<stintel>
also important to know that the additional % for social security in Bulgaria is calculated on max 3000bgn per month, so evertyhing above that is flat 10%
<stintel>
so you can get a 10k EUR bonus and manage to keep almost 9k of it
<mangix>
Wow. I had no idea.
<mangix>
Haha Isle of Man is 0
<stintel>
it used to be on max 2400bgn but the socialists had to increase that
<stintel>
of course everything might change with the new government
wulfy23 has quit [Remote host closed the connection]
Borromini has joined #openwrt-devel
rsalvaterra has quit [Quit: Leaving]
wulfy23 has joined #openwrt-devel
<wulfy23>
confirm> ethtool -K eth0 rx off as mentioned in upstream bug report tmp resolves
danitool has joined #openwrt-devel
rua has joined #openwrt-devel
Rentong has joined #openwrt-devel
Rentong has quit [Remote host closed the connection]
Rentong has joined #openwrt-devel
hitech95 has joined #openwrt-devel
rua has quit [Quit: Leaving.]
Rentong has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
<hitech95>
ping jow, allow fw3/fw4 to create refletion roules when IPv6 transitioning protocols are used, map, dslite etc. So far I only tested with mapt, since the map interface des not get a ip address the refletion roule is not creted.
<hitech95>
I'll open an issue but before that I waned to comment/discuss about the behaviour
hitech95 has quit [Remote host closed the connection]
CN_SZTL has quit [Remote host closed the connection]
CN_SZTL has joined #openwrt-devel
rua has joined #openwrt-devel
rsalvaterra has joined #openwrt-devel
Rentong has joined #openwrt-devel
Rentong has quit [Remote host closed the connection]
Rentong has joined #openwrt-devel
Rentong has quit [Remote host closed the connection]
Rentong has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
rsalvaterra has quit [Quit: Leaving]
rua has quit [Quit: Leaving.]
rua has joined #openwrt-devel
aiyion_ has quit [Ping timeout: 480 seconds]
brpr has joined #openwrt-devel
<brpr>
PaulFertser, I'm back. The dtsi has some info about Ethernet, but I can't understand it really. Hint please :P
aiyion_ has joined #openwrt-devel
<PaulFertser>
brpr: what dtsi?
zatwai has quit [Quit: ZNC 1.8.2+deb2~bpo10+1 - https://znc.in]
<stintel>
hauke: can I change the dependency of ptp-qoriq to @(TARGET_mpc85xx||TARGET_qoriq) in the same commit as the one adding the qoriq target, or prefer in a follow-up commit ?
zatwai has joined #openwrt-devel
<hauke>
stintel: you can do it in the same commit
<hauke>
did you check if there are more packets depending on TARGET_mpc85xx which should also work with your new target?
jlsalvador has quit [Quit: jlsalvador]
<stintel>
hauke: only crypto-hw-talitos - I don't think this is used in these boards
<brpr>
PaulFertser qca955x_zyxel_nbg6x16
<PaulFertser>
brpr: ok, and why can't you copy eth options from it if they worked for you?
<hauke>
stintel: ok tfine
<hauke>
*fine
<brpr>
PaulFertser, should I just copy it?
<PaulFertser>
brpr: I guess if it worked when you're booting it.
aiyion_ has quit [Remote host closed the connection]
aiyion_ has joined #openwrt-devel
rua has quit [Quit: Leaving.]
<brpr>
PaulFertser, I have included the dtsi and would you know - Ethernet freaking works :P
rua has joined #openwrt-devel
<PaulFertser>
brpr: I suggest to not include dtsi in your case, you should rather copy the sections before upstreaming.
<brpr>
PaulFertser, kinda. All ports are eth0, eth1 is unused
<brpr>
PaulFertser, weird. Even WAN is on eth0 and all the LAN ports :(
<PaulFertser>
brpr: probably that's just how your hardware is
<PaulFertser>
brpr: quite common to use single interface and do the wan/lan split on the switch.
<brpr>
so I should just connect wan and see if it connects?
<PaulFertser>
brpr: no, you should run "swconfig dev switch0 show" and write down a map of what port marking on the device casing corresponds to what port number in the switch.
<PaulFertser>
brpr: by attaching some ethernet device to each port in turn and observing link status changes
<PaulFertser>
in swconfig output
<PaulFertser>
The cpu port is expected to be always attached obviously.
<brpr>
PaulFertser, port0 is CPU right?
<PaulFertser>
brpr: you gotta figure that out
<PaulFertser>
If you disconnect all ethernet cables and see port 0 is still link up then yes.
<brpr>
because I have port 0 that's always UP and port 6 that's always UP
<brpr>
PaulFertser, and obviously MACs aren't working BUT 5GHz MAC works
<PaulFertser>
brpr: sometimes SoC has both ethernet ports connected to switch, that's also common
<brpr>
PaulFertser, alright, moving on. what to do with the dtsi? if I copy it then I get syntax errors
<brpr>
PaulFertser, and do I need to include eth1 if it doesn't do anything?
<PaulFertser>
brpr: copy it properly to avoid syntax errors...
<brpr>
I did lol
<PaulFertser>
brpr: probably it doesn't because you're not configuring it right yet. The vendor firware uses both eth0 and eth1 so should you.
