There are 2 ways to fix it. As switch still working but just disable only one 5ghz wl.
1. Copy and install eeprom on Fm ic (behind on the left of PEX8603).
2. implent code into FW.
Each router have specific code for switch to 2nd 5ghz wl. If then default to first 5ghz and disable the 2nd one.
here the software for dev:https://www.broadcom.com/products/pcie-switches-bridges/software-dev-kit
Hi
I have the same MP2 router and it turns on properly only on software R8500-V1.0.0.28_1.0.15 but does not work wifi [neither 2 nor 5 GHz]
I think it is a hardware problem. Perhaps only PEX8603 is damaged
This system is on BGA.
I would have to have copies of a flash router that works S34ML01G200 1Gb / 128MB
I have a programmer to upload.
I checked the bone and there are no bad blocks. She is not a problem. Maybe someone in the forum can do a flash copy of the router?
I will try to buy such a reliable router and flash flash [with substitution mtd4]
I think, however, that it would not do anything [the problem lies in some layout on the PCB]
Sorry for my bad English.
He also noticed that in the copy of the router [128MB] no data mtd7: 002c0000 00020000 "T_Meter1"
mtd8: 002c0000,00020000 "T_Meter2"
Maybe someone knows what it is?
Normal BCM 47xx Soc haves 3 pci-e, one for 2.4 ghz then another one for 5ghz. Because cut cost and efficiency , it used only 1 sdram at 32 or 64MB x 16 bit x 8 banks; then you see, arm cortex cpu is 32 bit, that means 32 i/o data line. But router only used 1 sdram ic with 16 i/o data line.(a mystery in computer x86). that is the bottom neck. So a 5ghz wl chip can transmit max 5GT. If the router has another one 5ghz then data rate will be spit into 2.5GT. In order to increase and maintain max data rate transmit then a pci-e switch need to be put in. As a doc say: 1 5ghz wl still get 5GT and the other is 2.5 GT which total 7.5GT(ASM1182e = 1x PCIe Gen.2 >>> 2x PCIe Gen.2 (packet switch) . also this concept cutting edge of technology is very well useful in future. But heat issues alway is a problem until interact of gravity unlock.
back to topic, we need to confirm that PEX 8603 is only disable or lock up. just copy whole img of working and original router then clone to bad one. if it work then the pci-e switch is disable. also ea9500 have same hardware , then you may be compare fw of 2 and find down what happen. that is easy way. Running software PEX8603 for debug may be last choice if you are a engineering tech.
So, I would be willing to try to write entire image to flash from a known good one; however, I do not have a known good unit that functions with stock firmware to get a source full image from. I did take the mtd partitions (less mtd1 and mtd4) and write each of those to the 16 other partitions (that was posted above) on the unit I have here. So, mtd1 is nvram of course, and mtd4 is boarddata...and I do have a good clone of my mtd4 of my unit. So short of nvram variables which were cleared, that is all pretty close to a full image restore that I could do with what I have. I anyone is willing to share a full image dump from a fully working unit, I am happy to try.
So, from the logs at boot, it is hard to say specifically that the pci-e switch is the cause. It seems like one of the bcm4366 chips is not recognized due to the mis-id of the chips Broadcom but perhaps it is in addition to the pci-e switch.
Not sure that ea9500 is same exact hardware - main stuff yes - but it has less ram for example. I am not sure how the PEX8603 would be disabled if one radio works for many folks?
There are 2 ways to fix it. As switch still working but just disable only one 5ghz wl.
1. Copy and install eeprom on Fm ic (behind on the left of PEX8603).
2. implent code into FW.
Each router have specific code for switch to 2nd 5ghz wl. If then default to first 5ghz and disable the 2nd one.
here the software for dev:https://www.broadcom.com/products/pcie-switches-bridges/software-dev-kit
Is it possible to determine the configuration for that eeprom from the PEX doc links you posted? Of course I am making the assumption that the contents are merely config/register settings needed to properly put the PEX into the proper state and not any type of custom firmware code. Is that about right?
Take a read through this if you have not already seen the thread. Sounds very much like what may be occurring here on some variants (i.e. MP2) of the R8300/R8500 Vietnams units:
It discusses the R8000; however the R8300/R8500 use the same PEX 8603 chip.
"The Netgear R8000 has a PEX8603 connected to the BCM53012 and if
it isn't configured during the bus scan the PCI layer goes crazy
trying to configure phantom devices."
Now looking at the some of the code (still reviewing), depending on the model, the proper setup of the downstream WL chips may not have occurred at time of PCI scan resulting in spazzing out and failure to properly register a particular WL chip/radio.
Furthermore, I actually witnessed this last night (me testing a theory), that improperly configuring one of the WL chips, can (and did last night) cause an inadvertent write to one or more areas on the nand chip and starting tossing out all kinds of corruption. In turn that produced a boot loop. I think this may ultimately be what be are seeing.