<rsalvaterra>
Or it should have been cc'd to stable, actually.
shibboleth has joined #openwrt-devel
Tapper has quit [Ping timeout: 480 seconds]
f00b4r0 has joined #openwrt-devel
Tapper has joined #openwrt-devel
arthur_melo has joined #openwrt-devel
minimal has joined #openwrt-devel
<hanetzer>
hrm. is it possible to use tplink-safeloader without the partition-table/etc? the vendor factory image *looks* similar but its missing those.
<ynezz>
ok, so to make it more clear, it would be nice to have `CONFIG_TARGET=platform|target|subtarget|env` ?
<pepe2k>
that was my initial idea, but you preferred to fix the wiki instead? I'm OK with both tbh
<pepe2k>
that "platform" thing there comes from the old build code I think
<ynezz>
ok, makes sense
<ynezz>
BTW wondering about following cargo cult for kernel modules: AUTOLOAD:=$(call AutoLoad,99,selftests)
<ynezz>
I mean, shouldn't it be just fine to leave available in the filesystem and let kernel decide when to load it?
<ynezz>
if it's not needed, then it's probably just wasting memory etc. right?
<ynezz>
ah, well, it's probably due to missing depmod
<pepe2k>
ynezz: regarding the UX of CONFIG_TARGET, if we decide to make some changes (e.g. add target==platform) then also the 'subtarget_platform' should be extended to support also 'subtarget_target' keyword... probably dropping 'platform' is not a good idea because of backward compatibility reasons?
<pepe2k>
on the other hand, sounds about right time to 'fix' that naming convention there (platform vs. target)? not that sure really
<schmars[m]>
RTX2080 should be finished cracking by now? :D
<Namidairo>
someone already ran it past a big dictionary
<Namidairo>
presumably it's some terrible leetspeek variation of a tp-link related phrase that you'd have to generate yourself
<hanetzer>
yeh.
<hanetzer>
stintel: hey. about the EAP615-Wall. You patched the tplink-safeloader.c for it, and said `.first_sysupgrade_partition = "os-image",`, but I don't see an "os-image" in your partition table?
valku has joined #openwrt-devel
<hanetzer>
hrm... slh seems the tree you pointed me at is having compile difficulties for firmware-utils
<neggles>
blergh yeah not a known hash
<neggles>
well
<hanetzer>
actually nvm.
<Slimey>
heh whats this we brute forcen'?
<Slimey>
i had to do the same against adtran bluesocket root account
<hanetzer>
Slimey: a few tp-link eap* devices seems to have that above zZDeY hash
<neggles>
the $$ is weird
<neggles>
but meh
<hanetzer>
the $'s are delimiters of a sort.
<hanetzer>
$type$salt$hash
<neggles>
yeah
<hanetzer>
$type$$hash << no salt
<neggles>
but mkpasswd et al. don't like doing empty salt
<Namidairo>
it's unsalted so you could probably just feed in the converted md5
<Namidairo>
unless crypt did things like "we ran it through 10000 rounds" or similar...
<Namidairo>
ah, it's 1000 iterations.
<hanetzer>
hrm.... yeah. slightly different tplink-safeloader format ehre.
<Slimey>
i think i ran the adtr pass thur rockyou rule with word dictionary it was that simple of a pass
<hanetzer>
heh
<Slimey>
because who doesnt like bandwidth
<Slimey>
¬_¬
<hanetzer>
so normally, tplink-safeloader is size:md5:filler:fwup-ptn:partition-table:SupportList:firmare/fit-image; this image appears to be size:md5(same salt?):some-stubbed-out-thing-with-just'support-list and soft-version':ubi:SupportList
<hanetzer>
(the ubi contains the kernel and rootfs)
<neggles>
aaaaaaand let me just go wake up the third one
<hanetzer>
more or less the same image alignments and theoretically the same algos. this'll be fun.
<hanetzer>
on that note. is there a quick'n'dirty way I can run the md5 algo for the tplink-safeloader to check it?
