<ynezz>
f00b4r0: so I've cherry picked those new devices to 22.03, but that QCA9561 fix has to wait for upstream review/feedback
svanheule_ has joined #openwrt-devel
cmonroe has joined #openwrt-devel
svanheule_ has quit []
svanheule_ has joined #openwrt-devel
svanheule has quit [Ping timeout: 480 seconds]
svanheule_ is now known as svanheule
cmonroe_ has quit [Ping timeout: 480 seconds]
Grommish has quit [Read error: Connection reset by peer]
Grommish has joined #openwrt-devel
danitool has joined #openwrt-devel
Tapper has joined #openwrt-devel
<f00b4r0>
ynezz: sure. Can I add your Tested-by for the upstream submission?
xback has quit [Remote host closed the connection]
Grommish has quit [Read error: Connection reset by peer]
Grommish has joined #openwrt-devel
goliath has joined #openwrt-devel
robimarko has joined #openwrt-devel
<ynezz>
f00b4r0: yep
<f00b4r0>
ynezz: thanks. I'll send it then. Just to confirm, did you test on XD3200 too?
<ynezz>
yes
<f00b4r0>
ok.
<f00b4r0>
last question: I'll send patches for 21.02 backport too: for the A930 I'm cherry-picking + changing nvmem to mtd-mac-address, is the following the correct way to show this in the commit message:
<f00b4r0>
(cherry-picked from commit a05dcb07241aa83a4416b56201e31b4af8518981)
<f00b4r0>
[switch to mtd-mac-address instead of nvmem-cells]
<ynezz>
yes
<f00b4r0>
thanks
cbeznea has joined #openwrt-devel
rua has quit [Quit: Leaving.]
rsalvaterra_ has joined #openwrt-devel
rsalvaterra is now known as Guest1995
rsalvaterra_ is now known as rsalvaterra
Guest1996 has quit [Ping timeout: 480 seconds]
AtomiclyCursed has quit [Quit: ZNC 1.8.2 - https://znc.in]
AtomiclyCursed has joined #openwrt-devel
<rsalvaterra>
Ok, a long time ago, in my tree, I removed the usb feature from mt7621/target.mk (because the RM2100 doesn't have USB). Now, make oldconfig is prompting me for USB-related stuff. And setting CONFIG_USB_SUPPORT=y. Even after removing /tmp. WTF…! o_O
<rsalvaterra>
And no, my kconfig doesn't have CONFIG_USB enabled.
<rsalvaterra>
I'm seeing weird things.
<rsalvaterra>
Imma pushing the big red button (make distclean).
srslypascal has quit [Ping timeout: 480 seconds]
srslypascal has joined #openwrt-devel
<rsalvaterra>
No use. (WTF)².
<mrkiko>
I would need a 1GB switch that runs OpenWRt or is unmanaged ("plain") but high quality enough to work reliably. If it runs OpenWRt, it would be cool if it had a recovery procedure not involving UART soldering or connection. Any suggestion?
<rsalvaterra>
Commenting the line that adds the usb feature depending on the kconfig "fixes" the problem, so something is wonky in my kconfig, perhaps… hm.
<mrkiko>
or may I use an openwrt router as a switch? :D
dangole has joined #openwrt-devel
<rsalvaterra>
I can't belive I'm going to have to hack include/target.mk…!
srslypascal has quit [Quit: Leaving]
srslypascal has joined #openwrt-devel
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
ekathva has joined #openwrt-devel
<slh64>
mrkiko: the 'safe' bet would be the ZyXEK gs1900 range of switches, sturdy metal case serial is available on an internal pre-populated 4 pin header with 2.54mm spacing. on some switches this header is even accessible from the outside, through the venting holes. but its usable flash size is a bit limited (7.6 MB), on the plus side, there are no different unsupported h/w revisions to account for
<slh64>
check in advance if your model already has OpenWrt support (two models have pending PRs, patches for a third haven't been submitted yet, but are available). midels range from 8 to 48 ports, with- or without PoE, the smaller devices are fanless
ekathva has quit [Remote host closed the connection]
ekathva has joined #openwrt-devel
ekathva has quit []
<slh64>
the D-Link DGS-1210 has more flash (and some models more RAM), but there you do have to solder, remove board and screwed-on heatsink to get to the unpopulated header - and you do have to be very careful to fetch the correct h/w revision
<Habbie>
is the openwrt wiki hardware table also the place to document devices that will never ever run openwrt? i ask because recently -i- would have loved to find exactly that information for a few devices
<slh64>
Habbie: that would be better suited to wikidevi
<slh64>
mrkiko: some of the larger gs1900 models might require soldering, check the wiki for your desired model
<svanheule>
mrkiko: Cisco SG220 models have an RJ45 (Cisco-style) RS232 port, but none of those are officially supported yet
<svanheule>
they also tend to use those shared clock line on bitbanged I2C busses (sensors, SFP cages), which are unfortunately not straightforward to support :-(
danitool has joined #openwrt-devel
robimarko_ has joined #openwrt-devel
<ynezz>
rsalvaterra: ugh, did I broke something with my recent changes in target.mk?
