;
EVE_cmd_flashspitx(1, tx_data);
EVE_cmd_flashspirx(0x000f5002, 5);

EVE_cmd_flashattach();

 flash_data[0] = EVE_memRead8(0x000f5000);
flash_data[1] = EVE_memRead8(0x000f5001);
flash_data[2] = EVE_memRead8(0x000f5002);
flash_data[3] = EVE_memRead8(0x000f5003);
flash_data[4] = EVE_memRead8(0x000f5004);
flash_data[5] = EVE_memRead8(0x000f5005);
flash_data[6] = EVE_memRead8(0x000f5006);

And the result is:
0x20 - Micron
0xba - 3V
0x21 - 1Gb
0x10 - remaining ID bytes
0x40 - Extended device ID: 2nd generation, standard BP scheme, 0 = HOLD, HW Reset not available, sector size = uniform 64KB

Ok, it is Micron, but this thing is a 512Mb, not 1Gb, so not a MT25QL512ABB?