<stintel>
jow: can I access parse_subnet in parse_ipsetentry somehow ?
<stintel>
nvm
JiiPee has quit [Quit: leaving]
JiiPee has joined #openwrt-devel
hanetzer has quit [Quit: WeeChat 3.3]
hanetzer has joined #openwrt-devel
goliath has quit [Quit: SIGSEGV]
kenny has joined #openwrt-devel
<will[m]>
when sysupgrade runs "ubus call system sysupgrade" and the json says "command": "\/lib\/upgrade\/do_stage2" and it comes back with "Command failed: Not found" what should i be looking for? the do_stage2 script is definitely there...
Ansuel has joined #openwrt-devel
Ansuel_ has quit [Ping timeout: 480 seconds]
kenny has quit [Ping timeout: 480 seconds]
Ansuel has quit [Ping timeout: 480 seconds]
strobo_ has joined #openwrt-devel
victhor has quit [Ping timeout: 480 seconds]
strobo has quit [Ping timeout: 480 seconds]
Tapper has joined #openwrt-devel
Tapper has quit [Read error: Connection reset by peer]
Tapper has joined #openwrt-devel
danitool_ has quit [Ping timeout: 480 seconds]
musashino has joined #openwrt-devel
Andi_ has quit [Read error: Connection reset by peer]
Andi_ has joined #openwrt-devel
<dwfreed>
will[m]: I imagine the escapes are the problem
<dwfreed>
like why are the slashes escaped
rua has quit [Quit: Leaving.]
floof58 has quit [Remote host closed the connection]
dangole_ has quit [Remote host closed the connection]
dangole_ has joined #openwrt-devel
dangole_ has quit [Remote host closed the connection]
dangole_ has joined #openwrt-devel
<stintel>
gahhh fscking Makefile
<stintel>
useless fscking errors all the time
enyc has joined #openwrt-devel
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
kenny has joined #openwrt-devel
Andi_ has quit [Read error: Connection reset by peer]
Andi_ has joined #openwrt-devel
goliath has joined #openwrt-devel
Tapper has quit [Ping timeout: 480 seconds]
robimarko has joined #openwrt-devel
<robimarko>
stintel: Whats the issue with 5.15? Its not gonna be used in the next release anyway
<robimarko>
I for one am kind of desperate for it as trying to backport everything is just wasting time
<stintel>
robimarko: we agreed not to merge that, before branching next release (see meeting notes september), if that takes too long, we can talk again
<stintel>
but adding it now will result in people seeing it and spending time on it - time that should rather be spent on things that are currently more important
<robimarko>
stintel: Since I am not one of the commits I dont really track meeting notes
<stintel>
I linked them earlier ;)
<robimarko>
Whats the desired priority now?
<stintel>
next release will have 5.10, focus on moving what is not on 5.10 to 5.10 and fixing what doesn't work
<stintel>
I'm working on firewall4 atm
<robimarko>
I know that it will use 5.10, but I dont see whats so broken currently
<stintel>
as that should be default in the next release too
<robimarko>
I assume that the next release is planned to be branched relatively soon then?
<stintel>
planned might be a stretch, let's call it aim :)
<robimarko>
I understand, been here long enough to see the release cadence
<stintel>
;)
<stintel>
but if we can avoid traps like merging 5.15 support now, maybe we can improve
<robimarko>
My honest opinion is that it wont change anything
<stintel>
well people can disagree
<stintel>
I've seen it happen before, new kernel and everybody drops everything and starts porting whatever they use to it
<robimarko>
Sure, that is why its an opinion
<robimarko>
Well, you gotta understand that we are in backporting hell now
<robimarko>
The diff is quite big for anything
<stintel>
it will be like that for the next release anyway, so that's a moot point
<robimarko>
Why?
<stintel>
because next release will use kernel 5.10
<robimarko>
I know that, but thing for example about new stuff?