robimarko has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
<rsalvaterra>
ynezz: Not sure, I think you're innocent, since you didn't touch the code I yanked out.
nixuser has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
nixuser has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
minimal has joined #openwrt-devel
<Habbie>
good news, i can boot a kernel on this very old very small device
dangole has quit [Remote host closed the connection]
dangole has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
wsh has joined #openwrt-devel
<hurricos>
Habbie: which device?
<hurricos>
mrkiko: GS1900-24HP!
rua has joined #openwrt-devel
<hurricos>
Or any of the GS1900's. If they fail to boot multiple times they switch partitions; they have an A and B partition; you leave the OEM firmware in the second partition.
wsh has quit [Quit: Using Circe, the loveliest of all IRC clients]
wentasah has joined #openwrt-devel
<hurricos>
GS1900 has full DSA support, so you can switch and route as needed.
<Habbie>
rsalvaterra, uh, whatever came with openwrt 15.05 i think
<Habbie>
rsalvaterra, 3.18.20
<Habbie>
rsalvaterra, i almost choked on my drink when i saw this last line pop up:
<Habbie>
[ 4.900000] ---[ end Kernel panic - not syncing: Out of memory and no killable processes...
<Habbie>
[ 4.900000]
<Habbie>
[ 82.730000] random: nonblocking pool is initialized
<Habbie>
hurricos, https://wikidevi.wi-cat.ru/Sitecom_WL-330 (I'll fill the other fields later, but it appears to be a 4MB/32MB device - although uboot tells the kernel it's 16?!)
<Kardy>
Hello openwrt-devel folks. Not sure if this is the correct forum for my questions about the build system...
mattytap has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
mattytap_ has joined #openwrt-devel
<Habbie>
Kardy, you won't find out until you ask
<Kardy>
Hi Habbie. I know, just wasnt sure about the forum.. :)
<Habbie>
i'll tell you two things
<Habbie>
one, both this channel and #openwrt sometimes just don't respond to questions - because the right person wasn't awake, or didn't have time, etc.
<Habbie>
two, if your question clearly belongs somewhere else, you'll hear it - unless condition one applies :)
rua has joined #openwrt-devel
<Kardy>
i see. ok, so...
<Kardy>
I would like to build ARCH=mxs kernel where the NAND/GPMI driver is included (currently is not). Modding the kernel config is obvious (what/where). However, is there something like a "defconfig" in/part of the buildsystem? It seems to me "everything" has to be manually "menuconfiged".
<Borromini>
there's lots of defconfigs, if you go through some makefiles you'll see certain stuff depending on ARCH x or y, or driver z or z' e.g.
<Borromini>
some targets have also some flags set that pick up some features by default afaik
mattytap has quit [Ping timeout: 480 seconds]
<Kardy>
so, just mod the kernel config and "make"?
<Kardy>
i dont need to do pkg selection or "anything"?
<Borromini>
no modding the kernel config is the wrong way, you want OpenWrt's buildroot to do that for you.
<Borromini>
after all, that's what ticking stuff in 'make menuconfig' does
<Borromini>
e.g. for target specific flags: see target/linux/ath79/nand/target.mk
<Borromini>
you can configure the kernel yourself directly in the buildroot but that is indeed fully manual (and not how you want to do things unless you're tinkering)
<Kardy>
sadly there no menuconfig for the GPMI/NAND to build as a module, so i assume it has to included in the kernel image (thus my attempt for manual kernel conf)
<Kardy>
i'll take a look at the file you've mentioned, i guess i've overlooked something obvious :)
<Borromini>
no, that's higher level, but you can create a makefile for drivers so they show up
<Borromini>
but stuff in package/kernel/ should give you some pointers I think
<Kardy>
ah, haven't thought of that, modding the menuconfig it self to do the job... thnx for that
<Borromini>
especially if you need it compiled in that's the way to go. If you want it to be a module, you can write a Makefile if you'd like it go get into OpenWrt proper
<Kardy>
is for a custom mxs board with NAND storage; Thanks to you both! The docu and the forum post are what i was looking for!