<Slimey>
heh nice
<Slimey>
too bad i cant use distributed hashcat on lab classrooms here :(
<hanetzer>
_NSAKEY
<neggles>
i'm impressed with how zippy this laptop A3000 is
<neggles>
doesn't quite keep up with the 2080 but it's not as far behind as it ought to be
<hanetzer>
someone try shoadmin against it.
<hanetzer>
ugh. I wish opencl wasn't such dix to get runnin :P
<neggles>
lol yeah
<neggles>
i dunno how to test md5crypt() without a salt
<neggles>
everything wants a salt
<neggles>
i guess i could throw it a one-word wordlist
<hanetzer>
yeh. did dat. no joy tho
<neggles>
its possible that it's salted and they just deleted the salt
<neggles>
making it effectively disabled
<neggles>
but that'd be mean and they're not generally that clever
<hanetzer>
ahhh, I know what I'll do. I'll just bodge src/md5.{c,h} into a standalone executable :P
<robimarko>
Nah, TP-Link aint that clever
<hanetzer>
robimarko: oh hey, it you. I have an ipq807x device :)
<neggles>
hanetzer: it's not quite that simple
<neggles>
crypt() uses base64 with a custom character set for output encoding >:(
<robimarko>
hanetzer: Which one?
<neggles>
he done got an EAP660 HD
<hanetzer>
yee
<robimarko>
Oh, that one is nice
<hanetzer>
neggles: I meant for checking the firmware out, not for the password thing :)
<neggles>
ohhhh
<hanetzer>
robimarko: would be nicer if they didn't leave out three jumper resistors on tx/rx/pwr of the presumptive uart tho :P
<robimarko>
If it wasnt TP-Link I would be surprised
<neggles>
did they just leave out the 0R or are they 1.8V IO?
<robimarko>
But they have been pulling that for a decade
<robimarko>
IPQ807x is 1.8V
<hanetzer>
anyone know offhand which binary actually flashes the the vendor firmware on tp links?
<robimarko>
So make sure your UART adapter is 1.8V, otherwise its gonna be costly
<hanetzer>
robimarko: k, with that said, what resistors should I use? :P
<robimarko>
0 Ohm should be fine
<neggles>
0-ohm and a 1.8V CP2102/FTDI/etc
<robimarko>
Or just bridge it with solder
<neggles>
all these blasted FSM99xx chips are 1.8V too grr
<hanetzer>
I'm using a blackpill running blackmagic firmware tbh.
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
<hanetzer>
my ttl died recently :P
<neggles>
hanetzer: that will kill it.
<robimarko>
For me the knockoff FT232R are best
<robimarko>
Been using them for years
<neggles>
those are 3.3V
<hanetzer>
neggles: figured.
<neggles>
i've become a CP2102N guy lately
<neggles>
cause you can get 'em for about three bucks
<hanetzer>
anywho, focus. which binary flashes updates? :P
<neggles>
there's a couple of good 5/3.3/2.5/1.8V FTDI breakouts
<hanetzer>
I'll grab some.
<neggles>
aliexpress <3
<neggles>
hanetzer: it's probably entirely dependent on the specific platform
<neggles>
but if cliclientd stopcs works that should kill the signature check like with eap615
<hanetzer>
well, it 'works' in that the command exists and it returns 0 :P
<neggles>
robimarko: if you're wondering what an FSM9900 is, "LTE femtocell chip aka an APQ8064 with a cubic assload of HEXAGON"
<robimarko>
It seems that everybody has got LTE cells these days
<neggles>
*AND THEY ARE ALL QUALCOMM FSM9900*
<neggles>
I have disassembled no fewer than *eight* different models and if FCC internals are anything to go by
<neggles>
it's FSM9910, FSM9916, FSM9900, and the occasional "actually we used an Analog Devices frontend and a Zynq UltraScale"
<neggles>
also they basically run android kernels.
<neggles>
and expose ADB. so that's neat.
