<dangole>
hurricos: i run into a problem where ntpd would not ajust the time on boot and was suspecting ujail. then turned out it was the hardware RTC of the SoC being used eventhough there was no battery feeding it on that board, hence it would come up as 2000-01-01 on boot which was just a bit too large of a jump for ntpd. disabling the RTC driver in DTS fixed it.
Tapper has quit [Ping timeout: 480 seconds]
szy_mat has joined #openwrt-devel
<szy_mat>
Hi! I'm trying to run gpiommc @ Chaos Calmer, the most important thing is that all my gpios are "inverted" by push-ups - I have GND level when i set them up (via /sys/class/gpio) and on the MISO the default level is HIGH, it could be brought down by connecting to GND and then it's LOW as an input
<szy_mat>
is it even gonna work?
<szy_mat>
right now of course
<szy_mat>
mmc_spi spi32760.0: SD/MMC host mmc0, no DMA, no WP, no poweroff
<szy_mat>
[ 980.870000] mmc_spi spi32760.0: no support for card's volts
<Slimey>
interesting i was just mentioning something like with with a certain ath79 unit
<Slimey>
the other day anyway
minimal has joined #openwrt-devel
szy_mat is now known as szy_mat_
<hurricos>
johnf: if anything it'll be because of the move to DSA
<hurricos>
there are two extra bytes at the end of some of the DSA frames which, not being word-aligned, slow everything down
<hurricos>
(specific to qca8k). At least that's what I've heard. I have no idea what this actually means :^)
szy_mat has joined #openwrt-devel
<johnf>
hurricos: lol, thank you
<johnf>
I did actually see some discussion about that I think
<hurricos>
he.
<hurricos>
heh*
<hurricos>
If I remember correctly, the folks pushing for qca8k are OpenWrt, though. I think there's some attempts at a fix somewhere I just don't know where they've gone
<huaracheguarache>
blocktrron: Hi! I've made some changes to the hostapd pull request we talked about yesterday.
<T-Bone>
damnit I hate systemd
T-Bone is now known as f00b4r0
pmelange has joined #openwrt-devel
zatwai_ has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
zatwai has joined #openwrt-devel
Tapper has joined #openwrt-devel
danitool has joined #openwrt-devel
arekm has joined #openwrt-devel
<arekm>
Hi, Tried to add few photos to hardware db wiki but can't register due to "Self-Registration is currently disabled or conf/users.auth.php is not writable. Please ask your DokuWiki administrator to create your account manually." Can I get access?
<Borromini>
PaulFertser: ^^
<PaulFertser>
arekm: please send me your desired nick and e-mail.
<Borromini>
guys, a question: say I have two trees for a device - one that builds NOR images, another NAND. Can I boot off the ramdisk the NOR tree provides, and still flash the NAND image from it?
<Borromini>
i'm not sure if sysupgrade looks to the firmware it's booted from for 'pointers', but I suppose it doesn't since one can use a ramdisk to flash sysupgrade images...
<dangole>
Borromini: devices having more than one type of flash which is suitable to install OpenWrt on are pretty rare. sysupgrade behavior in this case is board-specific and defined in /lib/update/platform.sh
<dangole>
Borromini: so as this file is board-specific, you will have to say which exact board you are using
<dangole>
Borromini: as (just to name examples) for gl-inet ar750 (with both nand and nor) it is solved using different images, while on more modern targets (say: bpi-r64) a single image can be used accross all storage types (spi-nand, emmc, sdcard in this case)
<dangole>
arekm: please send me your desired username, email address and whatever else should go to your profile by pm, i'll create an account for you then
robimarko has joined #openwrt-devel
<robimarko>
hurricos: IPQ40xx is not using DSA at all currently
<robimarko>
There is a PR and once everything is converted it will use it
<robimarko>
But there havent been any performance regressions with DSA at all
<arekm>
dangole: PaulFertser did that already, thanks.
<Borromini>
they both show the same behaviour, if i understood right i'd first need to boot the initramfs to replace RouterBoot
<Borromini>
i'll leave it as is for now, enough messing around for today :)
<Borromini>
dangole: MikroTik disabled it out of the box :^)
<Borromini>
<- first MikroTik device
<robimarko>
Yeah, thats the easiest way
<robimarko>
Or using jailbroken ROS
<Borromini>
hmm there was a jailbreak shared in the topic too right? Should look into that then.
