<bkallus>
I'm trying to get a new MT7620 router to boot over TFTP, and no matter what image I give it (even if I give it a dump of its own boot partition), I get this: Uncompressing Kernel Image ... LZMA ERROR 1 - must RESET board to recover. Given that this happens even when I use a stock image (but not when booting from NAND), I think this isn't due to being out of memory or LZMA dictionary size. Any ideas what it could be?
<Slimey>
nvidia bluefield-2 dpu :O
<bkallus>
Figured it out; I needed to leave more room for the decompressed image. Pushing the load address ahead worked.
<neggles>
bkallus: depending on your u-boot, you may also need to limit the size of the decompressed kernel image to 8MiB
<neggles>
using an uncompressed kernel or lzma-loader will get around that
<neggles>
I feel like I should write a wiki page about this since it keeps happening to people
<bkallus>
I tried selecting compression: none in the ramdisk compression section of menuconfig but it still gets stuck decompressing lzma (and binwalk shows lzma compression as well). How do I make an uncompressed kernel?
<bkallus>
wiki page would be much appreciated
<neggles>
bkallus: what does your section in the image makefile look like
<bkallus>
Just got it booting. The device is the Linksys EA6100, but I'm starting from the config for the E1700 (so I'm using a copy of that section in the image makefile)
<neggles>
adding `KERNEL := kernel-bin | append-dtb | uImage none` will override kernel to uncompressed, adding `$(Device/uimage-lzma-loader)` adds two lines from the main image/Makefile which in theory makes lzma-loader do its thing
<neggles>
(that probably needs more effort than just adding the makefile def)
<bkallus>
Thanks. I'm sure that will come in handy next time I run into decompression issues. Definitely will require more work that just changing the makefile. Wifi's up next.
<grift_>
basically ip6-based qos does not work because it doesnt differentiate between ip and ip6
<grift_>
i guess maybe the best approach would be to rename that function to qosdef_append_rule_ip4_limit() and to add a qosdef_append_rule_ip6_limit()
<grift_>
and then somehow determine whether the ip is ip4 or ip6 and call appropriate function accordingly?
<hanetzer>
don't suppose I could ask some non-openwrt (for now?) embedded arm questions here? I have a board I'm working on, some hisilicon chip (hi3521a), using stock buildroot to make an initrd, but its failing to give me a getty no matter what options I'm passing to buildroot or the kernel.
<hanetzer>
m not workign on a camera atm. but rather, a camera aggregator (nvr)
<swiftgeek>
how do i cut out that fishing troll out of google search xD it's such a hog on resources and searching for datasheets will often bring it up
danitool has quit [Remote host closed the connection]
<Grommish>
swiftgeek: You can try using a filetype:pdf to limit the results to pdf files only (which most datasheets I've seen tend to be)
<swiftgeek>
i have the opposite issue, 0 results :)
<swiftgeek>
tried removing div with it in µblock but it still loads it for few seconds
<swiftgeek>
ah i did stop load early and blocked in that moment and now it's gone
valku has quit [Read error: Connection reset by peer]
goliath has joined #openwrt-devel
ecloud_ has quit [Ping timeout: 480 seconds]
ecloud_ has joined #openwrt-devel
mattytap has quit [Read error: Connection reset by peer]
Tapper has joined #openwrt-devel
ecloud_ has quit [Ping timeout: 480 seconds]
mattytap has joined #openwrt-devel
<nick[m]1234>
@mrkiko ping
<mrkiko>
nick[m]1234: poing :D
<nick[m]1234>
whats the status of the glinet conversion?
<mrkiko>
nick[m]1234: thanks for helping out!! So, I am little bit difficulties in applying the patchset - because i experience some merge issues I don't know how to deal with
bluew has quit [Remote host closed the connection]
bluew has joined #openwrt-devel
<nick[m]1234>
you can also just checkout the branch
<mrkiko>
nick[m]1234: give me a git command and I'll do so
<mrkiko>
nick[m]1234: ok; for the DTS, should I use the one provided by worthel or yours?
<nick[m]1234>
did I provided one? xD
mattytap has quit [Read error: Connection reset by peer]
<nick[m]1234>
Thee is more than just the dts. you may also need to adjust the network setup files
<mrkiko>
nick[m]1234: yeah; I tried it and it seemd to not work, but it probably was also due to me not updating the network script, which I will do after pulling the DTS and replacing it
<mrkiko>
nick[m]1234: sure
mattytap has joined #openwrt-devel
danitool has joined #openwrt-devel
<nick[m]1234>
give me a sec, I have to find the patch
<mrkiko>
nick[m]1234: no problems; in the github openwrt issue, search for "mrkiko.txt" in case - dts provided by worthel
<nick[m]1234>
do you have serial access?
