topic FFS Wearleveling and Sector Size Questions in MQX Software Solutions

FFS Wearleveling and Sector Size Questions

naftalir — Mon, 11 Jul 2016 06:50:34 GMT

Hi,

I've done a test using the example of writing/reading to/from the NAND Flash directly and got really good speed results:

~1.563MB/Sec.

Then i've done a test using the example of writing/reading to/from the NAND Flash using the MFS installed on the FFS and got bad speed results:

~28KB/Sec.

I have a few questions:

1. Why is there such a big difference between the speed results? x57 times slower using the FFS.

2. Is there is any way to disable, bypass or even remove the wearlevel that the ffs driver is using completely?

The reason for that is the BIG overhead that the WL cause and the fact i dont need it at all, since im not going to overwrite data to the NAND flash device.

3. Is there a way to change the number of bytes that the FFS does auto flush (i saw its 4KB sector-size now by default)?

Thanks to all of you who are willing to help...

p.s: I'm using MQX 4.1.1

Re: FFS Wearleveling and Sector Size Questions

danielchen — Mon, 11 Jul 2016 11:11:04 GMT

Hi Naftalir:

1. Why is there such a big difference between the speed results? x57 times slower using the FFS.

[I think this difference is result from the MFS overhead ]

2.Is there is any way to disable, bypass or even remove the wearlevel that the ffs driver is using completely?

[

If you need to bypass wearlevel, you can use NAND flash IO driver. It provides some simple POSIX like abstraction layer between the user code (application), and NFC hw module. POSIX like means that the nand flash driver provides io functions: open, read, write, close, ioctl. we have an example

C:\Freescale\Freescale_MQX_4_2\mqx\examples\nandflash

]

3 Is there a way to change the number of bytes that the FFS does auto flush (i saw its 4KB sector-size now by default)?

[Not sure I understand your meaning, could you clarify it ?]

Regards

Daniel