Error Codes

[sbmotoracer]

Wanted to know if anyone knows where the error codes are stored? I'm thinking that if we find out where those codes are stored we might be able to fill in the blanks for the rest of the error codes. I'm thinking they might be in somewhere in the kernel but im not 100% sure about that. Would the kernel be up if lets say the GPU overheats or if not then maybe the SMC as from what i read it does do some temperature monitoring. If someone who has a copy of a kernel and has reversed it enough to find them and understand what they mean then if they could please add it to the existing error code list that we have.

[yivkX360]

We alreadey know almost all the error codes. And the ones we dont know are very rare.

[sbmotoracer]

True, but it dosnt hurt to have all of them.

[Oneohm]

No one out there truly knows the error codes. The guy on xbox scene "thinks" he knows the cause of them but a good portion of them are worng or vaguely jump around actual causes. It would be nice to get a legit cause effect list but many causes relate to the obvious. Hard Drive failures, GPU solder spheres not adhering/wetting, DVD drive I/O errors.

[tmbinc]

The published lists of error codes are *crap*, *crap*, *crap*. Seriously, they are all wrong, and not much more than a good guess "how, let's see - my console is too hot, i got xyzx - so xyzx means "overheat."".

I've done some reverse engineering of the SMC code, plus some board experiments, so here is my attempt at producing a "slightly better" error list, free for discussion, not implying that all are right:

Found by board testing / fuzzing / reparing:
0001 12V from power supply missing,
0002 CPU powergood deasserted
0003 GPU powergood deasserted

Found by SMC disassembly:
0011 CPU thermal overload
0012 GPU thermal overload
0013 edram thermal overload

found by 2BL disassembly:
0020..0023: some things in the initialization sequence didn't end up as expected. for example, if pci express didn't work (line cut, ...), you get this.
0022: this one comes when the CPU doesn't talk with SMC in some time period, but also doesn't indicate an error.

0030: temperature control didn't work for some reason. I can trigger this by messing around with the fan state via SMC commands.

0100..0111: they are all (some of them are unused) caused by the virtual machine code in the bootloader which initializes the GPU and size the memory.

0200+ are software generated errors, from kernel, and usually come up with a "fatal error code"

[Oneohm]

Just to add my 2 cents;

I repaired several code 0020 by reflowing the GPU only. Removed a GPU and found some traces were pulled up on the GPU.

Reproduced a 0021 by pulling open the PCI clock signal to the Southbridge. I also was able to correct many boards with a 21 by reflowing the southbridge. I believe the intention of this code was supposed to be a general PCI bus I/O failure.

Reproduced a Code 0022 by removing the Clock signal to the CPU/GPU After several seconds the code is produced. My theory is the code was intended for an I/O failure between the two.

Code 0110 tends to occurr if the GPU has open connections to the memory.

E74/1022 seems to be caused by an open line to the ANA chip from the GPU. Reflowing the GPU usually solves this but in some occasions it produced a code 0022.


I believe Robinsod reproduced an E79 reflashing the NAND without zeroing.

I read somewhere that the codes were linked to the serial eeprom on the WiFi/Power button Board but I never dumped the chip to see. I always thought it was storing controller ID's.