<robimarko>
Like IPQ807x
<robimarko>
Both MHI and QRTR that it requires are broken in 5.10
<robimarko>
And that is just the basics to get ath11k working
<robimarko>
Even worse all of the PCI cards require MHI and QRTR as well
<stintel>
I doubt even with 5.15 ipq807x could go into a usable shape
<robimarko>
Well, its relatively usable even on 5.10 now
<robimarko>
The wired networking is the issue
<robimarko>
And the fact that I have to do things twice currently to upstream anything
Luke-Jr has quit [Ping timeout: 480 seconds]
<robimarko>
Not to mention that trying to merge anything into IPQ40xx takes way too much
<robimarko>
As there are no reviewers for the target
<robimarko>
But I can tell you that firewall4 looks really interesting
<stintel>
making firewall4 default is going to cause some breakage
<stintel>
that should probably one of the focus areas
<stintel>
I'm almost ready with the firewall4 and luci part, struggling with miniupnpd Makefile
<stintel>
if you want to build firewall4 with my pending work you can use my firewall4 branch in my staging tree: https://git.openwrt.org/?p=openwrt/staging/stintel.git;a=shortlog;h=refs/heads/firewall4
<robimarko>
Thanks, will probably give it a go next week
<stintel>
cool, looking forward to some feedback
<stintel>
I'm using it on my main router since ~12h or so
<robimarko>
Thats the best way of testing for sure
<stintel>
yeah, having a redundant setup helps to do that
<stintel>
holy crap, I was just searching some irclogs ...
<stintel>
got reminded of hojuruku
<stintel>
hauke: gdb changes don't fix my issue
<aparcar[m]>
Hauke where?
goliath has joined #openwrt-devel
<hauke>
aparcar[m]: I asked Johannes in IRC ;-)
<hauke>
stintel: hmm not good
<stintel>
hauke: did you have any feedback on the KERNEL_{KA,UB}SAN patches I sent earlier ?
<hauke>
stintel: let me have a look
<enyc>
stintel: I understand, just ... might need assist in sorting out the ... commits to select ;p even if I set up the git tree locally and so-on, wouldn't necessary ly know which commits to pick
kenny has joined #openwrt-devel
<will[m]>
dwfreed: I can check on that but it's passed in via quoted CLI so I think slashes have to be escaped and also even quotes though I think my chat client unescaped some of them. I'd have to recompile ubus to see what actually makes it in, but I haven't changed this part of the system so I'd expect the slashes to be correct
T-Bone has joined #openwrt-devel
f00b4r0 has quit [Ping timeout: 480 seconds]
tomn has joined #openwrt-devel
shibboleth has joined #openwrt-devel
Borromini has quit [Quit: Lost terminal]
dangole has joined #openwrt-devel
dangole_ has quit [Ping timeout: 480 seconds]
ecloud has quit [Ping timeout: 480 seconds]
<Tapper>
hauke about saving space. What about letting dnsmasq do dhcp? Then you could drop odhcp and odhcpv6.
<Tapper>
I don't know how mutch space that would save.
ecloud has joined #openwrt-devel
Ansuel has quit [Quit: Probably my PC crashed or time to sleep.]
<shibboleth>
Tapper, building without luci and odhcp* will free up some space
<Tapper>
shibboleth I do remove odhcp but I need luci. I was talking about relece builds of OpenWrt tho.
<shibboleth>
ah, yeah... 5.15, luci, odhcp* will mean goodbye to 8mb devices
<shibboleth>
also, devs with funky kernel part sizes
<Tapper>
odhcpd is only 46kB
<Tapper>
odhcpd ip6 is just 40kB so that is 86KB saveings.
<Tapper>
Inside my lan dnsmasq does just fine without odhcpd and odhcpd ipv6, but I dont have ipv6 on my wan
ecloud has quit [Ping timeout: 480 seconds]
ecloud has joined #openwrt-devel
minimal has quit []
kenny has quit [Ping timeout: 480 seconds]
kenny has joined #openwrt-devel
ecloud has quit [Ping timeout: 480 seconds]
ecloud has joined #openwrt-devel
Andi_ has quit [Read error: Connection reset by peer]
Andi_ has joined #openwrt-devel
<JiiPee>
hauke: Could you point me for example of splitting kernel on devices what have size limitation?
rmilecki has quit [Ping timeout: 480 seconds]
<hurricos>
Is it possible to package particular packages along with a sysupgrade?
