[tasman_shn]

Foxconn have no intention to fix tis issue, but instead killed Netgear R8500

Same story with D7000 and D7000v2

/*foxconn Han edited start, 10/01/2015
*when R8500 didn't recognize all 3 interface, then do the software reboot*/
int isDhdReady()
{
char ifname[32] = "eth1";
int flags;
unsigned long addr, netmask;
int i = 1;
int ret = 0;
int max = 3;
int flag = 0;
char *max_pt = NULL;

if(acosNvramConfig_match("dhd_check_ignore","1"))
return 0;

if((max_pt = nvram_get("dhd_check_max"))!=NULL)
{
max = atoi(max_pt);
}

printf("\n--------------------isDhdReady()------------------------\n");
//system("ifconfig -a");
for(i=1; i<= max; i++)
{
sprintf(ifname,"eth%d",i);
ret=ifconfig_get(ifname, &flags, &addr, &netmask);
/*ENONET means interface don't have IP address,
*EACCES means interface don't exist
*EADDRNOTAVAIL Cannot assign requested address*/
if (ret != 0 && ret != EADDRNOTAVAIL)
{
printf("%s %d could not found %s ret=0x%X\n",__func__,__LINE__,ifname,ret);
flag ++;
}
else
printf("%s %d found %s ret=0x%X\n",__func__,__LINE__,ifname, ret);
}
printf("\n-------------------isDhdReady flag=%d-----------------------------\n",flag);

if(flag)
{
printf("DHD didn't bring up all the interfaces!\n");
system("reboot");
}
return ret;
}

[deslatha]

Lucky that PEX 8603 doesnt need a driver; here the layout of router.
from http://www.chinabaike.com/t/35933/2016/0320/4502371.html

[deslatha]

Then Asus AC3200 has same architecture that run very well but except 2.4ghz wl.
here the picture of ic with 15 TP.

[deslatha]

Here is a closed look ic with 128 pin out from Asus AC3200.
Solution??? may be r8500 need convert to ac3200. need mod gpio for compatible.

[routerhacker]

Well.. not sure what to think about the "no driver" statement. The PCI chip(s) / bus certainly need to be properly setup and devices (i.e. wireless chips) sitting on a given bus need to be properly setup and initialized. And that is what seems to be the issue at hand. The wifi chips are not properly being setup/initialized/recognized.

I have read elsewhere that the PEX 8603 used in the these units due to it nature has some trouble when wifi goes offline reestablishing proper communications due to a bug in the broadcom drivers (I can't personally verify that however).

It does seen that in reading much about this that Netgear never made proper changes to the bcm5301x_pcie.c "switch" code and instead just declared the router end-of-life to avoid as much replacement obligations/cost as possible. Seems the Openwrt guys found a bit of the issue but have not yet focused on the R8500 units at length but did sort out in EA9500 Linksys units. Cest-la-vie.

[deslatha]

[routerhacker]

True. But for most anyhow, the mtd calibration data is intact and viewable. It just is either a) not properly being sent or b) not properly being recognized or c) both or d) missing something else.

Perhaps some interactive manual initialization steps would be possible?

Way too late here and not thinking clearly More this weekend.

[Ja]

This is a hardware failure. I copied the flash from a working router and the thread did not change

The processor does not search for wifi devices.
Failure of wifi or PEX modules or CPU


the pcb version of mp1 and mp2 is different. A capacitor was added to the input and one 12V relay removed


Maybe there was a problem with the existence of low power and something damaged

[routerhacker]

Ja,

Which pcb version did you replace the flash on? Was is from a same exact variant board? You said it did not search for wifi devices... did you watch the logs under a serial connection? That is almost always where the bootloop issue occurs. The router tries to init the wifi chip and they are not responding in time due to either misconfiguration, failure, or something like that.

You mentioned the capacitor difference between the board revs. Here is a link that talks about that: https://forum.dd-wrt.com/phpBB2/viewtopic.php?t=309602&sid=7ebf9d0eee95444091b3e8592f13d71a

Now it well might be possible that failing (or missing) caps on the board lead to the wifi chips (and related coprocessors) not getting proper voltages and in turn the wifi chips not consistently or properly initializing and resulting in the reboots.

It is probably worth a very close inspection or maybe even complete replacement of caps on the board with new caps and see if that makes a difference fi that is something you can try.

-routerhacker

[Ja]

so it comes from the same variant

I have copies of the mp1 and mp2 versions


This is not a factory defect. Routers started to fall after some time or uploading a newer software



It is not the fault of software for sure. Unfortunately, I do not have parts to solder. Maybe I will order, but shipping from China will last. And do not want to spoil efficient routers.

[routerhacker]

Yeah, my soldering equipment here is on it's last leg and doesn't heat well now. I need to find a replacement. I will see if I can locate some replacement/fresh caps and give that a go here when I get a little time and see what happens.

Failing after a little time (not immediately from factory) does tend to sound like caps, especially if they have a super sensitive board design.

It would be nice if Netgear would just come clean about the issue so anyone choosing to do so might have a change to fix, even if they still don't replace or make good on their warranties for folks.

[Ja]

Only where to buy them cheaply?


They are quite expensive for 3 pieces
for example.
https://www.ebay.pl/itm/1x-BCM4366KMML-BCM43G6KMMLG-BCM436GKMMLG-BCM4366KMML6-BCM4366KMMLG-QFN-IC-Chip/401550673039?hash=item5d7e49048f:g:sZAAAOSw39NbYZu3


63 $ There is a risk that it will not work. Unfortunately, in my country I do not see that they were for sale

[routerhacker]

If you are thinking about replacing 4366 chips one option might be to get another broken router and salvage one working 4366 chip from it and replace it on your router.

I was not thinking of replacing 4366 chip however. I was talking about bad or failing capacitors. It is not uncommon for capacitors to go bad or start to not provide proper voltage/power levels to components. In this case, as their are multiple wifi chips with coprocessors and an extra PEX bridge involved, the power requirements are much different than other routers of the past. It may just be that the problems are related to one or more sketchy capacitors. After all, you have even seen that between board revisions MP1 and MP2 that Netgear added a capacitor which is indication they recognized something was not quite right in the original design. I suspect the problem may just be in capacitors due to heat and power drain. That would be a cheap and easier thing to determine than replacing 4366 chips.

[Ja]

There are too many types of capacitors to check


In my opinion the problem lies in the BCM4366 chip, which for various reasons crashes


Besides, there is a lot of possibilities. It would be best to start with BCM4366 but I do not have anywhere to buy them outside of China


e.g. Asus RT-AC5300 based on the same 4366 modules also has problems with 2 / 5GHz not working radios

[deslatha]