<ldir>
ynezz: It is native mac OS/hardware. No CI involved.
goliath has joined #openwrt-devel
<ynezz>
ldir: ok, so now it gets more interesting :p
<ynezz>
f00b4r0: I simply cancel workflows which are not need, saving some build time, so in this case I've kept just the last workflow for the tag
<ynezz>
aparcar[m]: you've already the highest permissions in gitolite, so I'm wondering what is blocking you
<ynezz>
aparcar[m]: Re: https://github.com/openwrt/buildbot/pull/10 its not clear to me what are your overall plans with this renaming, is it only about openwrt repository or you're going to rename master->main in all the repositories?
<ynezz>
aparcar[m]: if the later, then we should probably consider doing this in one go
<ynezz>
and I would prefer to prepare some issue with clear TODO tasks (so everyone is aware and can provide feedback) and do it once we finish the buildbot migration and 23.05.0-rc1 preparation
Lynx- has joined #openwrt-devel
robimarko has joined #openwrt-devel
dangole_ has joined #openwrt-devel
<f00b4r0>
ynezz: btw did you see hnyman's email earlier? Something's odd with g.o.o buildbot repo
<f00b4r0>
s/email/PR#9 comment/
<ynezz>
yes, thats the broken syncing, should be fixed now
<f00b4r0>
👍
mrnuke has quit [Remote host closed the connection]
robimarko has quit [Ping timeout: 480 seconds]
<ynezz>
does anyone have an idea how to install specific version of Make on macOS using brew?
<ynezz>
it uses version 4.4.1 as default and I would like to try out build with make 4.3, so tried make@4.3 but that didn't worked
<ynezz>
Warning: No available formula with the name "make@4.3". Did you mean make?
* f00b4r0
stays away from brew, can't help, sorry
fakuivan has quit [Ping timeout: 480 seconds]
<ynezz>
well, I don't mind to use macports/pkgsrc if that would allow me to try out different versions
<Lynx->
so as far as I'm aware it's not possible to restore DSCP from conntrack as in the tc call "action ctinfo dscp 63 128", so a combination of nftables and tc is needed, but right now I don't see the issue with that.
<Lynx->
(I mean this is for those cases where you want DSCP reapplied on ingress based on the DSCP set on egress)
<mrkiko>
lynxis: hi! didn't hear back from you...
bluew has quit [Ping timeout: 480 seconds]
<djfe>
what happens to PRs against master?
<djfe>
(since we switched to the name "main")
<djfe>
can openwrt bot handle this or do they all have to be merged manually in this case?
<oliv3r[m]>
Ansuel: any idea how to fix: `error: passing argument 1 of '__builtin_memcpy' discards 'const' qualifier from pointer target type memcpy(dev->dev_addr, addr->sa_data, ETH_ALEN);`? I get that dev_addr is const, but what's the point of having this in the netdev struct, if we can't modify it.
<oliv3r[m]>
d07b26f5bbea9ade34dfd6abea7b3ca056c03cd1 might be the answer to that ...
tmn505 has quit [Remote host closed the connection]
<mrkiko>
robimarko: eheheh! so happy qca-ssdk is compilable easily with make 4.4
<robimarko>
mrkiko: Well, I kinda made sense of some upstream commits and it eventually worked
FLD has quit [Ping timeout: 480 seconds]
FLD has joined #openwrt-devel
minimal has joined #openwrt-devel
<mrkiko>
robimarko: nice that qca develops ssdk openly
<robimarko>
mrkiko: well its GPL anyway so they just keep everything in one place
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
<mrkiko>
robimarko: nice to note that once switched to 6.1, ipq807x will be much much smaller in patches footprint, if I am looking at things correctly
<robimarko>
mrkiko: Yes, as a lot of stuff was upstreamed
<robimarko>
Remoteprocessor is currently broken so no WLAN on 6.1, I am looking for a fix instead of reverting 9 or 10 patches
<mrkiko>
robimarko: the remaining things will be network, qualcomm CR regulators, Qualcomm APm stuff, remoteproc patches... these are the biggest I see
<robimarko>
mrkiko: There was multiple revisions of CPRv4 upstream support but none landed yet, so there is hope for that
<robimarko>
Remoteproc is also feasible, but wired networking I dont see happening in future as there are no docs
dangole has joined #openwrt-devel
<mrkiko>
robimarko: but ok, IPQ807X is positioned better than I initially tought. As for the wired netowkring - is it handled inside SSDK or are there other pieces?
tmn505 has joined #openwrt-devel
<rmilecki>
Ansuel: patch I sent was for upstream
<rmilecki>
Ansuel: i just want nbd's comment on it
<rmilecki>
robimarko: is hardware offloading an option with some of QSDK's code?
<robimarko>
mrkiko: Well, it took a lot of time to upstream and I need to resume. Wired networking is split into SSDK for the switch and PHY and NSS-DP for ethernet controller
<robimarko>
rmilecki: Yes, in theory we can just use the NSS-DRV which is the core driver and hook into flow tables
<robimarko>
Currently, performance is quite limited as NSS-DP for EDMAv1 (IPQ807x and IPQ60xx) doesnt even have basic offloads like checksums etc
<mrkiko>
is nss firmware reddistributable my any means?
