Upgrading Iyonix to OS 5.24 fails

I agree that the Flash program should have better error handling

Does this program actually program the flash, or just set up parameters before calling an internal routine?
Because if there’s something you really really don’t want to go wrong, surely programming the flash is it.

Does this program actually program the flash, or just set up parameters before calling an internal routine?

DEF PROCasm(code%,flash%,image%)
p555=4
p2AA=5
c55=6
cAA=7
DQ5=&20
DQ7=&80
V_bit=1<<28
FOR X=0 TO2 STEP2
P%=code%
[OPTX
.goflash
SWI “XHourglass_On”
LDR R0, imageoff ; Image start
ADD R1, R0, #imagesize% ; Image end
LDR R3, flashoff ; The flash
ADD R4, R3, #&055*2
ADD R4, R4, #&500*2 ; R4 → &555
ADD R5, R3, #&0AA*2
ADD R5, R5, #&200*2 ; R5 → &2AA
MOV R6, #&55 ; R6 = &55
MOV R7, #&AA ; R7 = &AA
SWI “OS_EnterOS”
SWI “OS_WriteS”
EQUS “Erasing…”
SWI “OS_NewLine”


Because if there’s something you really really don’t want to go wrong, surely programming the flash is it

You could download the Tungsten flash archive and look at the contents for more entertainment.

SWI “OS_WriteS”

I think I just died a little inside…

Wow, what a lot of responses this has genrated and I haven’t even started on the Risc PC yet.
Colin, I have run your program on my Pandaboard and it comes up with one entry:
PlingBoot.Resources.PlingInternet.Bin.do .
There are 18 files in Bin, 17 Absolute and 1 Utility, but I don’t know how I could check where do occurs in these files. I could check it in !PSUtils on my Iyonix because the files were Obey files.

[Sorry about the plings above, but I couldn’t work out how to do exclamation marks without losing some of the filepath.]

I’ll have another go tomorrow, time for a beer now.
Thanks everyone for all the reponses.

Colin, I have run your program on my Pandaboard and it comes up with one entry:
PlingBoot.Resources.PlingInternet.Bin.do

Assuming that was getting used in preference to the command in the BootCommands module, that would cause the issue…

Indeed. There is no “do” in that directory in a standard HardDisc4…

David.

If you don’t see a file called ‘do’ in PlingBoot.Resources.PlingInternet.Bin then you don’t have a file on run$path with the name do which would cause a problem. If the file isn’t found OS_FSControl 37 (Canonise file name) will use the last subpath on run$path to create the full path name so presumably your run$path ends in PlingBoot.Resources.PlingInternet.Bin.

Edit: Ha. Just realised you mean !boot.Resources.!Internet.Bin I was wondering what PlingBoot was. Anyway !boot.Resources.!Internet.Bin. is normally the last path on run$path.

Perhap someone versed in ‘ARM’ code could give the Flash app a checkover!

I wouldn’t trust any changes made by anyone on this forum unless they had thoroughly tested the changes by flashing an iyonix. Even then would you trust a pretty program that had hardly been used or an ugly program which despite not having worked properly didn’t brick your machine.

Good point – but has Jon found out why – he has two bricked Iyonix’s?

then you don’t have a file on run$path with the name do which would cause a problem.

It isn’t a problem anyway if do is a module command as the module command is given priority.

but has Jon found out why – he has two bricked Iyonix’s?

Yes, faulty flash chips that have gone past their lifespan.

Personally, I wouldn’t risk attempting to flash any Iyonix unless you have the tools to externally flash and replace the chip. I replace both flash chips last week, but neither post! I need to try the process again as I flashed with the latest ROM, forgetting the botch nVidia patch is done prior to flashing, not in-line.

Is there any gain from – RO5.18 / softloading RO5.25 to Flashing RO5.24?

[edit1] by the way – what is the lifespan of the flash chips?

[edit2] could you socket the flash chips?

Yes, faulty flash chips that have gone past their lifespan.

Unless there is actually something wrong with the chip, I am wondering if there’s a programming/hardware issue.
Why? The AM29LV320DB90EI, which is what I believe the past is, is quoted in the datasheet as “Minimum 1 million erase cycles guaranteed per sector”. I don’t don’t believe that flashing an OS image a bunch of times would wear out the chip unless the Iyonix is doing something weird like storing the CMOS settings in part of it (and writing it frequently). To put that into context, the first Iyonix was created in 2002, so it’ll be about sixteen years old. To legitimately wear out the flash, you’d need to have programmed it about a hundred and seventy five times per day since purchase.
What I find more interesting was that the datasheet implies that the Vcc is 3.0v-3.6v, while at the same time stating 3.0V only. Perhaps if it is being run at 3.3V, this will reduce its lifespan? It also leaves less leeway for fluctuations.

What I find more interesting was that the datasheet implies that the Vcc is 3.0v-3.6v, while at the same time stating 3.0V only. Perhaps if it is being run at 3.3V, this will reduce its lifespan?

“Vcc for all devices . . . . . . . . . . 2.7 V to 3.6 V, Operating ranges define those limits between which the functionality of the device is guaranteed.” is the exact wording, which is taken to mean 2.7 min, 3.6 max with the 3.0V being the nominal voltage for normal operation.

3.6 is 20% above nominal, 3.3 is 10% above nominal.

Edit: The median for 2.7 and 3.6 is actually 3.15 so 3.0V would be 5% under real nominal and the quoted 3.3V is 4.76% above real nominal, so not sounding particularly off target.

Number of erase cycles per sector? Not looked at the erase code to try and figure out what’s what, Does it do an erase for every byte or erase a specific number of sectors? I know what it should do.

I have several AM29LV320DB90 that have developed faults, they start to suffer from bitrot, so bits either remain a 1 or a 0.

It could be down to bad batches, cosmic rays or just aging via heat/power cycles etc.