<PaulFertser>
brpr: using a single cpu port is possible too but that can become a bottleneck forwarding lan<->wan
<brpr>
it does transmit like 4k or 40k bytes last time I've checked
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
<stintel>
I tried to manually trigger it by adding the ipsets defined in /etc/config/firewall, I couldn't
<stintel>
but then it's still a kernel bug imo
<stintel>
s/still/also/
<hauke>
yes, but normally the kernel should catch this without a crash
<stintel>
that's why I say it's also a kernel bug
<stintel>
any idea to figure out exactly which rule in /etc/config/firewall triggers this ?
<stintel>
besides going over them one by one ?
<hauke>
I would add some extra printks into the function
<hauke>
into ip_set_sockfn_get
<brpr>
PaulFertser, ../dts/ath79.dtsi:47.22-68.5: ERROR (phandle_references): /ahb/eth@19000000: Reference to non-existent node or label "phy17"
<brpr>
PaulFertser, and also ../dts/ath79.dtsi:70.22-91.5: ERROR (phandle_references): /ahb/eth@1a000000: Reference to non-existent node or label "phy1"
<hauke>
in which line is this: .ip_set_sockfn_get+0xb4/0x380 [ip_set]
<stintel>
hauke: I linked it earlier
<PaulFertser>
brpr: it means what it says, it looks for phy17: defined elsewhere and I guess your file is missing that.
<brpr>
I have just copied that from the ZyXEL dtsi though! :(
<hauke>
stintel: could you add a printks into the ip_set_sockfn_get() function to see what input we get
<hauke>
probably into the error handling of copy_to_user()
brpr has quit [Remote host closed the connection]
brpr has joined #openwrt-devel
<stintel>
hauke: the weird thing is that the code does goto done, which skips the copy_to_user call
<stintel>
hauke: so I'm not sure if the faddr2line output is to be trusted
<stintel>
I'll add some printks
<hauke>
yes sometimes faddr2line is wrong
<stintel>
not sure if that's going to be for today still though
<hauke>
the compiler rearranges the code in the optimization
<stintel>
goddamnit, I'm moving /etc/config/firewall out the way in failsafe then reboot yet it keeps appearing and the bug keeps triggering
<stintel>
fuck it, I'll just wipe the sd
<stintel>
great, it gets worse. copy /etc/config/firewall and /etc/firewall.user from my main router to the 2nd M300, fw3 restart, no problem
<stintel>
reboot the device, bug triggers
<brpr>
PaulFertser, It worked!
<brpr>
PaulFertser, now what about MACs
<PaulFertser>
brpr: first get wan/lan/switch working properly, then proceed to caring about macs.
<brpr>
right, forgot about that entire crap
<brpr>
PaulFertser, how would I go forward with getting all these ports to work?
<PaulFertser>
18:39 < PaulFertser> brpr: by attaching some ethernet device to each port in turn and observing link status changes
<brpr>
I already have that
<brpr>
Port 1 is WAN
<brpr>
Port 3 is LAN2
<brpr>
Port 5 is LAN4
<brpr>
Port 4 is LAN3
<brpr>
Port 2 is LAN1
<PaulFertser>
brpr: and port 0 and port 6?
<brpr>
they're always UP
<PaulFertser>
brpr: which one is eth0
<brpr>
all of them are eth0
<brpr>
eth1 sends 4k bytes and then just does nothing
<brpr>
probably at initialization
<brpr>
PaulFertser, also is it normal that ethernet interfaces go up after a long pause? kmodloader finishes at 15s and eth interfaces go up on 40-ish seconds
<PaulFertser>
brpr: yes, probably normal, takes plenty of time to start all the scripts
<PaulFertser>
brpr: guess either @phy0 or @phy17 isn't correct then, they can't be both eth0.
<brpr>
I have copied them from zyxel, didnt change a thing
<stintel>
hauke: printk shows *op is always 6 or 8 during boot, faddr2line was probably wrong
<PaulFertser>
brpr: but your board isn't that zyxel so
<brpr>
PaulFertser, that would be the problem ;)
<stintel>
ok so I guess I'll be disabling the rules one by one :/
<stintel>
I'll also enable CONFIG_DEBUG_BUGVERBOSE
danitool has joined #openwrt-devel
<hauke>
stintel: it goes into the "if (*op < IP_SET_OP_VERSION)" with 6 or 8
<stintel>
hauke: yes a lot and when it crashes it was also with 6
<stintel>
doesn't help that this is forced ext4 due to squashfs being unusable, and every time the crash occurs the filesystem goes corrupt
aleasto has joined #openwrt-devel
<stintel>
[ 36.560294] usercopy: Kernel memory exposure attempt detected from SLUB object not in SLUB page?! (offset 0, size 8880)!
<stintel>
it appears to be somewhat random. right now it crashed on size 40
<brpr>
PaulFertser, what should I do to fix this?
<PaulFertser>
brpr: I do not really know. I would read how other boards based on same SoC configure ethernet interfaces, what PHYs and parameters they're using. Can't do it for you, too tired.
<brpr>
No worries man. I'll look into it tomorrow.
<brpr>
Thanks for all your help.
<stintel>
oh. it crashes in different parts
<stintel>
yeah I'm just going back to my original conclusion that HARDENED_USERCOPY is broken on ppc64be
Borromini has joined #openwrt-devel
Borromin1 has joined #openwrt-devel
<stintel>
attack of the Borrominis!
Borromini has quit [Ping timeout: 480 seconds]
<stintel>
hauke: I've mailed linux-mm and netfilter mailinglists
Borromin1 is now known as Borromini
brpr has quit [Ping timeout: 480 seconds]
<Borromini>
and the world was whole again ;)
<Borromini>
happens so infrequently had to look up my pw >_>