<neggles>
5G ones are FSM10k and now I am going the heck to bed before I rant about LTE in the wifi router software development channel again heh.
<neggles>
(my fav thus far has been the sprint magicbox 544/587, both of which contain three independent SoCs for really dumb reasons that boil down to "Qualcomm Sucks". ok night y'all)
goliath has quit [Quit: SIGSEGV]
<hanetzer>
robimarko: so, your openwrt tree/ipq branch. you taking prs for that? :P
<robimarko>
Yeah, off course
<robimarko>
Most boards there were contributed by users via PR
<hanetzer>
so apparently the runtime dts differs significantly from the on-disk one :P
<robimarko>
Thats bootloaders fault
<hanetzer>
for instance, the nand partitions are not listed unless its a live system.
<robimarko>
They are patching it via the bootloader
<robimarko>
You can use the SMEM parser instead
<hanetzer>
splain :)
<robimarko>
Those partitions are just parsed out of SMEM by the bootloader
<robimarko>
And then it will patch those in
<robimarko>
And in a really nasty way if there is a partitions subnode
<robimarko>
You can just use the in-kernel SMEM parser to populate the partitions
<hanetzer>
ah, so don't bother putting them in the dts manually?
<robimarko>
Yeah
<robimarko>
I mean, you could used fixed-partitions
<robimarko>
But then you have to use the trick I used in AX9000 with the fake disabled partitions node
<hanetzer>
tbqh i'm a bit miffed. quite a lot of the tplink-safeloader using images specify 'os-image' as the first partition but don't evne have it listed in their list of partitions
<robimarko>
For anything IPQ807x related, jump in the AX3600 thread or make a new one in the forum
<hanetzer>
well, its not ipq per se. its just a safeloader image that's not quite the same as others :P
<robimarko>
I meant in general regarding that device
<robimarko>
Anybody else having issue with building the gpio-button-hotplug?
cbeznea has quit [Quit: Leaving.]
cbeznea has joined #openwrt-devel
shibboleth has quit [Quit: shibboleth]
guidosarducci has quit [Remote host closed the connection]
guidosarducci has joined #openwrt-devel
<ynezz>
with 5.15?
<ynezz>
robimarko: ^
<robimarko>
Yeah
<robimarko>
It fails on the No rule to make target
<robimarko>
Nothing more usefull than that
<ynezz>
it works fine here with armvirt/64 and CONFIG_ALL_KMOD=y
<ynezz>
imx is fine as well, sunxi too
<robimarko>
Lets try cleaning again
<rsalvaterra>
robimarko: Building fine here too in both targets I've tested (ramips/mt7621 and mvebu/cortexa9)…
<robimarko>
OK, its gotta be a dirty buildroot issue(Though I distcleaned today)
<rsalvaterra>
It's possible. The weirdest errors I've seen happened when musl was bumped.
<robimarko>
Yeah, that is why I distcleaned after rebasing since musl update was there
<robimarko>
But I know why it failed
<robimarko>
I ran make clean
<robimarko>
And then tried building mac80211
<robimarko>
But the kernel wasnt built
<rsalvaterra>
Oh, chicken and egg… :)
<Tapper>
Hi people. Can any one test that they can make a new env for me pleas? I did a distclean and ./scripts/feeds update and then install are working for me, but ./scripts/env will not work.
<Tapper>
tapper@TappersPC:~/openwrt$ ./scripts/env new r7800
<Tapper>
error: unknown switch `b'
<Tapper>
./scripts/env: Failed to initialize the environment directory
<Tapper>
I am using Ubuntu on WSL2 windows 11.
<Tapper>
If some one can let me know that it is working for them then I will know that my wsl install is fucked!
pepe2k has quit [Remote host closed the connection]
<hanetzer>
what even is scripts/env ?
