rehius / usk

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Picofly does not glitch with rebuild nand(original prodinfo present) but boots into ofw.

hanswurst31 opened this issue · comments

Picofly does not glitch with rebuild nand(original prodinfo present) but boots into ofw.

Error code:

Picofly FW 2.75 short long short
Picofly FW 2.76 long short long short

-The installed picofly glitch to hekate with corrupt nand.
-The picofly glitch to hekate with empty emmc.
-The picofly(including same flex) glitches fine in another switch where the original nand is present.

-NOT SURE: i might be remembering wrong but the picofly glitches to hekate with rebuild nand WITHOUT original prodinfo present.

Tested with erista unpatched/patched and Mariko. Not tested with lite and OLED.

I use @sthetix tutorial. I asked him what the issue could be but he unfortunately told me, that he doesn't use picofly.

Do you guys think that this is fixable by FW update for picofly?

Edit: tested also with hwfly. Same result.

commented

Did the picofly blink a few times slowly blue followed by a short yellow blink after first startup after flashing the firmware? This is indicating, if it succeeded in learning the glitch timing. If this does not happen at all, normally a connection is likely bad. Diode measurements and cleaning all flux residues especially on B (reset point) would be my first step.

And btw: 2.76 has been revoked afaik - so 2.75 is the latest stable release.

I guess, this guide is already known:
https://github.com/Ansem-SoD/Picofly/blob/main/PicoFlyGuideV.6.2.pdf
= No eMMC block 0 read (eMMC init failure?)

Normally, the pico should glitch and boot the payload, if all connections at least to the CPU side are present - the eMMC can even be removed, as I understand it.
I had once an unavailable eMMC with a CMD resistor partly lifted off one pad but it was still connected to CPU side (OLED) and I was able to reach hetake as usual but eMMC was just not showing up at all.

No it doesn't. Installation is perfect. If it would be an installation issue, it wouldn't glitch with corrupt nand, or empty emmc. I never have problems with picofly installation when orig nand is present. Even OLED always works.

Btw tested with hwfly also. Same result.

commented

Than it seems not to be picofly related, if HWFly does not resolve it as well.

It's Nand structure related it seems. Picofly fw doesn't seems to be able to handle the nand structure of a rebuild nand. This should be fixable by FW adjustment.

commented

if HWFly can not, why should picofly be able to do it?

And how did you rebuild your nand in the first place? Did you use another switch or is it an OLED?
Just because some simple USB-to-eMMC sticks will produce issues.
I have this one for rebuilding: https://www.tindie.com/products/ignas/emmc-reader-for-hac-emmc/

Erista unpatched/patched and Mariko. Not tested with lite and OLED.

Dump Nand(dumping prodinfo and prod keys is enough) via Hekate and proceed with the following tutorial:

https://m.youtube.com/watch?v=FDAOghK2tHE

Then flash the rebuild nand via nxnandmanager/Hekate to emmc.

A previously "bricked" console will boot into ofw after this. The picofly/hwfly just doesn't glitch to hekate anymore afterwards.

commented

Nice to know, that rebuilding is already possible just via USB-C.
Could be, that during mmc_initialize() the emmc response is not fitting to the expected result.
For example as it tries to step through the emmc tables with this for-loop:
for (int i = 0; i < sizeof(mmc_init_table)/sizeof(mmc_init_table[0]); i += 4)

Are really all partitions available and accessable on the rebuild nand? Tegraexplorer can browse them all?
image

Just tested a console with a completly dead emmc. So I rebuild a Nand without orig prod info present and it glitches fine.

Tutorial:

https://youtu.be/mH8osmtPFlA

That's what I was suspecting. Maybe the rebuild nand is written in a slightly different position where the switch can handle it but picofly cant. Just talked with sthetix (tutorial creator) and he told me that instead of restoring the rebuild nand via nxnandmanager straight to the emmc, I could restore it via restore section of hekate. Maybe this will solve the problem.

I would guess that every partition is available otherwise the ofw wouldn't boot. Unfortunately I don't have those console here anymore. I will factor reset, update a working unpatched erista to the latest fw and rebuild a Nand. Then I can check and also compare both Nand. Maybe you are interested in the nand copies?

commented

I never tried with any donor images to rebuild - always had all backups till now.l and used hetake.

I am not really deep into picofly firmware dev so the images would not help me much.

Picofly FW 2.75 short long short
Picofly FW 2.76 long short long short

Either hardware issues or unsupported eMMC, try another eMMC here.
It does not matter what is written within the eMMC, should always work
However there is one intentional thing - the first boot after system update / eMMC BOOT rewrite is done into the OFW to properly burn fuses. But the following boots should go into hekate / preloader since the eMMC is burned with the custom loader.