<nick[m]1234>
otherwise, I patch always the basefiles and remove the "disabled=1" part of the wifi, so I can connect via wifi
<mrkiko>
nick[m]1234: no; but I did set things up so wi-fi will get enabled by default after "sysupgrade -n"
<mrkiko>
nick[m]1234: I am using the uci-defaults approach for re-enabling wi-fi
mattytap has quit [Read error: Connection reset by peer]
<nick[m]1234>
mrkiko: I see. even better than mine approach ^^
<nick[m]1234>
ip l did not show any network interface? or dmesg?
<mrkiko>
nick[m]1234: well, I tried long time ago, so I do not remember details. In my opinion eth' was present, but I did not update network script so who knows.
mattytap has joined #openwrt-devel
<mrkiko>
nick[m]1234: btw, did my uboot modification turned to be useful on the 7530 ?
<nick[m]1234>
so just network scripts were missing?
<mrkiko>
Yes. But then wortel proposed a DTS update, so I was confused what to use
<mrkiko>
DTS are little bit hard for me
<mrkiko>
qca8075 as far as I can tell from the specs
ecloud_ has joined #openwrt-devel
<nick[m]1234>
mrkiko: i am also a noob in dts things. However, i like to learn. So maybe ypu can do a "concert glinet..." commit AS far AS you can and send link to me?
<mrkiko>
nick[m]1234: sorry, I do not understand. You mean - try now to build / convert my device and send you somehow a patch to do so basedo n the sartura branch?
<nick[m]1234>
yes. or fork openwrt, and push it into your github
mattytap has quit [Read error: Connection reset by peer]
<mrkiko>
prefer to send the patch to you somehow. That said, would you mind to give me some hints on how I might update the network config scripts after dts changes?
mattytap has joined #openwrt-devel
<nick[m]1234>
the device has a lan and wan port?
<mrkiko>
nick[m]1234: I have only the gl-b2200 from glinet (the other glinet devices I have aren't part of the ipq4xx family))
<mrkiko>
nick[m]1234: yes
<mrkiko>
nick[m]1234: ok, so maybe I can do this with more time and let you know once I finished
bluew_ has joined #openwrt-devel
srslypascal is now known as Guest3243
srslypascal has joined #openwrt-devel
<nick[m]1234>
mrkiko: the device tree you gave me has already labels for lan, and wan
<nick[m]1234>
just add the device to the network.sh at alfa-network,ap120c-ac
bluew has quit [Read error: Connection reset by peer]
<nick[m]1234>
that should be enough
<nick[m]1234>
mrkiko: I am 24/7 in this irc. just ping me when you need help :)
Guest3243 has quit [Ping timeout: 480 seconds]
<mrkiko>
nick[m]1234: thanks a lot!! I'll try it !!
<mrkiko>
nick[m]1234: you mean the "mrkiko.txt" ?
<nick[m]1234>
yep
<mrkiko>
nick[m]1234: ok, I'llcontact you back. Sorry, I did not know you where the polynomialdivision :D
<robimarko>
He is in covert mode on IRC
<nick[m]1234>
actually, I think my github name is not a real nickname for me ^^ so it is more the other way around
<mrkiko>
robimarko: :D :D
<mrkiko>
but guys, looking at the differences between the two DTS I see a lot of them
<mrkiko>
robimarko: yes, nick[m]1234 wants REALLY this to get merged :D :D
<robimarko>
mrkiko: That makes two of us
<nick[m]1234>
I take care of the "MikroTik Wireless Wire Dish LHGG-60ad"
<mrkiko>
:D I am happy to help out, since it seems a pretty nice improvement especially for long-term
<mrkiko>
can you help me compare the dts and understand why there are so many differences?
<robimarko>
Well, we are in a state where conversions are needed
<robimarko>
Until that is done, its not getting merged
<robimarko>
Other stuff should be ready, especially since Serhii sorted the calibration out
<robimarko>
Though, I need to look into cleaning that up
<mrkiko>
the gl-b2200 is the only ipq4xx device I have that's not been converted - Fritz!box 7530 and fritz!box 4040 are already converted
<nick[m]1234>
So we have 15 of 64 devices. So 49 left. We are 3 persons. So each of us need to take ~17 devices into account. Can we maybe open a git where all of us can push new device support?
mattytap has quit [Read error: Connection reset by peer]
<robimarko>
Well, thats the trick
<robimarko>
In theory we can blindly convert all of them
<robimarko>
But, MAC adresses and port ordering is gonna get broken for sure
<mrkiko>
I rememb Ansuel trying to work on an automated conversion tool
<mrkiko>
so - should I replace blindly the dts with the new one and you will sort out differences or is it better to understand them now and merge the whole thing up? I am worried something has changed in the meantime
<robimarko>
Well, there shouldnt be that much changes
<mrkiko>
robimarko: I didn't expect it as well, but still...