<karlp>
using git to magically have multiple config trees in the same build root
sorinello has quit [Ping timeout: 480 seconds]
<hanetzer>
ah
indy has quit [Ping timeout: 480 seconds]
lelux_ has joined #openwrt-devel
ekathva has joined #openwrt-devel
lelux has quit [Ping timeout: 480 seconds]
<mrkiko>
robimarko: hi, can you help me out with the gl-b2200 DSA conversion?
<robimarko>
Not right now
<mrkiko>
robimarko: np
<robimarko>
Send me a PM on the forum or something, will reply when I have time
<mrkiko>
robimarko: is IRC ok for you?
<robimarko>
Better forum PM
<mrkiko>
robimarko: :(
<robimarko>
Cause, I will forget the IRC PM for sure
<mrkiko>
robimarko: can understand, maybe I'll send you an e-mail directly
<robimarko>
That works as well
<robimarko>
Even better
<mrkiko>
robimarko: sahrtura address, thanks
<robimarko>
Better on my private email: robimarko@gmail.com
<mrkiko>
fine, thanks!!
<hanetzer>
question, how can one definitively tell the difference between ipq807[0-9] devices?
<svanheule>
hanetzer: I was the one who figured out "cliclientd stopcs" exists. Cost me quite a few nights of staring a decompiled code >_>
<hanetzer>
svanheule: many thanks, then. what exactly does it achieve? still need the tplink-safeloader header?
<svanheule>
hanetzer: firmware layouts in tplink-safeloader usually concat "os-image" and "file-system" into one "firmware" partition
<svanheule>
hanetzer: the image generator then just takes the first 64kB of the unified firmware partition, and calls that the "os-image" partition, the rest is put in "file-system" IIRC
<svanheule>
and since flash layout is contiguous, that results in a working device :)
danitool has joined #openwrt-devel
<svanheule>
hanetzer: the stopcs disables the signature check, but there were other checks on total file size. So something with the size of a signature still needed to be present
<hanetzer>
ahhh
<svanheule>
I was actually worried TP-Link would insta-plug that stopcs hole allowing unsigned FW-s to be uploaded; but it seems they don't care
<hanetzer>
so, this device is a 'dual boot' with ubi :P
<hanetzer>
"Them fuckin' nerds will just figure something else out" or something :)
<hanetzer>
but yeah. this device has something akin to a tplink-safeloader header, but its split into a header and a footer, and the header part is a bit different :P
<svanheule>
you should have a look at the commit messages for the EAP245v1/EAP225v1. Those are older FW-s where stopcs didnt' exist yet
<svanheule>
are there already other NAND devices supported by tplink-safeloader?
<hanetzer>
not that I've found yet.
<hanetzer>
but yeah. its 0x1814 header, ubi(kernel+rootfs), and a footer
<svanheule>
I've been following the EAP6xx device releases, but so far they've all turned out to be ipqXXXX :(
<svanheule>
except for the EAP615-Wall, of course
<hanetzer>
yee. this one is ipq807x
<svanheule>
I think there's also ipq60xx devices; the SoC (family) name is the filename of the available GPL archives
<hanetzer>
yeh
<hanetzer>
the eap660 gpl archive is incomplete. where linux-4.4 should be, it just has a broken symlink lmao
<svanheule>
there is *an* archive, that's already a step up from some other vendors :P
<hanetzer>
but yeah. what one would think of as the partition list partition is ... nonstandard, and the 'SupportList:\r\nFoobar' bits are at the end of the firmware :P
srslypascal has quit [Quit: Leaving]
rua has joined #openwrt-devel
<svanheule>
the safeloader format doesn't require any particular order to the included partitions
Borromini has joined #openwrt-devel
<svanheule>
hanetzer: oh, great fun, they changed the safeloader partition table format :(
<hanetzer>
svanheule: yeh.
<hanetzer>
got it up in kaitai right now.
