dgcampea has quit [Remote host closed the connection]
dgcampea has joined #openwrt-devel
ptudor has joined #openwrt-devel
maciekb has quit [Read error: Connection reset by peer]
maciekb has joined #openwrt-devel
stintel has quit [Ping timeout: 480 seconds]
robimarko has joined #openwrt-devel
Tapper has joined #openwrt-devel
djfe has quit [Read error: Connection reset by peer]
djfe has joined #openwrt-devel
G10h4ck has joined #openwrt-devel
danitool has joined #openwrt-devel
<PaulFertser>
djfe: I'd say the main question is how come mmc1 appears at all.
<PaulFertser>
djfe: it's not wifi-specific, it's apparently an issue with the mmc subsystem.
<PaulFertser>
Because why should it ever change at all?
<djfe>
because they switched the detection system and now it runs asynchronous
<djfe>
they are detected in different orders every reboot, it's random
BWhitten has quit [Ping timeout: 480 seconds]
<djfe>
the device uses mmc for more than just wifi, most of the time wifi is detected first (mmc0) but sometimes the flash (or whatever) is detected first
<PaulFertser>
djfe: are both eMMC and this brcmfmac really on the same MMC controller?
<djfe>
no clue, I don't own the device but this issue probably affects the whole sunxi target
<PaulFertser>
One 1c10000.mmc mmc host controller serving two different MMC devices, are you sure?
<djfe>
there are 1c0f000.mmc, 1c10000.mmc and 1c11000.mmc
<PaulFertser>
But how come mmc1 appears on 1c10000.mmc then?
<djfe>
because sometimes linux numbers 1c10000.mmc as mmc0 and sometimes as mmc1
<djfe>
mmc0 and mmc1 are assigned by order of detection and that is asynchronous
<djfe>
at least I think that's the gist
<PaulFertser>
djfe: so in other words it's the order of the detection of the host controllers themselves, not the mmc devices, do I get you right?
<PaulFertser>
djfe: you could just disable the other two controllers in the DT if you have nothing connected to it, and if you can't really connect anything to it.
<djfe>
I was wondering if we could fix this for openwrt, but the thing is: the dts file is Linux. So any change to the dts file would be a patch
<djfe>
there's got to be a way to detect this in userspace. I mean it could be done with a hotplug script. But I'm not sure that's the best approach.
<djfe>
1c10000.mmc should probably be enough to address the wifi chip. there's probably only one thing connected at any given time
Borromini has joined #openwrt-devel
<djfe>
weird enough that the path still contains the arbitrary number then
<PaulFertser>
djfe: I see your point.
<PaulFertser>
djfe: what's bad in using "option phy phy0" instead of option path for this usecase?
<PaulFertser>
djfe: alternatively, one could allow glob expansion in package/kernel/mac80211/files/lib/wifi/mac80211.sh in check_mac80211_device() to allow platform/soc/1c10000.mmc/mmc_host/mmc0/mmc0:0001/mmc0:0001:
<PaulFertser>
With mmc0 there becoming "mmc?"
<PaulFertser>
djfe: or you could use "option macaddr ..." to select the right phy instead of option phy and option path.
<PaulFertser>
So far it seems like option macaddr is the sanest way.
<djfe>
I'll check those out, thx for the ideas!
<dwfreed>
you'd need to patch mac80211.sh to not produce an /etc/config/wireless that uses path, though
<dwfreed>
xdarklight: that works for disks, not wifi devices that are accessed via mmc
<russell-->
you can make it deterministic by not using mmc connected wifi!
<G10h4ck>
I am trying to build hostapd with source tree override as explained here by Jow https://forum.archive.openwrt.org/viewtopic.php?id=46916 the problem is that I am getting an apparently unrelated compiler error even if I checkout the same identical commit used by the Makefile (without source overriding) as you can see here https://bpa.st/5H2LW wome suggestions on what is happening? I need this to have a reasonably workflow testing
<G10h4ck>
modifications on hostapd
<G10h4ck>
ping jow nbd
<PaulFertser>
G10h4ck: are you building with all OpenWrt patches applied?
cbeznea has joined #openwrt-devel
<PaulFertser>
G10h4ck: looks like not.
<PaulFertser>
They're in package/network/services/hostapd/patches/
<G10h4ck>
PaulFertser I tought the patches would be applied automatically, doesn't source tree override just use that dir as source instead of downloading it? or does it skip also the patches?
<PaulFertser>
G10h4ck: your log shows patches are not applied
<PaulFertser>
G10h4ck: probably you can force it if you do make QUILT=1 package/yourpackage/compile , not sure
cbeznea has quit [Read error: Connection reset by peer]
<PaulFertser>
Or probably just apply manually and proceed with your real target.
<G10h4ck>
PaulFertser seems I am getting the same error even with QUITL=1 :( https://bpa.st/XQSAY
stintel has joined #openwrt-devel
<PaulFertser>
G10h4ck: so apply manually
cbeznea has joined #openwrt-devel
<G10h4ck>
PaulFertser a suggestion on how to do that without breaking all the workflow? The Idea of give the source tree override a try is to improve the workflow so i don't need to push every change to a repository to test it
<PaulFertser>
G10h4ck: apply everything manually with "patch" or quilt, then add a temporary commit with all of that, work on top of it.