<robimarko>
You can drop the disabled switch port nodes
<robimarko>
They are disabled by default
<robimarko>
Also, why does B2200 use clk_ignore_unused?
<robimarko>
That stinks like copy/paste from the stock FW DTS
<nick[m]1234>
are you planing to do further conversions? it would be nice not two people doing same conversions
<nick[m]1234>
otherwise, my plan is to do conversion, make image, send it to device maintainer and ask for feedback
<robimarko>
Ahh, I think I converted all of the devices I have
<robimarko>
And I dont expect to have time for more blind conversions in the following week
<robimarko>
*weeks
<nick[m]1234>
okay, good to know. I will try going on with that list.
<nick[m]1234>
maybe I will write a post
<mrkiko>
robimarko: I know I am asking a lot, but ... could you please prepare patch to change the dts to conver it and maybe clean it up by removing that part, so I can try it but still getting something maintenable later? Because if I try things as they are right now, we are going to face this issue anyway later.
<robimarko>
Well, if you can tell me which switch ports match the ports on device I can do it
<mrkiko>
nick[m]1234: I will for sure do my best to get the b2200 converted
ekathva has joined #openwrt-devel
ekathva has quit [Remote host closed the connection]
<robimarko>
Also, how are the MAC-s set on B2200?
<mrkiko>
robimarko: telle mw what I need to do and I'll do it; I am a blind person, so I can't read labels, but i can well use swconfig :D
<robimarko>
Basically, I just need to know to which switch port is WAN connected and to which one LAN
<robimarko>
But if there are only ones, current swconfig defconfig should be enough
<mrkiko>
yes the devices has only 2 ports, one lan and one wan
<robimarko>
Uhh, is it just me or does the current swconfig config specifiy a 5 port switch instead
<mrkiko>
robimarko: you're right
<mrkiko>
of these, port 0 is lan
<robimarko>
Are you sure?
<robimarko>
0 is the CPU port
<mrkiko>
can't tell you, give me a moment - I'll go to play with cables :D
<robimarko>
Port 0 should always be up since its connected to the CPU
<mrkiko>
ok
<mrkiko>
0: CPU port; 5: LAN port; 4: WAN port
<robimarko>
Great, thanks
<mrkiko>
for mac-addresses, where can I look?
<robimarko>
You can see whats set with ip or ifconfig
<robimarko>
I assume they are set by using aliases
<Tapper>
after typing ./scripts/env new 6260 it just says error: unknown switch `b'
<Tapper>
Grommish: Has the same thing.
<mrkiko>
Tapper: what distros are you using ?
<mrkiko>
for me it seems to work
<Tapper>
I am using ubuntu under WSL on windows 11
<Tapper>
It was working for me.
<Tapper>
But after a new check out it's broken
<mrkiko>
Tapper: well, that message doesn't seem to come from the script
<bkallus>
Working on a dts for the linksys ea6100. I think I've set everything up right in the nand section, but I don't see any mtd devices in /dev when I boot the iamge. Any ideas what the problem could be?
<mrkiko>
Tapper: may you launch it with a "strace"
<mrkiko>
Tapper: sorry
<mrkiko>
Tapper: try something like bash -x ./scripts/env ...
<Grommish>
mrkiko: This is a new VM, my old pre-set VM doesn't have the issue.. the issue is that if the branch already exists, git checkout -b master isn't a valid command
<Grommish>
mrkiko: because checkout -b fails on an existing branch
<Grommish>
mrkiko: The commit specifies the checkout -b master in the init
<Grommish>
mrkiko: and that borks the show
<Grommish>
Reverting the commit will fix the issue, whether reverting the revert after it's "properly" init'd by the script will then work, I dunno, but on a fresh install, it fails
<mrkiko>
nick[m]1234: robimarko: started build, will keep you posted here and maybe on the openwrt github issue about what happens
<nick[m]1234>
I am also building right now toolchain to start converting some boards :D
<nick[m]1234>
however, my pc is shitty, and takes a lot of time...
<mangix>
converting from what?