<robimarko>
Yes
<Borromini>
thanks for your 5.15 tree btw :)
<robimarko>
Lets just say that I doubt his "loader2" will ever be used in OpenWrt
<robimarko>
He had something similar for IPQ40xx until we finally managed to find a generic soluton
<robimarko>
Cause, licensing is a bit sketchy with it
<Borromini>
yes, i understand. and the binary bit
<robimarko>
Well, source is public but a lot of it was pulled from U-boot
<robimarko>
And is reimplementing FDT support
<robimarko>
So, I would personally preffer to either replace RouterBoot with U-boot
<robimarko>
Or boot U-boot using RouterBoot and then use NAND normally
<robimarko>
Cause MikroTik uses damn YAFFS2 again
<Borromini>
yes, I saw that
<robimarko>
Adron is great at hacking things and reverse engineering but he usually doesnt upstream
<robimarko>
And there is a big difference in getting it to work and upstreaming
<robimarko>
So that is where our approaches differ
<Borromini>
i am seeing some parallels with certain Realtek hackers :^)
pmelange has left #openwrt-devel [#openwrt-devel]
<robimarko>
Currently lighter version of that I would say
swalker has joined #openwrt-devel
<robimarko>
So, RB5009 in mainline OpenWrt is gonna take some time
<robimarko>
Gotta figure out SFP pinout anyway
<robimarko>
My breakout PCB should arrive soon for that
<Borromini>
robimarko: haha. yes. You're right about that :)
<Borromini>
you said SFP+ was working for you no? or just at 1G speeds?
<robimarko>
It works
<robimarko>
10G as well
<robimarko>
But its gotta be plugged in before booting
<Borromini>
ah.
<robimarko>
As since there are no GPIO-s for the various functions defined
<robimarko>
It defaults to a module being plugged
<robimarko>
And attempts to read the EEPROM
<robimarko>
So if its plugged in then it will decode it fine and work
<Borromini>
:)
GNUmoon has joined #openwrt-devel
<dangole>
robimarko: having routerboot do all the low-level init and then load U-Boot with proper NAND/UBI support sounds like a good plan :)
<robimarko>
Yes, that would be the easiest
<dangole>
robimarko: building U-Boot for mvebu should be pretty straight forward, I believe most of the things are upstream for a long time
<robimarko>
As the switch works fine in U-boot as well without having to port DSA ono MV88E6393X
<robimarko>
Yes, I have a working prototype already
<robimarko>
Since I had to upstream other Armadas, 3720, 7040, 8040 etc
<robimarko>
But, there is a rather big issue
<robimarko>
And that is that RouterBoot will only boot from NAND and it has to be inside of their new changed YAFFS
<robimarko>
This can be done from jailbroken ROS, you just replace the kernel ELF with U-boot
<robimarko>
But it means you cant really update U-boot easily
<robimarko>
And also, they messed up DRAM config in ATF
<robimarko>
Its reporting 2GB as they are using x16 DRAM and used x8 config
<robimarko>
And its also really old
bluew has joined #openwrt-devel
robimarko has quit [Quit: Page closed]
srslypascal has quit [Remote host closed the connection]
srslypascal has joined #openwrt-devel
<Slimey>
so like a 2nd stage bootloader thing :P
<Borromini>
wouldn't be the first OpenWrt device to do that i think :)
Borromini has quit [Quit: leaving]
huaracheguarache has quit [Quit: Page closed]
srslypascal has quit [Remote host closed the connection]
srslypascal has joined #openwrt-devel
srslypascal has quit [Remote host closed the connection]
srslypascal has joined #openwrt-devel
robimarko has joined #openwrt-devel
srslypascal has quit [Remote host closed the connection]
<robimarko>
Yeah, like a second stage bootloader
srslypascal has joined #openwrt-devel
<robimarko>
I think that AVM devices already use something similar
<robimarko>
Some Mediatek boards as well
<robimarko>
Thats not the issue, issue is how to deal with YAFFS where it must be placed in
<robimarko>
Cause MikroTik just started using UBI and UBIFS on IPQ40xx and then went back to YAFFS
* dangole
*facepalm*
<robimarko>
Yeah, they somehow just keep f*cking up
<robimarko>
I was kind of feeling enthusiastic to work on RB5009 cause of HW
<robimarko>
And HW is really good but man, how they managed to screw everything else up
<robimarko>
Also, just in case you want to replace ATF+Bootloader you gotta fit it into 700k unless you want to erase or move their soft/hard cofnig
<robimarko>
And that doesnt really fly as ATF alone is 670+k
<dangole>
robimarko: $220 for that hardware is also pretty cheap... all we need is a convenient exploit in ros which allows us to use it to write U-Boot (or maybe just SPL, so we can update U-Boot later without having to do that again...) to their yaffs mess.