<Kardy>
private use
<Borromini>
:)
<Kardy>
The help is much appreciated, i was getting desperate "doing all the menuconfig" by hand; "make download world" to the rescue!
<Kardy>
Also reading the correct docs (was fighting with the SDK & imagebuilder) helps :D
cbeznea has quit [Quit: Leaving.]
<hurricos>
Habbie: RIP :( 16/32s are really common with RT3052s and ramips in general
<Habbie>
hurricos, the chip I can identify really is 16 - wonder why uboot says 32
<Kardy>
Once more, thanks to you both for your help and time. Time to RTFM.
<Habbie>
hurricos, i haven't found an openwrt image (on downloads) that will run from RAM on it, sadly
<Habbie>
hurricos, (because i want to dump the flash)
<hurricos>
Otoh ... 32 should die >:)
<Habbie>
oh sure
<Habbie>
i have no illusions about -usefully- booting openwrt on it
<hurricos>
and you won't find me clinging to the last moment on my uh, my MR16s. Nope.
<Habbie>
i'm just using this EUR 2,50 device to practice my openwrt and hardware hacking so i don't blow up a 80 EUR device later
* hurricos
suppresses a tear
<hurricos>
Nice!
<Habbie>
i did actually install 18.06 on a 4/32 device last week for actual usage
<Borromini>
i've managed 21.02 on an 8/32 device so far, but 22.03/master will break a lot of those, with WPA3 and its crypto dependencies
<Borromini>
if anyone gets 22.03 on 8 MiB devices with WPA3 (no PPPoE or LuCI or anything needed) I'd be happy to hear, but it seems the bigger kernel does them in as well.
<Habbie>
does wpa3 require recent hardware or is it just software?
<Habbie>
(that 4/32 box has luci, so i could probably improve the disk space situation if i wanted to)
<hurricos>
Habbie: It requires SAE which requires your driver to be happy about managing encryption keys on a two-tuple MAC-addess basis
<hurricos>
pretty much all hardware can. Some, though ... I doubt brcmfmac or mwlwifi will be happy about that.
<hurricos>
b43 doesn't use hardware cryptography, so no worries there.
<Habbie>
right
<Habbie>
if they leave enough to the software, it'll work if the driver does it
<hurricos>
well, no driver can realistically do the encryption. So if the hardware has complicated key management for its hw crypto engines ...
<Borromini>
a lot of ath9k devices still seem to be able to manage
<Habbie>
so, related unrelated, what i'm really looking for right now is some uImage that'll boot on 16MB, in RAM, and gives me enough to read flash
<Habbie>
hurricos, so, this means the CPU/SoC has the capability to map SPI NOR?
<hurricos>
Yes!
<hurricos>
For example, the MR16 uses the bootcmd `bootm 0xbf0a0000`.
<Habbie>
it's 0xbfsomething here too
<hurricos>
It clearly doesn't have >3GB of RAM ....
<Habbie>
will experiment in a moment
<hurricos>
might as well try 0xbf000000
<hurricos>
if you read invalid memory it'll just hang :P
<Habbie>
oh i have the number in my pastebuffer on the other desk
<hurricos>
fwiw, you actually DO copy the RAM into the right memory region
<hurricos>
it just does so silently
<Habbie>
ack
<hurricos>
and probably with some compression.
<Habbie>
oh, any clue how i would pass kernel cmdline args in this case?
<hurricos>
probably `setenv cmdline`.
<hurricos>
the kernel command line arguments get copied into a memory region which is pointed to by one of the initial boot registers
<hurricos>
or at least that's how it's done in PowerPC
<Habbie>
ack
<hurricos>
I'm sure it's no different in MIPS
<Slimey>
hurricos do you want to play a game?
<hurricos>
Slimey: No :^)
<Habbie>
it boots into some ncurses-like thing that asks me user/pass, and i have no idea what it wants from me
<Habbie>
if i hold the reset button for 10 seconds, i get a root prompt for 1 second :D
<hurricos>
Habbie: Yes, but as long as you get the `md` dump you can send us the content
<Slimey>
its one of those puzzle things where you have to find something, so in this picture find the EEPROM chip, https://i.imgur.com/N6SONAW.jpeg
<hurricos>
at which point we can fish for an md5 or md2 table and find the hash
<Habbie>
hurricos, yep
<Habbie>
hurricos, don't even know if it's a hash, but we'll find out
<hurricos>
Slimey: How did you get a photograph of my room?