<mrkiko>
Mangix: ipq4xx from swconfig to DSA driver
<mrkiko>
nick[m]1234: robimarko: I am not generally favorable to "forceful" approaches - still hope this can get merged soon!
<nick[m]1234>
Mangix: rom sw to dsa
<nick[m]1234>
from*
<mangix>
nick[m]1234: ccache is your friend
<nick[m]1234>
Mangix: yeah, but I don't have enough disk space either
<nick[m]1234>
I had mt7622 buildchain. and now switching to ipq40xx I need to delete everxthing again xD
<mrkiko>
nick[m]1234: rt3200?
<Tapper>
Grommish: thanks
<nick[m]1234>
rt3200 and banana pi r64
<Grommish>
Tapper: yar sir.. I tried to catch you last night, but you had already left
<mrkiko>
nick[m]1234: what about the r64? Never tried it but was curious
<nick[m]1234>
I love it
<mangix>
nick[m]1234: maybe cfarm then :P
<Tapper>
If I do git revert b5ab6d0bc577e81a77f39211883f6c0c4155b6ed will it come back when I do a git pull.
<Grommish>
Tapper: I don't use git pull, I use git rebase but you'll probably get conflicts. However, you can remove the revert after you set it up
<Grommish>
Tapper: the issue seems to be that once it's setup, it's not an issue to be there
<Tapper>
OK cool
<Tapper>
let me get this shit working now then so I don't forget.
<Grommish>
Tapper: It's only when it hasn't already been setup and goes to do the initial env setup
<nick[m]1234>
Mangix: never heard of it before. interesting. :) would you recommend?
<mrkiko>
ok, since I will be "cutting" my connection to the Internet, I go for now. Coming back later to let you know how it went
mrkiko has quit [Ping timeout: 480 seconds]
MAbeeTT2 has joined #openwrt-devel
<hanetzer>
does 'stock' linux support mtdsplit partitioning?
<robimarko>
Nope
<hanetzer>
blerg. it should :P
MAbeeTT has quit [Ping timeout: 480 seconds]
<robimarko>
Well, somebody has to contribute support for it
<robimarko>
OpenWrt has had the mtdsplit for ages, no idea if anybody tried upstreaming it
mrkiko has joined #openwrt-devel
<mrkiko>
nick[m]1234: robimarko: it works! And MAC addresses seem to be as they where before
<mrkiko>
eth0 and lan have the same mac btw
<robimarko>
Great
<robimarko>
Yeah, eth0 uses ethernet0 alias and LAN will inherit that
<robimarko>
WAN uses ethernet1 alias so bootloader populates its MAC separately
<mrkiko>
robimarko: so, te bootloader actually patches the DTS
<robimarko>
Yeahž
<mrkiko>
robimarko: :D :D
<robimarko>
Yeah, sorry for the croatian/slavic "ž"
<mrkiko>
robimarko: would you like a dmesg or ifconfig or something?
<robimarko>
Nope, if it works than thats it
<mrkiko>
robimarko: actually it made me smile ...
<mrkiko>
robimarko: thanks for the help! Also to nick[m]1234
<mrkiko>
robimarko: add my Tested-By tag if you want, and hope you can incorporate this so the device will be ready when this gets merged :D
<mangix>
nick[m]1234: yes
<mangix>
building OpenWrt on PPC was...fun.
<robimarko>
Great, can you comment on the PR and I will add it tommorow
<robimarko>
There is one or two more device PR-s for the conversion I need to pick anyway
<robimarko>
IPQ807x has been sucking my time
<Tapper>
Grommish: How do I change back. I have env set up now thanks to you.
<mrkiko>
robimarko: sure! BTW - do you have any hypotetical timeline for merging this?
<robimarko>
Turns up cleaning up for upstreaming is not fune
<robimarko>
mrkiko: Into OpenWrt master?
<Grommish>
Tapper: git reset --hard to the commit just before the revert
<Grommish>
Tapper: That'll remove the revert
<mrkiko>
robimarko: I imagine it's a lot of work. Yes, OpenWrt master
<Tapper>
can I just do git reset --hard
<Tapper>
I want to go back to head of master
<robimarko>
mrkiko: IPQ807x or DSA?