<robimarko>
If thats the case than that already exists
<robimarko>
And its one exploit that they cannot fix
<robimarko>
Adron developed it when we were working on RB3011 and LHGG60ad
<dangole>
robimarko: why on earth is ATF so huge for that platform?! i'm used to 100kB for bl2 AND bl31... U-Boot can fit in 500kB...
<robimarko>
Well, there is a "secure" blob
<robimarko>
That implements PSCI that you have to embed into
<robimarko>
Thats 166k on its own
<dangole>
robimarko: ouf... and probably all we use PSCI for is to trigger reboot...
<robimarko>
And I also thought it will be smaller
<robimarko>
Maybe I am leaving DEBUG symbols or something but I dont think so
<dangole>
robimarko: you are sure it doesn't include OP-TEE as well? because then the size would kinda add up...
<robimarko>
I am not sure
<robimarko>
U-boot on its own is like 700k
<robimarko>
So total its around 1.5M currently
<robimarko>
Stupid thing is that MikroTik puts 16M NOR just for bootloaders and their config
<robimarko>
Everywhere, but they around 700k for the bootloader
<robimarko>
And then you end up with 15M of the NOR being empty
<robimarko>
That blob I mentioned is called: MSS Management SubSystem Firmware
<robimarko>
It probably does more than PSCI but I have not dug into it, looks to be FreeRTOS based
<dangole>
robimarko: U-Boot size can probably be explained by UEFI support now being enabled by default... just kick that out and you win 150kB...
<robimarko>
Sadly, I already disabled UEFI
<dangole>
robimarko: :(
<robimarko>
But there is tons of development or debug stuff enabled
<robimarko>
Thing is that I dont actually mind it being as big as needed
<robimarko>
I hate when basic stuff gets disabled or you have to worry about updating it anytime soo
<robimarko>
With 1GB of RAM and 16M of NOR I couldnt care less about couple hundred k
<robimarko>
*NAND not RAM
<robimarko>
Ill try to wrap up U-boot soon
<robimarko>
As well as OpenWrt once my SFP breakout comes
<robimarko>
Now that Serhii ported Amethyst support for 5.10 I might as well add it ASAP
<robimarko>
And not wait for 5.15
<robimarko>
Which I hope will come soon, I really want to upstream IPQ807x
robimarko has quit [Quit: Page closed]
cmonroe has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
noltari has quit [Quit: Bye ~ Happy Hacking!]
noltari has joined #openwrt-devel
rua has joined #openwrt-devel
cmonroe has quit [Ping timeout: 480 seconds]
Strykar has quit [Quit: /quit]
Tapper has quit [Ping timeout: 480 seconds]
guerby has quit [Remote host closed the connection]
guerby has joined #openwrt-devel
Piraty has joined #openwrt-devel
piraty_ has quit [Ping timeout: 480 seconds]
Strykar has joined #openwrt-devel
<hauke>
The OpenWrt 19.07 and 21.02 branches are looking fine to me and I would like to tag a new release on Tuesday evening
<hauke>
are there any objections?
<hauke>
anything missing?
<hauke>
jow: is LuCi fine in both branches?
<Thagabe>
@hauke: Fushia OS (nest hub old) is having issues with openwrt 21.02.1. Logs show good handshake and a correct assisnment of dhcp address. However, the device stays saying "no internet connection found" How would I be able to proceed to collect info for possible debug?
<hauke>
thagabe: you could use tcpdump to look at the traffic
<hauke>
probably the device is unable to connect to google
<Thagabe>
I'll do that.
<hauke>
no connection to google means no internet ;-)
<Thagabe>
But my question is, I didn't change dns settings or no AdBlock.