topic Re: IAP write block not blank in LPCXpresso IDE

IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:12 GMT

Content originally posted in LPCWare by blasiis on Wed Apr 20 08:56:29 MST 2011
I'm using LPC1758 with LPCXpresso
I have try to use the IAP command "copy ram to flash" when the sector flash it isn't blank, but i need only to move a bit from 1 to 0.
Using this function other bit is write to 0 !!!
Why ? I cannot write a 256 byte block already write ?

Thanks

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:13 GMT

Content originally posted in LPCWare by larryvc on Wed Apr 20 14:15:21 MST 2011
Ciao blasiis,

Could you post your IAP call or your code so that we have more information to help you.

Bouna notte,
Larry - O quando mi trovo in Italia, Lorenzo:)

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:13 GMT

Content originally posted in LPCWare by blasiis on Wed Apr 20 23:20:59 MST 2011
Buongiorno

This is my problem:
I write the last 256 block in the sector 26, this loop run every 1 ms

uint8_t flash_buf[FLASH_BUF_SIZE];

addSett = 0x00060F00;

     add = addSett ;

     // read 256 block
     for (i = 0; i <FLASH_BUF_SIZE; i++)
       {
       flash_buf = (*(uint8_t *)add);
       add ++;
       }

tmp = flash_buf[pnt] <<1;       // force to 0 the last bit at level 1
flash_buf[pnt] = tmp;
IAP_WriteFlash(26, 26, (unsigned)addSett, (unsigned *)flash_buf, FLASH_BUF_SIZE);

void IAP_WriteFlash(unsigned Start, unsigned End, unsigned flash_address, unsigned * flash_data_buf, unsigned count)
{
    IAP_SectorPrepare(Start, End, SystemCoreClock/1000);
        IAP_WriteData(SystemCoreClock/1000, flash_address, flash_data_buf, count);

void IAP_SectorPrepare(unsigned start_sector,unsigned end_sector,unsigned cclk)
{
    __disable_irq();
      param_table[0] = PREPARE_SECTOR_FOR_WRITE;
      param_table[1] = start_sector;
      param_table[2] = end_sector;
      param_table[3] = cclk;
      iap_entry(param_table,result_table);
    __enable_irq();
}

void IAP_WriteData(unsigned cclk, unsigned flash_address, unsigned * flash_data_buf, unsigned count)
{
    __disable_irq();
      param_table[0] = COPY_RAM_TO_FLASH;
      param_table[1] = flash_address;
      param_table[2] = (unsigned)flash_data_buf;
      param_table[3] = count;
      param_table[4] = cclk;
      iap_entry(param_table,result_table);
    __enable_irq();
}

When I force to 0 the bit, in lpcxpresso memory display, other bit in other addres (3-4 byte previuos)go to 0 !!!!!
In this phase I don't use erase sector, but i whish only delete 1 bit each loop !
I think the problem is that i try to write a 256 block not erased, it is possible this ?

Thanks

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:14 GMT

Content originally posted in LPCWare by TheFallGuy on Thu Apr 21 00:08:29 MST 2011
You cannot write a single bit. The minimum you can flash is a sector, and to completely write it, it must be first be erased. The problem is the way that flash works. It is erased to all 1's and a write, can only change a 1 into a 0.

You should also re-think how you are writing to flash, as it has a maximum number of times it can be written to. I think NXP claim 10,000 write cycles. If you are trying to write every 1ms, that would imply you would write 10,000 times in 10 seconds! Even if they are conservative and are an order of magnitude out, that would still only last 100 seconds!

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:14 GMT

Content originally posted in LPCWare by blasiis on Thu Apr 21 01:01:30 MST 2011
Sorry for misunderstanding, 1 ms is only a time between 2 write command, but obviously after 8 write cycle the function is stopped (8 write cycle = 8 bit)!!
I know that is possible only force a bit from 1 to 0 in the memory flash.

Sequence Example :
before write, byte = 11111110
write without erase with byte = 11111100
after write, byte = 11111100, but in other address (3-4 before) a bit is go to 0 !!!
My question is: Before write command i must erase all sector also if I move bit only from 1 to 0 ??

By

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:15 GMT

Content originally posted in LPCWare by Ex-Zero on Thu Apr 21 03:14:21 MST 2011
1. The correct sequence is: prepare-erase-prepare-write

2. If you want to change 1 bit of a sector, you have to read this sector in a buffer, change that bit and then write changed sector back to flash

3. You should read the return code of the IAP command to control if your command is executed:

Quote:

32.8.10 IAP Status Codes
0 CMD_SUCCESS Command is executed successfully.
1 INVALID_COMMAND Invalid command.
2 SRC_ADDR_ERROR Source address is not on a word boundary.
3 DST_ADDR_ERROR Destination address is not on a correct boundary.
4 SRC_ADDR_NOT_MAPPED Source address is not mapped in the memory map. Count value is taken in to consideration where applicable.
5 DST_ADDR_NOT_MAPPED Destination address is not mapped in the memory map. Count value is taken in to consideration where applicable.
6 COUNT_ERROR Byte count is not multiple of 4 or is not a permitted value.
7 INVALID_SECTOR Sector number is invalid.
8 SECTOR_NOT_BLANK Sector is not blank.
9 SECTOR_NOT_PREPARED_FOR_WRITE_OPERATION Command to prepare sector for write operation was not executed.
10 COMPARE_ERROR Source and destination data is not same.
11 BUSY Flash programming hardware interface is busy.