<Grommish>
Tapper: Well, unless you'd added other stuff, the commit before the revert is your HEAD
<Tapper>
OK thanks
<mrkiko>
robimarko: DSA conversionof ipq4xx
<robimarko>
Really no idea, we are kind of stuck on board conversions
<robimarko>
Other than that it should be readyy
<mrkiko>
robimarko: should admit I am under the impression ipq807x will be difficult to use due to bootloaders requiring signature, but I may well be wrong
<robimarko>
The calibration stuff could use some cleanup
<robimarko>
mrkiko: Depends on the IPQ807x board
<robimarko>
Secure boot is still optional
<robimarko>
And even those that use it, use it in a stupid way
<mrkiko>
robimarko: great, so I shouldn't lose my hope
<robimarko>
In which you just replace the bootcmd with a manual one instead of bootipq/bootqca
<mrkiko>
robimarko: that said, I find your work very useful and great in any case
<robimarko>
Thanks
<robimarko>
Have started upstreaming various drivers for IPQ807x so thats sucking time
<mrkiko>
robimarko: ehehe... can understand
<mrkiko>
robimarko: I guess the ipq4xx DSA conversion will helpin upstreaming some pieces there as well or am I wrong?
<robimarko>
Unfortunatelly not
<robimarko>
Ethernet controller looks brand new despite the same EDMA name
<robimarko>
Switch is semi-similar but now has 2 10G ports
<robimarko>
Completely different PCS
<robimarko>
BTW, they even have EDMA v2 for the upcoming WiFi 7
<mrkiko>
robimarko: sorry, I didn't explain myself correctly. I mean - the ipq4xx DSA conversion will help in upstreaming some ipq4xx code and help long-term maintain the target
<robimarko>
Oh yeah
<robimarko>
The ethernet controller has been submited upstream in a cleaned up version
<robimarko>
We have hired Bootlin to take care of that
<mrkiko>
robimarko: ehe yeah I "knew" IPQ807X is a "different" story - I knew the NSS thing, and saw some traces of it's different versions with names such as "nss deluxe" or things like that. Crazy.
<robimarko>
Well, those are feature profiles
<robimarko>
You have Premium and Enterprise
<robimarko>
Weirdly, Premium has more features and everybody uses that
<mrkiko>
robimarko: ... yeah
<robimarko>
As it has most of the stuff autoconfigured, while Enterprise doesnt enable anything by default
<mrkiko>
robimarko: I mean - I seen that names. Didn't know everyone used that. Maybe enterprise was tailored for more stability? Don't know.
<robimarko>
No idea, its basically meant for the end user/vendor to be able to configure everything as they wish
<robimarko>
Without any "help" from defaults etc
<robimarko>
Havent seen anybody using it though
<robimarko>
Deluxe was from the IPQ806x days, I think they dropped that a long time ago
danitool has quit [Remote host closed the connection]
<mrkiko>
robimarko: is "nss" firmware loaded kinda like ath10k firmware is, or is it flashed somehow in the switch
<mrkiko>
robimarko: in other words - how do you "choose" what variant to use
<robimarko>
Its loadable
<robimarko>
So you just ship the premium or enterprise blobs and thats it
<mrkiko>
robimarko: ok
<robimarko>
But for their config to not interfere you gotta build with enterprise profile
<mrkiko>
robimarko: from one side you build with enterprise to not have autoconfiguration, and on the other you use the blob you want, so that it autoconfigures stuff?
<robimarko>
No
<robimarko>
Blob probably exposes more interfaces
<mrkiko>
robimarko: ... thanks for the timeyou spent helping me out. I did commented on the github issue. Hope the conversions can go on
<robimarko>
Will probably pick it all tommorow evening
<mrkiko>
robimarko: thanks! In case you can pick up even the BDF fix :D :D
<mrkiko>
robimarko: p.s.: now I see "clk: not disabling unused clocks"
<mrkiko>
the system works fine in any case
danitool has joined #openwrt-devel
<robimarko>
mrkiko: I dont have commit rights, so somebody else has to push that commit
<robimarko>
mrkiko: That print is due to the clk_ignore_unused in bootargs
<robimarko>
That was there already
<mrkiko>
robimarko: ok... does it make sense for me to try to remove it some day and in case, if it works, remove it completely from dts?
<robimarko>
No idea why it is there, probably copy/paste from the stock DTS
<robimarko>
As everything should work without it
<robimarko>
It basically forces the kernel not to disable clocks which are not used by any driver
danitool_ has joined #openwrt-devel
<robimarko>
Anyway, enough for today, bye
robimarko has quit [Quit: Leaving]
danitool has quit [Ping timeout: 480 seconds]
linusw__ has quit []
PaulFertser has quit [Ping timeout: 480 seconds]
PaulFertser has joined #openwrt-devel
mangix_ has joined #openwrt-devel
mangix has quit [Read error: Connection reset by peer]