<svanheule>
hanetzer: I think the table format is: name [0x10 bytes], offset (starting from 0x1814) [UINT32_BE], size [UINT32_BE], partition type (?) [UINT32_BE] for entries. It starts with the UBI payload size, "0x00000002" (partition count?), "0x00000000" (delimiter?)
rua has quit [Ping timeout: 480 seconds]
<hanetzer>
that's what I was getting at, yeah
<stintel>
hanetzer: check the reviews on the ML, I believe there's something about that
<hanetzer>
stintel: hmm? remind me what you're replying to :P
<stintel>
eap615 partition
<stintel>
I'm in the middle of nowhere with sporadic access to irc :P
<hanetzer>
svanheule: see, I'm not sure about your hypothetical definition of the new part table, because 0xe80000 is exactly the length between UBI# and SupportList:
<hanetzer>
wait a minute.
T-Bone has joined #openwrt-devel
f00b4r0 has quit [Ping timeout: 480 seconds]
goliath has joined #openwrt-devel
shibboleth has joined #openwrt-devel
Borromini has quit [Quit: leaving]
pepe2k has quit [Remote host closed the connection]
xes_ has quit [Quit: bye..]
xes has joined #openwrt-devel
robimarko has quit [Quit: Leaving]
<russell-->
if i modify network.wan.dns with uci on a temporary basis (not committed) what do i reload so that the new value is respected? after a few minutes my /tmp/resolv.conf.d/resolv.conf.auto is reverting to the commited values
<Habbie>
russell--, if i recall correctly, .auto is written by the dhcp client in most setups
<jow>
russell--: ifup wan should do the trick
<russell-->
that's tricky, because i'm remote and i don't want to lose access
<jow>
russell--: you might want to set network.wan.peerdns=0 too
<russell-->
jow: that's set already
<jow>
russell--: in that case: (ifup wan; sleep 60; reboot) &
<russell-->
is ifup wan idempotent?
<jow>
yes
<russell-->
ah, cool
<jow>
at least it is supposed to, quality might vary slightly depending on the involved proto handler
shibboleth has quit [Quit: shibboleth]
<jow>
in any case, it might make sense to start a "reboot -d 60 &" before working on the wan config
<jow>
if stuff continues working after "ifup wan", do a killall reboot
<russell-->
i did some testing locally, it seems to work
<russell-->
fwiw, i'm diagnosing a dns poisoning
<jow>
hmm
<russell-->
so i'm pushing resolution through a management vpn
<jow>
as ins dnsmasq serving wrong replies?
<russell-->
but to set up the vpn, i need regular dns, lol
<russell-->
something upstream is feeding me 0.0.0.0 replies to things like time.apple.com
<russell-->
it's local to the particular network
<russell-->
that is, same isp elsewhere works fine
<minimal>
russell: a pi-hole (or similar) I'd guess
Misanthropos has joined #openwrt-devel
<minimal>
I've done that myself for time.apple.com in my local network
<russell-->
the replies appear to come from 8.8.8.8 in tcpdump
<russell-->
gateway.fe.apple-dns.net is another victim
<russell-->
jow: ifup wan worked fine, thank you!
arthur_melo has quit []
<hanetzer>
yeah there's something fucky going on here.
rua has quit [Ping timeout: 480 seconds]
<hanetzer>
think I got it...
<russell-->
minimal: does pi-hole forge dns?
Pepes has joined #openwrt-devel
rua has joined #openwrt-devel
<hanetzer>
svanheule: yeah I misread what you meant. so basically there's a 12-byte description of the actual ubi firmware, then two 44 byte support-list and soft-version 'partition entries' which come $fw_size and ($fw_size + $support-list-size) after the start of the firmware?
<dwfreed>
russell--: trace the mac address? assuming you have access to the network devices comprising the LAN
<Pepes>
Guys, I would like to ask as I was told that next month in Hungary there is going to be Free Software Conference, where is going to be OpenWrt community and even some OpenWrt devs. What is the truth about this thing?
<Pepes>
(my fault, I posted this first to #openwrt)
<minimal>
russell: pi-hole is used to "blackhole" some dns names to point instead to something like "127.0.0.1" or "0.0.0.0" (not sure which it uses)