FRDM-K82F error when flashing large program to QSPI

[jozefl]

I have a program which is 42880 bytes large when built at Os in KDS3, however when I try to flash it to QSPI using blhost from the K80 bootloader tools, I get the following error:

Boot Section 0x00000000:

  ERAS | adr=0x00000000 | cnt=0x00000800 | flg=0x0000

  LOAD | adr=0x20000000 | len=0x00000200 | crc=0x0719eed2 | flg=0x0000

  ENA  | adr=0x20000000 | cnt=0x00000004 | flg=0x0100

  ERAS | adr=0x68000000 | cnt=0x00008000 | flg=0x0000

  LOAD | adr=0x68000000 | len=0x00000200 | crc=0x0719eed2 | flg=0x0000

  LOAD | adr=0x00000000 | len=0x000003c0 | crc=0xe83ccd71 | flg=0x0000

  LOAD | adr=0x00000400 | len=0x00000010 | crc=0x0e6bd07a | flg=0x0000

  LOAD | adr=0x68001000 | len=0x00009e88 | crc=0x5aad77e1 | flg=0x0000

Inject command 'receive-sb-file'

Preparing to send 42880 (0xa780) bytes to the target.

Successful generic response to command 'receive-sb-file'

usb hid detected receiver data abort

Data phase write aborted by status 0x2712 kStatus_AbortDataPhase

Response status = 403 (0x193) QSPI Flash Command Failure.

Wrote 31200 of 42880 bytes.

I have a different program which is 25952 bytes when built at Os and flashing it using blhost completes successfully and the program runs correctly.

If I build that same program at -O0 then the program is larger than 31200 bytes and I get the same error as before:

Boot Section 0x00000000:

  ERAS | adr=0x00000000 | cnt=0x00000800 | flg=0x0000

  LOAD | adr=0x20000000 | len=0x00000200 | crc=0x0719eed2 | flg=0x0000

  ENA  | adr=0x20000000 | cnt=0x00000004 | flg=0x0100

  ERAS | adr=0x68000000 | cnt=0x00008000 | flg=0x0000

  LOAD | adr=0x68000000 | len=0x00000200 | crc=0x0719eed2 | flg=0x0000

  LOAD | adr=0x00000000 | len=0x000003c0 | crc=0xe812f756 | flg=0x0000

  LOAD | adr=0x00000400 | len=0x00000010 | crc=0x0e6bd07a | flg=0x0000

  LOAD | adr=0x68001000 | len=0x00007d84 | crc=0xb93b887a | flg=0x0000

Inject command 'receive-sb-file'

Preparing to send 34432 (0x8680) bytes to the target.

Successful generic response to command 'receive-sb-file'

usb hid detected receiver data abort

Data phase write aborted by status 0x2712 kStatus_AbortDataPhase

Response status = 403 (0x193) QSPI Flash Command Failure.

Wrote 31200 of 34432 bytes.

Why am I getting this error and how can I fix it? As far as I'm aware the QSPI memory is much larger than 30kB, in the linker script ~1MB is allocated for the text section in QSPI and the web page for the FRDM-K82F says they are 4MB each.

[isaacavila]

Hello Jozefl,

Could you please be so kind to attach your images? (The one bigger than 30kB and the one that can be programmed correctly).

Best Regards,

Isaac

[konradgräfe]

I have the same problem on the TWR-KL82Z72M, except that the "cut-off size" seems to vary depending on the actual image. I attached my images to this post. 

First, hello_world_qspi_small.sb, which is basically the demo application from the SDK, is working:

hello_world_qspi> blhost -V -p COM5 receive-sb-file hello_world_qspi_small.sb
Ping responded in 1 attempt(s)
Framing protocol version = 0x50010200, options = 0x0
Inject command 'receive-sb-file'
Preparing to send 11776 (0x2e00) bytes to the target.
Successful generic response to command 'receive-sb-file'
(1/1)100% Completed!
Successful generic response to command 'receive-sb-file'
 - took 3.421 seconds
Response status = 0 (0x0) Success.
Wrote 11776 of 11776 bytes.‍‍‍‍‍‍‍‍‍‍‍

Second, hello_world_qspi_large.sb, which is the same but extended by a very large constant array, is not working:

hello_world_qspi> blhost -V -p COM5 receive-sb-file hello_world_qspi_large.sb
Ping responded in 1 attempt(s)
Framing protocol version = 0x50010200, options = 0x0
Inject command 'receive-sb-file'
Preparing to send 44640 (0xae60) bytes to the target.
Successful generic response to command 'receive-sb-file'
(1/1)97%Data phase write aborted by status 0x2712 kStatus_AbortDataPhase
Possible JUMP or RESET command received.
 - took 11.676 seconds
Response status = 403 (0x193) QSPI Flash Command Failure.
Wrote 43648 of 44640 bytes.‍‍‍‍‍‍‍‍‍‍‍

If I increase the size of the constant array, more bytes will be transferred, before it gets aborted.

I am running blhost from a Cygwin shell on a Windows host but I have the same issue with the KinetisFlashTool (with less output) so I guess that shouldn't matter.

[konradgräfe]

In my case the QSPI config block delivered with the SDK was wrong: the QSPI flashes don't have 4MiB each but 8MiB:

diff --git a/qspi_config_block_generator.c b/qspi_config_block_generator.c
index 7e7da1a..1169621 100644
--- a/qspi_config_block_generator.c
+++ b/qspi_config_block_generator.c
@@ -183,8 +183,8 @@ int main(void)
 const qspi_config_t qspi_config_block = {
 .tag = kQspiConfigTag,
 .version = {.version = kQspiVersionTag},
- .sflash_A1_size = 0x400000, // 4MB
- .sflash_B1_size = 0x400000, // 4MB
+ .sflash_A1_size = 0x800000, // 8MB
+ .sflash_B1_size = 0x800000, // 8MB
 .sclk_freq = kQspiSerialClockFreq_Mid, // Mid frequency
 .sflash_type = kQspiFlashPad_Quad, // SPI Flash devices work under quad-pad mode
 .sflash_port = kQspiPort_EnableBothPorts, // Both QSPI0A and QSPI0B are enabled.‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

My images are far away from those 4MiB, but with this change it is indeed working for me.