<mrkiko>
robimarko: thanks
<robimarko>
mrkiko: Yes, QCA even has it publicly on their Github, though its not the latest one
<robimarko>
Its even possible to push the traffic via NSS FW to utilize basic offloads like checksums and TSO
<robimarko>
Ansuel even ported it to use the generic DMA API
<mrkiko>
oh, I didn't imagine there was some WIP on this. Thaks.
<robimarko>
Well, its not really WIP but rather previous work so it "works"
<robimarko>
But it needs more love
<robimarko>
SoC also has an EIP197 crypto engine, it seems to be partially compatible with the upstream driver but its completely lacking interrupt support it seems
<mrkiko>
:D :D
<robimarko>
Remoteproc is driving me nuts, commit which seemingly has no relation to it is breaking it but I dont see how
<robimarko>
SCM call just magically starts failling after that commit
<djfe_>
schmars[m]: are the maintainers able to do that? I'm not talking about my own PRs but the 300 out there
f0ghua has joined #openwrt-devel
<f0ghua>
In order to understand the usage of 'reload_config
minimal has quit [Quit: Leaving]
<f0ghua>
In order to understand the usage of 'reload_config', I searched the git log of openwrt and found that it comes from this change 'procd: add a small script that handles config reloads until configd is ready'. Does anyone know what 'configd' is? I didn't find the package on openwrt source tree.
<Ansuel>
djfe_ it's a no problem... we don't use github to merge pr... and master and main will be synced currently in this transition phase
<Ansuel>
new pr will target main automatically
<djfe_>
ok great
<djfe_>
I'll read the logs if others answered. I still got to figure out why quassel can't handle relogging for my second login when my ip get's invalidated by my isp
<djfe_>
I registered djfe_ and linked it but it still didn't login
<robimarko>
I have the same issue
f0ghua1 has joined #openwrt-devel
<robimarko>
Plus hexchat despite having SASL setup tries to join the channel before even registering the nickname
<f0ghua1>
Sorry, I was dropped
<f0ghua1>
repeat the question. In order to understand the usage of 'reload_config', I searched the git log of openwrt and found that it comes from this change 'procd: add a small script that handles config reloads until configd is ready'. Does anyone know what 'configd' is? I didn't find the package on openwrt source tree.
f0ghua has quit [Ping timeout: 480 seconds]
<djfe_>
robimarko: afaik this irc doesn't support sasl
shibboleth has quit [Remote host closed the connection]
shibboleth has joined #openwrt-devel
FLD has joined #openwrt-devel
Borromini has quit [Ping timeout: 480 seconds]
[Pokey] has joined #openwrt-devel
<[Pokey]>
Hey OpenWrt, wondered if anyone here has played with an RTL8812BU with OpenWrt yet? If not, I have found a few source packages with a Makefile and patches from other people's OpenWrt builds, and I wondered how hard (if at all possible) it would be to compile that for the existing official builds?
<[Pokey]>
All I need is to connect to a WiFi network with this adapter and then connect a device through an ethernet port and get bidirectional network access through any means
<Borromini>
date suggests it's ancient as well.
<PaulFertser>
[Pokey]: then route a subnet as usual, please consider refraining from NAT :)
floof58 has quit [Remote host closed the connection]
<[Pokey]>
If I can do that I will do that
<[Pokey]>
I want to avoid NAT too
<[Pokey]>
but I will take whatever works
<PaulFertser>
[Pokey]: 802.11s can be considered, apparently this realtek driver supports it.
<PaulFertser>
Znevna: problem is that people often see this word but do not (and often actually can not) understand what it really implies because over they years many different vendors used it to mean very different incompatible things.
<Znevna>
wds client was used, and LuCI present an option called exactly that "Client (WDS)"
<xdarklight>
[Pokey]: I suggest checking the upstream rtw88 driver. at around Linux 6.2 it gained USB support, including RTL8822BU
<Znevna>
presents*
<[Pokey]>
xdarklight: I'm using RTL8812BU
<PaulFertser>
Znevna: the problem is that OpenWrt means one thing there, and many other vendors mean another. Including Linux the kernel which used to have a special "WDS" interface mode (and it was shown as such by "iw" and other nl80211 tools) which is different and incompatible.
<xdarklight>
[Pokey]: ah, I missed that
<xdarklight>
[Pokey]: RTL8812BU is not supported by rtw88 as far as I know
<Znevna>
isn't that just the crappy single stream version of 8822bu?
<[Pokey]>
So I selected it as a kernel module in menuconfig, do I need to build the entire thing or can I just build that one module?
<[Pokey]>
I would preferrably like to end up with a .ipk
floof58 has quit [Read error: Connection reset by peer]
floof58 has joined #openwrt-devel
<[Pokey]>
I ended up doing a full make.... we'll see how that goes
<PaulFertser>
Znevna: I'm never just saying that. I always clarify about the hostapd "wds_sta" option meaning 4addr mode and that OpenWrt "option wds" refers specifically to that.
<PaulFertser>
[Pokey]: btw, you'll have to run images that you build yourself, not just the additional kmod packages. Unless you take special care and follow special procedure to reproduce buildbot config so that the kernel .config hash matches.
<[Pokey]>
PaulFertser: I feared this. Would I just be able to take buildbot's .config and just enable my option on top of it without changing the magic?
<PaulFertser>
[Pokey]: for a wireless driver, probably, since it's built against backports.