<G10h4ck>
ok thanks
noltari has quit [Quit: Bye ~ Happy Hacking!]
noltari has joined #openwrt-devel
Danct12 has quit [Quit: WeeChat 3.8]
goliath has joined #openwrt-devel
robimarko has quit [Remote host closed the connection]
robimarko has joined #openwrt-devel
<Borromini>
AFAIK pointing to an external tree does just that, it does not use OpenWrt's patches on top like the default build process would
BWhitten has joined #openwrt-devel
djfe has quit [Read error: Connection reset by peer]
<rmilecki>
is that stable? comprable for QSDK in performance, coverage?
[BWhitten] has joined #openwrt-devel
<[BWhitten]>
Hi all, is there any work going on for kernel 6.1 LTS? Looking at backporting for a platform and would like to see how far I need to bring it back, if I can avoid back to 5.15 it might make it easier.
<rmilecki>
[BWhitten]: i guess that once we switch all remaining targets to 5.15 and branch then we'll see 6.1
<[BWhitten]>
rmilecki: how is that looking presently?
<rmilecki>
[BWhitten]: last time i checked 2 targets only were at 5.10
<[BWhitten]>
rmilecki: okay sounds pretty close then
<[BWhitten]>
rmilecki: looks like omap and ath25
djfe has quit [Read error: Connection reset by peer]
djfe has joined #openwrt-devel
<f00b4r0>
ath25 is now source only so it shouldn't be a blocker
MaxS0niX has joined #openwrt-devel
Borromini has joined #openwrt-devel
MaxSoniX has quit [Read error: Connection reset by peer]
<Ansuel>
hgl btw working on nginx and been like 4 h on trying to understand why the damn thing cause recursive dependency when selecting luajit...
<hgl>
Ansuel: thanks for your work. I don't remember encountering recursive dependency, maybe a new challenge for a new version? In the mean time, I'm also trying the easy way out :P: https://github.com/openwrt/packages/pull/20849
<Ansuel>
as soon as you replace liblua with luajit everything explode -.-'''
<Ansuel>
and the cause it's luajit supporting only a few arch...
<hgl>
oh, I only checked x86, I guess that's why I didn't encounter compiling issues
<hgl>
Ansuel: stintel mentioned yesterday that LuCI is dropping lua in favor of ucode, and given the time you invested, I guess it's probably not worth it to continue this path.
<Ansuel>
sure but the lua module wasn't used for luci stuff either... my concern is for user that use it for other stuff
cbeznea has quit [Quit: Leaving.]
<hgl>
Ansuel: that's reasonable, but even if you manage to get it compiled, the lua module still won't be happy with the vanilla nginx it lives. I doubt many people use it for serious work.
<Ansuel>
this way dependency should be easier to handle...
<Ansuel>
(cause with the correct dependency and config the lua module can still be used)
<hurricos>
blocktrron: Will do. Thank you.
<Ansuel>
rmilecki perf are not comparable with qsdk.... they use wifi offload that can be supported but that require more work (and more qcom patch... only 2 patch needed to mac80211 and the rest to ath11k codebase)
<Ansuel>
with no wifi offload we reach 800mbps with ath11k
Gramdalf has joined #openwrt-devel
<Gramdalf>
Is anyone with write access able to review this pr? https://github.com/openwrt/openwrt/pull/3802 It would be nice to have it officially supported for my project (KlipperWRT-EX). More info is in the latest comment.
schwicht has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
[BWhitten] has quit [Remote host closed the connection]
schwicht has joined #openwrt-devel
BWhitten has quit [Ping timeout: 480 seconds]
schwicht has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
schwicht has joined #openwrt-devel
schwicht has quit []
schwicht has joined #openwrt-devel
valku has joined #openwrt-devel
schwicht has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
schwicht has joined #openwrt-devel
schwicht has quit [Read error: Connection reset by peer]
schwicht_ has joined #openwrt-devel
schwicht_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
schwicht has joined #openwrt-devel
schwicht has quit []
schwicht has joined #openwrt-devel
rsalvaterra has quit []
rsalvaterra has joined #openwrt-devel
minimal has quit [Quit: Leaving]
schwicht has quit [Read error: Connection reset by peer]
schwicht has joined #openwrt-devel
madwoota has quit [Remote host closed the connection]
danitool has quit [Ping timeout: 480 seconds]
madwoota has joined #openwrt-devel
madwoota is now known as Guest11693
Borromini has quit [Quit: Lost terminal]
MaxS0niX has quit [Quit: Konversation terminated!]
cbeznea has joined #openwrt-devel
cmonroe has quit [Read error: Connection reset by peer]
cmonroe has joined #openwrt-devel
schwicht has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Ansuel>
hgl btw finally manage to find time to implement and test the dynamic module and works correctly
<Ansuel>
and that indirectly solve the dependency hell
<Ansuel>
:D
cmonroe has quit [Ping timeout: 480 seconds]
<rmilecki>
Ansuel: i've zero experience with nginx, can't help you with that i'm afraid
<rmilecki>
Ansuel: ah, nevermind, i thought you pinged me about pull 16842 (nginx)
<rmilecki>
Ansuel: thanks a lot for ath11k iinfo
<Ansuel>
if you want some raw performance with the offload you can reach 3gbps but that results in high ram consumption so 1gb is required
schwicht has joined #openwrt-devel
Ryncewynd has joined #openwrt-devel
BWhitten has joined #openwrt-devel
<slh>
rmilecki, Ansuel: in practice, with HE80 and 2x2 clients (ax200, ax210, mt7921au) you rarely see more than 750-800 MBit/s with proprietary QCA APs either (nor any real difference on 6 GHz); 4x4 clients would be another topic