<Slimey>
:P
<Habbie>
Slimey, why is there a blanket on my other desk? :)
<Slimey>
i moved a to new house and just started unpacking
<Slimey>
haha
<hurricos>
I can't find the eeprom.
<Habbie>
i suspect there are roughly 12 eeproms in this picture
<Slimey>
its on the wap sitting on the white label
<hurricos>
there are at least a few WAPs in this photograph.
<hurricos>
Habbie: let me know as soon as you ahve a copy of the dump, or even just the raw console output
<Slimey>
still looking for that hash?
<Habbie>
raw console output of md you mean?
<hurricos>
yes.
<Habbie>
got it
<hurricos>
I'm assuming you're at 115200 baud so
<Habbie>
57600
<hurricos>
if not, ouch
<hurricos>
not bad, its only what, 5 minutes of `md`.
<Habbie>
9.26 minutes multiplied by however md formats things ;)
<hurricos>
I assumed 4 chars per byte and said
<hurricos>
(4096*1024)/(57600/4)
<hurricos>
anyways
<Habbie>
57600 is in bits
<hurricos>
oh wtf.
<Habbie>
:)
<Habbie>
there's no rush here anyway
rua has quit [Ping timeout: 480 seconds]
Borromini has quit [Quit: leaving]
rua has joined #openwrt-devel
KGB-2 has quit [Remote host closed the connection]
Kardy has quit [Quit: Leaving.]
Kardy has joined #openwrt-devel
Misanthropos has joined #openwrt-devel
KGB-2 has joined #openwrt-devel
<PaulFertser>
57600 is in baud, so with 8 bits, no stop bits, 1 stop bit (and one start bit) per byte it'll be 10/8 slower than bit/s.
<Habbie>
that too :)
<PaulFertser>
I was told recently to go read about 8b/10b SERDES way of coding to see how this traditional 57600 baud for UART can be viewed as 5760 symbols per second transmission, and there was a fair point in it.
<Habbie>
symbols, but not bytes?
<Habbie>
also, my 9.26 minutes missed 8/10
<PaulFertser>
Symbols, yes.
<PaulFertser>
Talking about traditional UART as something that transmits those 8b symbols would be stretching the definition, I agree, but still interesting learning opportunity.
robimarko_ has quit [Quit: Leaving]
<Habbie>
indeed :)
<Habbie>
.. 7 minutes later, of course this also means 5760 bytes per second
<hurricos>
It looks like ServeTheHome documented all of their firmware manipulation process for all of their hacked-up cracked-open decade-old top-of-the-shelf cloud-native enterprise hardware.
<hurricos>
neggles: Including how to flash an LB6M (24 x SFP+) over to a layer-3 compatible firmware
<hurricos>
and it's *powerpc*
<hurricos>
MPC8541 :eyes:
<hurricos>
BCM56820 :eyes:
<Habbie>
hurricos, more memory mapped SPI NOR? :)
<hurricos>
Oh, do you have some for me, or?
<Habbie>
no, just responding to brokeaid.com
<hurricos>
oh
<hurricos>
No, this is just ... crazy stuff. It's good to finally see this get documented. I don't think we're ever gonna see DSA support for BCM56820, but then again, maybe I should try to be the change I want to see in the world?
<hurricos>
(I have no idea how to even go about starting to implement DSA support. :^)
<Habbie>
hehe
<Habbie>
as for the change, either you do it or somebody else - but is there somebody else?
<hurricos>
Or I just settle for learning Fastpath which is apparently the best firmware here :^)
<hurricos>
or, sorry. Brocade's CLI is what I'd have to learn, which is apparently very similar to Cisco's. Brocade's software stack is on top of Fastpath.
<hurricos>
Looks like any hardware with this switching silicon in it will be running using Fastpath. First I've heard of it to be honest.
wentasah has joined #openwrt-devel
<hurricos>
realistically though, this hardware fits in the very tiny niche of "more SFP+ ports than Realtek", and that's it. And it won't keep that for long as more vendors pick up RTL93xx's
<hurricos>
thanks to folks here like svanheule getting way ahead of that eventual inrush