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:16 GMT

Content originally posted in LPCWare by blasiis on Thu Apr 21 03:23:28 MST 2011

Quote: Zero
1. The correct sequence is: prepare-erase-prepare-write

2. If you want to change 1 bit of a sector, you have to read this sector in a buffer, change that bit and then write changed sector back to flash

3. You should read the return code of the IAP command to control if your command is executed:

I work as in the point 2 !!!

 // read 256 block 
     for (i = 0; i <FLASH_BUF_SIZE; i++)
       {
       flash_buf = (*(uint8_t *)add);
       add ++;
       }

// change a bit
tmp = flash_buf[pnt] <<1;       // force to 0 the last bit at level 1
flash_buf[pnt] = tmp;

// prepare and write
IAP_WriteFlash(26, 26, (unsigned)addSett, (unsigned *)flash_buf, FLASH_BUF_SIZE);

The problem is that when execute write flash the second time with byte not a FF, but I change only a bit from 1 to 0 (tmp = flash_buf[pnt] <<) is changed also other bit. !!

The return code is CMD_SUCCESS !!

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:16 GMT

Content originally posted in LPCWare by Ex-Zero on Thu Apr 21 03:35:22 MST 2011
Can't see there any erase command in your code :confused::confused:

Write without erase will not work !!

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:17 GMT

Content originally posted in LPCWare by blasiis on Thu Apr 21 04:34:29 MST 2011

Quote: Zero
Can't see there any erase command in your code :confused::confused:

Write without erase will not work !!

This is the problem, i want write without erase, but only bit 1 to 0 !

Quote: Zero
2. If you want to change 1 bit of a sector, you have to read this sector in a buffer, change that bit and then write changed sector back to flash

I want execute this operation.

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:17 GMT

Content originally posted in LPCWare by Ex-Zero on Thu Apr 21 04:44:02 MST 2011

Quote: blasiis
This is the problem, i want write without erase...

Then you have to use something external like EEPROM / FRAM or RTC GPREG0-4.

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:18 GMT

Content originally posted in LPCWare by blasiis on Thu Apr 21 05:11:14 MST 2011

Quote: Zero
Then you have to use something external like EEPROM / FRAM or RTC GPREG0-4.

I want use flash as eeprom, but i don't have it !
I want use flash without erase, any time that I write flash the modification is only bit 1 to 0, NOT 0 TO 1 that it isn't possible, I know !
When all bit is a 0 I execute the erase sector, in this way I saving many life cycles
I am in debug mode with LPCXpresso

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:19 GMT

Content originally posted in LPCWare by TheFallGuy on Thu Apr 21 08:00:55 MST 2011
Have you looked at the LPC1777x and LPC178x families? They have a 4K EEPROM.

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:19 GMT

Content originally posted in LPCWare by Ex-Zero on Thu Apr 21 08:23:02 MST 2011

Quote: TheFallGuy
Have you looked at the LPC1777x and LPC178x families? They have a 4K EEPROM.

Sounds lika a good idea, but

1. Product status: Development :mad:

2. Package: >= LQFP144 :mad::mad:

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:20 GMT

Content originally posted in LPCWare by rmteo on Thu Apr 21 08:39:49 MST 2011
Several companies (such as Microchip and ST) have published app notes describing how to use FLASH memory as EEPROM. Using these software techniques you can achieve [B][U]effective[/U] [/B]FLASH endurance equivalent to EEPROM up to several million write cycles. I have used the MCHP AN1095 app note - code is in C and should be easily adaptable.

______________________________________
Why use 8 bits when you can have 32?
ARM CORTEX Rules!!! :D:D:D

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:20 GMT

Content originally posted in LPCWare by blasiis on Thu Apr 21 09:19:20 MST 2011

Quote: rmteo
Several companies (such as Microchip and ST) have published app notes describing how to use FLASH memory as EEPROM. Using these software techniques you can achieve [B][U]effective[/U] [/B]FLASH endurance equivalent to EEPROM up to several million write cycles. I have used the MCHP AN1095 app note - code is in C and should be easily adaptable.

I cannot change Microprocessor !
I know the techniques for use flash as eeprom (2 bank in alternation).
But this case is different: I need to write many time (15 min ) a timer value decremented, for solve the endurance flash cycle (10.000), I have think to start from a 4k sector with all bit in at 1 (flash erased) and any 15 minutes erase a single bit in byte in sequence.
Here comes my problem: When write a byte from FF to FE in a 256 block where other byte it isn't a FF some bit goes wrongly to zero in the same block !!

