To anyone following this:
I am running Linux 4.4.88 for this test.
There are reports that Linux 4.4.111 has network detection problems.
Skip to Linux 4.4.113 at least.
My latest report is that Linux 4.4.115 solves the problem on Slackware.

Owner has come back, have to go.
Got 4.4.115
Did not find a didentd slack package with -2, and I dont know where that came from.
It might be why I am missing the libcrypto.so.0.
May have to reinstall opensll.

Tried testing the DNS system.
I can get DNS for all the sites, both IPv4 and IPv6.

I have verified that NetworkManager is setting up the DNS with the proper DNS sites.
I verified that the cable modem DOES NOT have the utopia DNS exploit.


I can access big sites like google, youtube.
It seems that I can access sites that have IPv6 address.
I cannot access reddit.com, linuxquestions.org, sourceforge.net.
I can get DNS entries for them (dig, nslookup) and they have IPv4 addresses.
Oddly, slackware.com has only an IPv4 address, yet I can access slackware.com
(but I cannot access the slackware store).

routetrace fails for the same sites, usually with "network inaccessible".
I feel am going to have to reverse engineer something to find out exactly what service is determining that and from what settings. It will probably be a kernel service.

I connected a router to my eth0 cable, just to see the blinking lights. If I access that router, or ping one of those addresses, I can see the router lights blink. They DO NOT blink for any of the IPv4 addressed sites like linuxquestions.org, so I do not think that those sites are getting misrouted to eth0. Access to the eth0 router uses IPv4, and that seems to still work on eth0.

NetworkManager keeps setting up route as empty (it just leaves 2 entries, one of which is loopback). I don't know where to look to see what kind of routing it is actually doing.

There are a number of /etc files like hosts.allow and hosts.deny that control host access.
I think that those two only control access from outside to services provided by my machine. Reading the man pages does not make it any clearer. But, I don't think that those particular files are the problem. I tried commenting out hosts.deny, just to test it, but it had no effect.
I do wonder what other /etc files need to be checked over.

I suppose there could be a Linux firewall running and it is just running with random settings. I never needed the firewall before, so I have not done any firewall setup on this Linux box. There has not been any sign of a Linux firewall either, not in dmesg, systemlog, nor ps -A (assuming that it would be easily recognizable).

will show you the routing from your Slackware box.

if /etc/rc.d/rc.firewall is executable, then there should be some type of firewall running.

Who is the ISP? Once your packets hit their network, it's out of your hands.

Thank you Alien, I did eventually find the slackbuild for the patches. I almost always go directly to the patches download page. I had to back up a directory from where I was looking, to find the patch source directory.

It is an Xfinity cable box, a cisco DPC.

The XP on the same cable router can see linuxquestion.org. That is where I am at now (until I get harrassed off, at least).

When the Linux box is connected by LAN cable to the router box (the long LAN cable up the stairs), the linux box cannot see the XP box that is on another LAN port of the same router. I even tried giving the XP box a static address.
The Linux box cannot access the router admin page using the IPv4 router address.
I can access this admin page from the XP box.

Even if the cable router does not like Linux, the local network through the cable router box (LAN cables) should allow me to ping the XP box or see its open ports. It should have a transparent local network for it's 4 LAN ports, shouldn't it ??

I can ping with IPv4 the router on the eth0 cable and access that router's admin page.

I will try
> route -n

When I tried
> route
the only thing I could see was the two local settings that NetworkManager left there.
It seems to clear everything else out.

There is only one commonality that I can think of. Every time I want to test something on the cable router, I end up using NetworkManager to connect. This gets an address from the cable box via dhcp. I keep end up wondering if NetworkManager is so hung up on using IPv6 that it is blocking the IPv4 through its controlled connection somehow.

On Ubuntu: NetworkManager IPv6=automatic breaks IPv4
https://bugs.launchpad.net/ubuntu/+s...er/+bug/284874
This is from 2008, but there are still posts in the forum.
The description does not match up with my problem, but it is the most interesting bug report seen yet. Final post says fixed in 2011.
Note that slackware runs an older version of NetworkManager than the current release.

With so many sites still using IPv4, I would think that most everybody runs dual stacks.
They recommend setting IPv6=ignore in NetworkManager.

What do you mean by "cannot see the XP box"?

It is possible that the router has the IP for the XP box whitelisted for accessing the admin page.

Beats me. It could be configured not to allow that as well as a security measure.

It's probably something similar to One entry for the subnet matching your machine and another entry for everything else. I'll bet that your addresses will be somewhere in the 192.168.x.x space, since that's the more-used private IPv4 network space.

"Cannot see" == As if it was not connected, cannot ping its address, scan cannot see ports.

> route -n
1 entry for 127.0.0.0 lo (loopback)
1 entry for 192.168.... eth0 (my lab network)

> route -A inet6
all the entries from NetworkManager for wlan0
1 entry with a gateway address, ::1/0 ... wlan0

I think that this kind of route setup would not let any IPv4 addresses out on the wifi (wlan0).

All the NetworkManager setups stay hidden when the standard commands are used.
The special IPv6 commands must be used to see the IPv6 setup that the
NetworkManager does.

Turning off IPv6 (as suggested by Alien) did not help. Then I got no network at all.
It may be that there is still something blocking IPv4 setup.

The IPv4 on NetworkManager is set to Auto, the IPv6 is set to Auto.