<Habbie>
yes
<hurricos>
I think the ImageBuilder is probably the best way to do it, but I actually don't know what side-effects that has
<Habbie>
hurricos, what kind of side effects are you thinking of?
<hurricos>
Habbie: Correct me if I'm wrong, rqworker is limited to effectively using the ImageBuilder, yes?
<Habbie>
i have that impression, yes
dedeckeh has quit [Remote host closed the connection]
<hurricos>
Not just rqworker, but I'm guessing the premise of the API is, "add packages [and potentially configuration] to an imagebuilder tarball and build"
<Habbie>
i guess
<hurricos>
... Which is OK, actually, probably better workflow than I have.
<Habbie>
i don't think it adds configuration to the image(builder)
<hurricos>
ah
<hurricos>
The problem then is if I have to override a package's configuration to avoid a crash, then I'm stuck :^)
<Habbie>
override in what way?
<hurricos>
MR24 with an ath10k card swapped in will kernel panic if ath10k_pci is loaded without irq_mode=1
<hurricos>
but I should just throw that away or PR a fix if I have one.
<hurricos>
The former really, Openwrt doesn't support modified hardware outside of x86 if I recall policy correctly
robimarko has quit [Quit: Page closed]
<hurricos>
ASU is perfect for more realistic uses though, like: I just need to upgrade a mesh node that is connecting a printer; just have to make sure wpad-mesh-wolfssl is installed so it rejoins the 802.11s mesh after
<hurricos>
thank you Habbie :)
dangole has quit [Ping timeout: 480 seconds]
<Habbie>
hurricos, np!
<Habbie>
hurricos, i was amazed by this tool when i found it :)
<Habbie>
hurricos, modified hardware as in, you physically modded some device?
Ansuel has joined #openwrt-devel
<hurricos>
Habbie: I think so. I don't think swapping wireless cards on device tree targets is really supportable
<Habbie>
right
<hurricos>
For obvious reasons ...
<Habbie>
sure
<dwfreed>
you could build your image to accomplish this
<Habbie>
yes
<dwfreed>
you just can't use the existing imagebuilder services
<Habbie>
in which case you'd have to set up your own asu too
<dwfreed>
I mean, you don't have to use asu for it if you do it manually
<Habbie>
sure, but it sounds like asu really hits the spot for hurricos outside of that
<dwfreed>
asu can make automating it easier
<hurricos>
So, there's the question of drawbacks: when I build packages into an image, <*> will put them in the kernel's space / initramfs, whereas <M> will put them in the image itself, is that right?
<hurricos>
using the full build system, when using menuconfig
<Habbie>
i ended up not using asu; turns out i somewhat misunderstood its purpose at first, and my problem is building openwrt packages from our git master :)
<Habbie>
hurricos, i thought M just built the packages and not included them in anything
<hurricos>
oooh ....
<dwfreed>
and * for packages would put them in the root image, not the kernel
<hurricos>
I think the initramfs picks them up though
<hurricos>
even if the root kernel does not. It's been months since I last tinkered with building additional packages into images, I probably should have been using the imgae builder all along
<Habbie>
i was wondering how openwrt packages handle architectures
<Habbie>
should opkg list, install, etc. warn the user that they have no indexes?
<slh>
Tapper: odhcp6c is a dhcpv6 client, not a server - dnsmasq can't replace that. odhcpd-ipv6only still has the edge (quite massively even) when it comes to dynamic prefix changes, that's where dnsmasq is 'not good' to say the least - and don't forget, currently dnsmasq is not compiled with IPv6 feature, if it were, the package would grow, so the math doesn't quite work out as easy as you make it to be
jlsalvador has quit [Quit: jlsalvador]
<slh>
Tapper: in a way it would be more likely to replace dnsmasq with the full odhcpd and a slightly hacked up umdns package than the other way round