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:21 GMT

Content originally posted in LPCWare by larryvc on Thu Apr 21 10:19:02 MST 2011

Quote: blasiis

Here comes my problem: When write a byte from FF to FE in a 256 block where other byte it isn't a FF some bit goes wrongly to zero in the same block !!

This sounds like a hardware problem, bad flash.

Do you have another chip to test?

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:22 GMT

Content originally posted in LPCWare by blasiis on Thu Apr 21 23:47:12 MST 2011

Quote: larryvc
This sounds like a hardware problem, bad flash.

Do you have another chip to test?

You had a Good idea :D !!
But unfortunately after I tried the same result . . . . :(

Viewed with lpcxpresso vindow memory but also with array in code:
The 26 sector is blank (all bye a FF)
Write 0x00060000 = 00 - no problem
Write 0x00060001 = 00 - no problem
Write 0x00060002 = 00 - also the address 0x0006000B = BF !!!!!

Remember that when write, befor i not erase the sector, but i move the bit only to 0 !!!!!

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:22 GMT

Content originally posted in LPCWare by larryvc on Fri Apr 22 00:07:40 MST 2011
blasiis,

Bestia maiale.:D

I see you have

tmp = flash_buf[pnt] <<1; // force to 0 the last bit at level 1
flash_buf[pnt] = tmp;

What is "pnt" defined as and what is it set to?

If you want, export your code and attach it here. I can try it on an LPC1769.

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:23 GMT

Content originally posted in LPCWare by blasiis on Fri Apr 22 00:35:21 MST 2011

Quote: larryvc
blasiis,

Bestia maiale.:D

I see you have

tmp = flash_buf[pnt] <<1; // force to 0 the last bit at level 1
flash_buf[pnt] = tmp;

What is "pnt" defined as and what is it set to?

If you want, export your code and attach it here. I can try it on an LPC1769.

I have change the code now for make test.
Try this wuth lpcxpresso:
1) Mass Erase
2) Load application with write flash in a loop (1 second) like this:

uint8_t flash_buf[FLASH_BUF_SIZE];
addSett = 0x00060000;
 add = addSett ;
 
// main every 1 second for example 
     // read 256 block 
     for (i = 0; i <FLASH_BUF_SIZE; i++)
       {
       flash_buf = (*(uint8_t *)add);
       add ++;
       }

IAP_WriteFlash(26, 26, (unsigned)addSett, (unsigned *)flash_buf, FLASH_BUF_SIZE);   <--- break
// end main


void IAP_WriteFlash(unsigned Start, unsigned End, unsigned flash_address, unsigned * flash_data_buf, unsigned count)
{
    IAP_SectorPrepare(Start, End, SystemCoreClock/1000);
        IAP_WriteData(SystemCoreClock/1000, flash_address, flash_data_buf, count);


void IAP_SectorPrepare(unsigned start_sector,unsigned end_sector,unsigned cclk)
{
    __disable_irq();
      param_table[0] = PREPARE_SECTOR_FOR_WRITE;
      param_table[1] = start_sector;
      param_table[2] = end_sector;
      param_table[3] = cclk;
      iap_entry(param_table,result_table);
    __enable_irq();
}

void IAP_WriteData(unsigned cclk, unsigned flash_address, unsigned * flash_data_buf, unsigned count)
{
    __disable_irq();
      param_table[0] = COPY_RAM_TO_FLASH;
      param_table[1] = flash_address;
      param_table[2] = (unsigned)flash_data_buf;
      param_table[3] = count;
      param_table[4] = cclk;
      iap_entry(param_table,result_table);
    __enable_irq();
}

3) Add break at IAP_WriteFlash call (not run for too many write)
4) Run
5) When break change in flash_data_buf [0] = 0;
6 ) Run
7) When break check in debug window mwmory address 0x00060000 = 0
8 ) change in flash_data_buf [1] = 0; ([0] already = 0)
9) Run
10) When break check in debug window mwmory address 0x00060001 = 0
11) change in flash_data_buf [2] = 0; ([0 and 1] already = 0)
12) Run
13) When break check in debug window mwmory address 0x00060002 = 0 and 0x0006000B = BF !!!!!!!!!! ????????????

Re: IAP write block not blank

lpcware — Wed, 15 Jun 2016 21:53:24 GMT

Content originally posted in LPCWare by larryvc on Fri Apr 22 01:11:30 MST 2011
Why don't you try using a printf in semihosting mode to read and display the values in sector 26? This way you will be able to determine if the debugger is returning correct values from your hardware or not.

It will be slow, but there are ways to speed up the output to the console. See this post:

http://knowledgebase.nxp.com/showpost.php?p=6761&postcount=3