Philips QFU1.1 Kernel Panic

[Beeldbuisje]

Typical issues of a QFU chassis: doesn't start sometimes, sometimes nothing after some hours etc.
When not starting, this thing gives a Kernel panic when logging.

So first things first: reballing of the Fusion, same behavior. Tried a NAND from a working SSB from exactly the same tv (40PFL8007K/12), the same Kernel panic.

I know too little from a Linux system to interpret this error. Could a Linux professional give me some help pls.
When I log the working SSB, I've got exactly the same till the Kernel Panic.


Please find the log in attached PDF.

[Diah]

Re: Philips QFU1.1 Kernel Panic

swap NAND will bring nothings to you because each boards had its owned security Keys generated at update, any panic log dont think its firmware or Kernel.... so the main fault remain the Fusion CPU. how much you sure on yours reballing work.. still the PCB sandwich invisible to you.. so it was always advised after removing the chip we exam the resistances of the soldering joint under the chip before placing the chip again.

[Maalobs]

Re: Philips QFU1.1 Kernel Panic

If you search for "error" and step through all the hits, you'll find these two:

Reading out data:
04 unit:
waiting ECC result ready
01 bits error in the unit.
CWord at 0x1C, Error Mask:0x2000000.

06 unit:
waiting ECC result ready
01 bits error in the unit.
CWord at 0x45, Error Mask:0x100.

ECC error sounds like bad RAM, and from the location in the log output (preceding U-Boot starting and reading the NAND) it appears that it is RAM-testing.
This might be the cause of your troubles.

I was expecting there to be an error code for bad RAM detected during POST which would throw the TV into protection mode, but when inspecting the error code overview in table 5-2 in the service manual, apparently there is no code for this error.
That would explain why you get random kernel panics after the system has been booted on RAM with bit errors.

If you also search for "warn" in the log output, you'll find this line:

UBI warning: ubi_eba_init_scan: cannot reserve enough PEBs for bad
PEB handling, reserved 4, need 40

It's surrounded by a great many "fixable bit-flip detected", soft errors in NAND consistency, before and after that line.
I think the warning is telling you that the Bad Block Table of the NAND has already been maxed out and no more bad blocks can be mapped away.

The 0x2C38 Device ID indicates that it's a Micron MT29F8G08ABABA 1 Gbyte NAND.
They have been out of production for long time now, but if you search on AliExpress, maybe you can still find someone selling them in a "slightly used" condition.

On the very last line you have this:

Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(249,0)

Is the filesystem mount failing because of a too badly damaged NAND or by bad RAM, I honestly can't tell.

PS:
It's completely reasonable to attach log textfiles as txt-files in the future.

[Davi.p]

Re: Philips QFU1.1 Kernel Panic

From what i know the most possibile answer of theese behaviours is that reballing is not come well, in other cases even if reballing is ok since theese are smart tv so them saves data on to flash rom, with the previous state of the cpu, data loss could happen and have ruined sectors content, so a FS fixing or software rewriting is neccessary.

[Wiebotje]

Re: Philips QFU1.1 Kernel Panic

I talked to Beeldbuisje on Circuitsonline.net and asked to check the reference voltages of 750mV for the DRAM's. He came to the conclusion that one of the resistor chips was broken. Replaced and tv was repaired.