SystemUpdate.Load() minimum required block size?

Archive
1

If I’m using SystemUpdate.Load() like so:


SystemUpdate.Load(SystemUpdate.SystemUpdateType.Deployment, data, data.Length);

Does the “data” byte-array have a minimum required size? When data happens to be 16 bytes long, for example, I get an “Invalid param. Data is too small” exception (screenshot). Is this normal?

2

@ andre - LOL, I wish I had an app size of only 16 bytes!

Imagine this setup from GHI’s SystemUpdate Class documentation. To paraphrase:

  • We are reading the application hex file from some source. In the example, the source is a SD card.

  • We are using SystemUpdate.Load() to load the hex file in block increments. The block size in the example is 10 * 1024 bytes = 10240 bytes long.

  • Finally, at the very end, there will likely be a remainder. This remainder is (ApplicationSize modulo 10240). We still need to use Load() to load the remainder.

So if my application size is 40976 bytes, I will end up with a 16-byte remainder. And this will throw an exception once I finally Load() this portion.

If this is not intentional behavior, it can actually be quite insidious. Imagine in-field update working for some hex file sizes and not others. It would be very confusing.

3

Will check on this for you. Our IFU expert (Dat) will be back on Monday, him and his wife had a baby :slight_smile:

1 Like
4

Congrats to Dat!!! Something must be in the water at GHI!

5

Congrats, Dat!

At this rate, the new GHI building might need a daycare.

6

Congrats Dat!

7

Thanks all :)). I hope in the future, my baby will be also a programmer same as all of you here. :)))

@ Iggmoe: You are right, we will correct it in next.

Temporary solution:
if you can, how about if we write a small C# application to add 0xFF into end of that hex file? and remember correct its checksum too :)).

8

@ Dat - Congratulations to you and your family! I assume it is a boy. :wink:

9

@ Dad - Thanks for the tip on the workaround. If the new firmware will be released in 4 weeks or less, I think I can wait for the official fix. Out of curiosity, what kind of checksum is it? Simple additive checksum or CRC?

And no . . . that’s not a typo :slight_smile:

10

FYI the runtime crashes if you call SystemUpdate.Abort() twice in a row.

EXCEPTION 0x05:
cpsr=0x01000000
pc =0xa09c1cb8
lr =0x00000016
sp =0x100011a8
r00 =0xa09c1f28
r01 =0x00000000
r02 =0x00000001
r03 =0x00001000
r04 =0xa099e018
r05 =0x00000016
r06 =0xa09c1cac
r07 =0x05000000
r08 =0x00000000
r09 =0xa0024cd8
r10 =0x00000000
r11 =0x00000000
r12 =0xa099e018
EXCEPTION 0x05:
cpsr=0x01000000
pc =0xa09c1cb8
lr =0x00000016
sp =0x10001188
r00 =0xa09c1f28
r01 =0x00000000
r02 =0x00000001
r03 =0x00001000
r04 =0xa099e018
r05 =0x00000016
r06 =0xa09c1cac
r07 =0x05000000
r08 =0x00000000
r09 =0xa0024cd8
r10 =0x00000000
r11 =0x00000000
r12 =0xa099e018

11

@ Iggmoe -

They have been fixed.

cs = ~(dataCount + data[0] + data[1] +…data[dataCount -2]);

data[0,1,2,3]: address
data[dataCount -1] : where you put cs byte.

12

Or we can re-define the block size to make sure every block is bigger than 46 bytes