Netgear R7800 exploration (IPQ8065, QCA9984)

And the physical characteristics of the NAND chips have not changed since they were installed. It still knows exactly the same as it did then...

Like I said, I think you may be right about the missing drivers, but u-boot knows nothing about those. It has enough knowledge about handling NAND chips to read and validate the image and if ok, loads it to "Load address", passes the control to the kernel at Entry point (or something like that...). And then Linux kernel starts and uses dts and needs the correct drivers etc.

It's not only about drivers, it's also about partition layout. But you msy be correct indeed. Let's just see bootlog with my changes.

Wonder if it may have such side effects
https://git.lede-project.org/?p=source.git;a=commit;h=7d00cfe9bb693e376ac9d035e13f8ce8a5ff572c
Because as it seems it can't find rootfs.

That is a good find. Might explain why the structure fails from u-boot's perspective.

But why would it still work for 4.4 images?
Some peculiar rounding error with the 4.9 size just on the wrong side of a boundary?
(there is a change in ipq806x padding logic.)

EDIT:
original discussion about that change is in
https://github.com/lede-project/source/pull/745

Fix found and implemented: https://git.lede-project.org/?p=source.git;a=commit;h=e01b034cdc64123baaca8d15f15b2b90e8a39db9

Does not help much. I tested that a few hours ago.
(It fixes just the u-boot part, but not anything inside Linux itself :frowning: )

Bummer :frowning:

I was able to get some more infos using earlyprintk ... http://snap.reifferscheid.org/linux-4.9-bootlog.txt
any suggestions anyone?

Nice!! Will check it out.

Is it with my build or trunk? My build has cpufreq/clock/mdio DT drivers included

https://github.com/dissent1/r7800/tree/bl49-11

http://snap.reifferscheid.org/linux-4.9-bootlog.txt was built with lede-trunk

A build with dissent1/bl49-11:
http://snap.reifferscheid.org/linux-4.9-bootlog-dissent1-bl49-11.txt
http://snap.reifferscheid.org/lede-dissent1-bl49-11-config
http://snap.reifferscheid.org/lede-dissent1-bl49-11-kernel-config

Seems some driver crashes even earlier.

I've rebased my tree on top of current trunk https://github.com/dissent1/r7800/tree/bl49-12

Could you please test run this branch with every commit starting from

If you want I can premake images for you

That would be the most easy thing for me. Please just send the initramfs images to my private email address and/or contact me on IRC. Thank you.

@blogic here's the scm patch for testing


apq8064 used as workaround for now

Btw the issue seem to be with incorrect memory mapping leading to paging failure.
We've tried a couple of variants on IRC, none seem to solve the issue.

I've figure what commit breaks booting even more, it's the one with renaming and adding dts files

Just in case it helps anyone else, here is some information on the switch port setup (which I have put as comments in my /etc/config/network so I don't have to work it out again when I change something):

# All ports are connected to one switch (switch0)
# Switch ports:
# Port 0: eth0
# Port 1: port labelled 4
# Port 2: port labelled 3
# Port 3: port labelled 2
# Port 4: port labelled 1
# Port 5: port labelled Internet
# Port 6: eth1

There are reports on the web of problems with VLANs on the R7800 (with the Netgear software). At the moment I have only tested using VLAN support to define additional sub-interfaces for ports. Has anyone else tried things like tagging externally?

I use the R7800 to connect to a FTTH connection.
There the internet, IPTV, and VOIP comes in on different vlans.
Using tags to detangle that works perfectly fine

1 Like

Same here, using R7800 for FTTH with 2 VLAN on WAN (port5) and 4 in LAN (port1), working perfectly fine !

A small thin i see is that in Luci the port numbering is mirrored vs the real numbers on the back.
so if i plug a device in port 1, Luci shows me port 4 going online.

Do you guys have the same experience?

Yes. See my list of switch port mappings to port labels above.

ah actually missed to understand that part of your post. so you see the same mirrorring. ah well, minor issue

I can reproduce a problem where multicast packets are not received over LAN (wired) interfaces (but work fine over WiFi). See https://bugs.lede-project.org/index.php?do=details&task_id=580 for details.

I would be interested to hear if anyone has a different experience.

Hmm, well i have the r7800 in use with FTTH at 2 family members, and both use them to als watch IPTV. The IPTV streams are multicast based, and working fine. Ofcourse that is only wan-> lan and not lan->lan.

What build / kernel are you on?
These boxes don't run the latest code but R3196, kernel 4.4.47