Trying to setup IPv4 manually has stimulated NetworkManager to actually generate some routing for IPv4. However, it still does not let me access the router using its IPv4 address, nor to access linuxquestions.org (which has an IPv4 address).
It wants me to set up the addresses and gateway manually, and I cannot figure out exactly what it expects me to put into those fields. The documentation goes on and on about what syntatically would be correct, and says nothing about the semantics of what you would want to put into those fields.

Some of the settings I have tried for the manual IPv4 has caused NetworkManager to stop IPv6 traffic too.

I have tried to setup the routing manually too, but I must be getting that wrong somewhere.
I believe the gateway address is the same as the router admin page.
The 10.0.0.1 is the default for cisco router (I am going to change it later).
This has been obfuscated, so don't take it too literally.
In /etc/networks: cableboxnet 10.0.0.0
In /etc/hosts: 10.0.0.1 cablegw
> route add cablegw wlan0
> route default add 0.0.0.0 gw cablegw wlan0
(the above is from memory, so probably not quite right, but the commands were accepted, and the routing showed up)
(Tried several variations, too many to remember them correctly.)

It did no good though. Still could not access the router admin page from firefox.
Could not ping it, could not ping6 it either (it probably has pings blocked, but why would that also block them from the local network side).

The DNS probes succeed.
> dig linuxquestions.org
gives me the address (which is IPv4).

I have tried to watch the network traffic, but have not been able to decipther it.
There seems to be a regular packet exchange between Linux and the cable box, that is seen as packet about every 2 seconds. The wifi dongle led blinks too.
When trying linuxquestions.org, there is consistently a 4 packet exchange. I have not been able to prove that it is an exchange with the cable router as I have not been able to decipher that network traffic display. The constant background traffic ensures that the it scrolls off the display in about 10 seconds, and it takes me at least 2 seconds to get that console back after stimulating it from firefox. There is logging, but that would generate a long listing where I would not know which part was the response to my stimulus.


Trying traceroute for 10.0.0.1 and linuxquestions, gives me "network unavailable" (or some variation of that wording).
If I try it with a host name, I get "host unknown" responses.
It looks to me like IPv4 is being blocked somewhere, dependent upon address ranges.
The IPv4 going to the lab router over the eth0 cable worked, so it cannot be all IPv4 being blocked.
The NetworkManager auto setting of IPv4 might be failing for the same reason, it fails
somewhere in detecting the IPv4 support.

Thank You, Richard Cranium, for the route settings. I will compare that to what I am trying.

I just decided to try NetworkManager over this lab router in various combinations to access one of the lab computers. I will use a simple WiFi router that I understand better.
I will see if I can get that setup, or if I get the same problems. At least I will get some idea if the problem is with the cable router, or in the Linux, or in NetworkManager.

I really want to figure out what is going wrong before I try something like wicd (which I do not have installed yet).

I probably will try wifi-radar again in the lab tests, as I can do that quickly enough if I just disable NetworkManger in /etc/rc.d and reboot. If that works, it will confirm that the Linux support is there, and it is a NetworkManager setup issue.

Also will try totally manual setup. I found some useful information on Ubuntu sites.
I have to give credit to Ubuntu for providing useful help information. I requires some translation into slackware appropriate commands though.

If the windows box you've been using is using wireless to connect and if your ISP supports IPv6 (Xfinity appears to), then the wireless setup may not allow IPv4 at all. There wouldn't be much of a reason to allow IPv4 in that case.

You may want to attempt to enable IPv6 on your slackware box and merely use that; see https://notes.benv.junerules.com/tag/ipv6/ as well as asking folks around here for help.

dunno whether this might be at play:

from /etc/dhcpcd.conf, which had to be edited from duid to clientid to get network access at work:
hth

This posting is from the Linux computer over the WiFi, using IPv4.
I have got something that works, and is repeatable. I am still working out the details of exactly what happens, because I intend to post the most useful information.
- I have started making scripts that activate the hardware. This is necessary because having done so many variations, I do not trust anymore typing anything from memory. Once having figured it, the script records the correct command and is repeatably does it the same each time.
A necessity for debugging.
- I made a script that will shutdown and startup the various network manager tools. It now does that reliably and correctly.

The current steps to connect manually.
1. Disable NetworkManager, stopping process and disabling rc.networkmanager.
2. Put all connection details into /etc/wpa_supplicant.conf
2b. Cannot save network connection setups from wpa_gui because of authorization.
2c. Network setups ( in network{ } ) persist and are usable from the wpa_gui.
2d. Slackware network setup doc by Alien said to have scan_ssid=0. I found that you must enable it with scan_ssid=1, or else
it will be a disabled setup that is DISABLED in the wpa_gui, which cannot override this.
3. Make it a fixed network address in inet1.conf, then do DHCP in wpa.
3a. DHCP in inet1.conf is limited to one setup, and if that fails it creates messages, and leaves the hardware less usable.
4. Use wpa_gui to select a network config and connect.
4b. wpa_gui will sometimes start dhcp itself and reconfigure the routing.
5. Be prepared to change the route tables if incomplete, learn how to setup gateway.

Ran out of time, so I had to abbreviate this.
It was not anything in Linux kernel, was not missing packages, just setup problems, too many files with same info, must shut most down.
Have not figured out how to configure NetworkManager yet.