« first day (1796 days earlier)      last day (3147 days later) » 

5:47 AM
morning
 
 
3 hours later…
8:51 AM
@crasic Do I remember rightly that you were fiddling with u-boot a while back?
 
9:01 AM
Should I ask a ... very narrow scope ... U Boot question on Unix/Linux stack?
 
 
7 hours later…
3:41 PM
In case someone here knows bundles of u-boot, this is my problem now: unix.stackexchange.com/questions/232548/…
 
@Asmyldof did you check browser history for the blog? Firefox can show detailed history per day, but I don't know what you are using.
I'm not familiar with the board and the mtd device. Does the output of /proc/partitions make sense to you? Does it use partition tables and if so what does that look like?
also interesting is that you didn't find a filesystem, but rather plain text config file. If I understood correctly.
does fdisk -l /dev/mtd work? (WRT partitions)
Does it use some sort of boot sector that you missed to update?
 
3:58 PM
@jippie Good questions. I'll have a look
The whole thing is that u-boot is the boot sector, so to speak
Those env variables are what it uses to unpack and init the kernel
 
hmmm there is a difference between boot sector and boot loader
but I'm not really familiar with u-boot
 
It's an Embedded ARM, so it just loads from 0x0000000
which is where u-boot lives
 
is it memory mapped?
 
To some degree, but what degree is all in ChingChong documents
EmbedSky is some Guangdong / HongKong company as far as I could see and most references online confirm that assuming the TQ2440 to be similar to Samsung's own dev board is hazardous at best
Therein lies my 2 month frustration with "Koole Kontrols" in Goes. Avoid them if you ever see them bid on a project you are linked to, they will cost you decades of your life
To be honest, I was semi-happy to find that the board comes with u-boot, because at least the German mainly responsible for that knows that writing shit down is useful
Something KK hasn't heard of, the concept of "writing shit down" alone
 
isn't offset related to the absolute location on mtd, so disrecarding the partitions?
divide by four on ARM could very well be related to 32 bit words
 
4:09 PM
Possibly.
Fair point
Almost going home, so can do maths and experiments in a bit
At least I do now know for 100% sure the block is in the first 256kbyte, being MTD0
 
no offset?
 
The u-boot stored the config in plain ascii
MTD0 reports start at 0x00000
 
are you telling me there is at 0x0000 a plain text config file?
dunglish</
 
Thing is, those defines may very well not be right, the 0x1F0000, and such, since this project is riddled with "this source here, oh and there, oh, and, by the way there"
So while there is only one u-boot source in the dev-env provided, there is no guarantee that what is provided is all there is and that that is what they used
No
At 0x00000 all kinds of blarb lives
but in the sector that has that blarb, mtd0, also plain text lives
Which one can search for with "edit->search" in whichever is one's favourite hex editor
things like "baudrate=" and such
And I found this block of text at that 0x367FC / 0x367F8 point I mention, which turned out to be a location made almost entirely out of unobtainium as far as the tools I have are concerned
 
Are you being confused by a filesystem layer?
 
4:15 PM
Which leaves: 1. prod the NAND flash myself on 3 different system configs
No, this is how u-boot works
just is
 
m
 
2. try all possible numbers between 0 and end-of-mtd0 for all colums of the config
3. Give up, refund, flush 110 billable hours down tha drain
I find 1. an option not befitting a final delivery to a customer for a system that will not ever be recoverable if I do it wrong, or it could go wrong in one isolated instance
I find 3 very unfavorable
And 2 is my fall back, unless a stackerman/woman tells me how to translate 0x367F8 +/-4 + 0x01C0(??) into a number I can insert into the .config file that goes with the tool
 
/me shrugs
I'm happy to have a look at the flash image if you didn't overlook anything obvious.
 
When I'm back home I can mail it to you. No propriatary code to speak of in it anyway
But I don't think it'll help much, to be honest
 
neither do I
 
4:23 PM
I have put 14 hours of google into the results I have now
 
but I may have some tools that spit out useful info
 
I'm happy to send you the u-boot folder zipped as well, with, if you like, the EmbedSky.h separate
or, for in advance, it's in include/configs/
Do I know the whole structure by heart? Yes I do! Why? Because I already have nightmares about EmbedSky boards!
>.<
@jippie Is there a rationale for everything from the headerfile definitions to have to be divided by 8?
 
@Asmyldof /me shrugs
64 bit words?
 
Maybe, if, possibly, the OFFSET is to the end of the block of variables, 1F0000 / 8 => 0x03DFFF, is inside MTD0, subtract 0x04000 ....
Nope, still not covering the ascii values I found
:/
64 bit words would be highly curious, as all parts of the Samsung proc (the one component I DO have a readable datasheet of) are 32bit ARM
 
how about endianess?
that shouldn't be a problem I guess
 
4:39 PM
I don't know. I tried navigating through the u-boot env-tools sources to figure out how it uses the numbers, but it kept giving me stack-overflows
Admittedly, I did only use Bluefish inside the dev-env. Nice enough, but by all measures not the most easy system to nagivate between function calls and returns, AFAIK
(re stack-overflows: me personally, inside my brain)
Well, if for nothing else, I do have 10 extra rep on unix now :-)
Eh
I just re-clicked a file I had already deleted in the browser downloads bar and it decided to re-download....
:/
 
@Asmyldof Other stacks just give away the rep like it's going out of style. On EE, you'd have been downvoted and closed by now for not already knowing the answer.
 
@ThePhoton To circle back to your earlier comment re: Q&A here ?
:-P
 
@Asmyldof What?
 
High number of autists here, etcs
I wanted to reply to your correction, but I forgot with what, by the way. Brain haunted with EmbedSky daydreams
Well, anyway, even if it would get closed, at the very least I tried, and I made a copy in html>odt>word/pdf to send to the customer to show I didn't clack off all weekend
slack*
So all the typing isn't wasted either way
ld home, me
 
4:58 PM
@Asmyldof EX HOME,HOME' in Z80 speak
 
5:32 PM
:-)
dinner making time
 
 
6 hours later…
11:45 PM
Hello @Ondeckshooting
 
Hello @ThePhoton I appreciate your willingness to help me
 
First question, do you have an idea what your design will cost to build, and do you have that much money backing you up?
@Ondeckshooting You might want to read some stuff like Bunnie's blog for some background on working with manufacturers.
 
Not a single board alternative one which is what I was hoping for. But if I were to buy the parts and than assemble them it would be $110 each with the display more or less guessed depending on if I can get a driver in the mix for the display. I have no idea where I can find drivers or if you get get a driver.
Right now I am trying to create a proposal for a game changing device for our industry that would be widely adapted it I can deploy it right. We work in a sports community that works as it did in the early 2,000's with minimal computer influence. Our company is a software company that needs an idea to stay alive. I am their 3rd and latest employee.
 
@Ondeckshooting Where are you located (just country or state if you're in the US)
 
Mainly the United States and Great Britain I do not know our demographics for any other region.
 
11:58 PM
I am currently working with one of the biggest CM companies, and they are willing to do runs of 10 and 20 boards for me. I don't know if they are willing to do these small jobs as a normal business or if they are hoping my company will give them much bigger jobs in the future. (Also, my product is more like $5k/unit rather than $100)
 

« first day (1796 days earlier)      last day (3147 days later) »