<pkgadd>
aside from the incorrect signed-off-by, there is no indication that this change really has an effect on the archer c2600. ipq806x is using cortex a15 since day one (so roughly 2016), it would (imho) need a whole lot more of a justification to change that after half a decade
zaolin has joined #openwrt-devel
lynxis has joined #openwrt-devel
hbug_ has quit [Ping timeout: 240 seconds]
<mangix>
pkgadd: huh that's funny. look at target/linux/ipq806x/patches-5.4/0065-arm-override-compiler-flags.patch
<mangix>
upstream kernel.org seems to disagree
nslu2-log_ has quit [Remote host closed the connection]
nslu2-log_ has joined #openwrt-devel
<pkgadd>
more verbose commit messages would certainly help, but that equally (respectively even more-) applies to that new PR
<pkgadd>
it would be interesting what lineageos is doing as well
<mangix>
pkgadd: fewer toolchains would be nice
<mangix>
even if there is no real problem with a15
<pkgadd>
mangix: armvirt is also using cortex-a15
<pkgadd>
sure, reducing the set of architectures would always be nice, but there should still be a justification in the commit message - that may very well be that cortex-a7 works well enough, but not that it "might eventually" fix the c2600 without clarifying if it really does
victhor has quit [Ping timeout: 258 seconds]
<pkgadd>
(I wouldn't assume that there'd be a significant performance difference for typical routing tasks either)
<mangix>
pkgadd: that armvirt seems to be wrong
<mangix>
given that it is not true with most devices
<mangix>
i would think qemu targets should be using something close to hardware
<pkgadd>
well, so far both targets have the justification that there's the other one also using cortex-a15 ;) I've never used armvirt and have no idea if using cortex-a15 might be faster there
tobleminer-tSYS has quit [Quit: AS4242423214]
tobleminer-tSYS has joined #openwrt-devel
<pkgadd>
interesting enough, ipq40xx was apparently using CPU_TYPE:=cortex-a15 && CPU_SUBTYPE:=neon-vfpv4 as well, without any ill effects
<pkgadd>
(while ipq40xx was still supported within the ipq806x target)
<lipnitsk>
seems to hide/bypass a bunch of modpost checks
<lipnitsk>
like: FATAL: modpost: sound/soc/ralink/snd-soc-ralink-i2s: struct of_device_id is not terminated with a NULL entry!
early has joined #openwrt-devel
<Grommish>
Silly question, but what does the build system mean when it returns "Nothing to be done for 'compile'"? I can see it not actually doing anything, even when I force the define with $(call Build/Compile/Default). it never calls the make invocation
<lipnitsk>
did you misspell the target?
<Grommish>
lipnitsk: I'll check, but I didn't see anything in particular
<lipnitsk>
paste your command
<Grommish>
make -j1 V=sc package/feeds/packages/libhtp/{clean,prepare,configure,compile}
koniu has quit [Remote host closed the connection]
<aparcar[m]>
mangix: please create new openssl patch which does the rename and the refresh at once
<aparcar[m]>
actually, please just use my genius script and I'll close the PR
<guidosarducci>
rsalvaterra: have you pinged jow? Not much I can personally do, and I'm loath to do cleanups in "security" components I don't fully understand anyway. Do your changes fix bugs? How are you testing it? Regressions? I'd probably be more a PITA than jow TBH. :^)
<rsalvaterra>
guidosarducci: Heh… just code consolidation and cleanups, nothing fancy… :)
<mangix>
aparcar[m]: what genius script?
<aparcar[m]>
in the patch description
<mangix>
maybe tomorrow. turns out these lua packages also have host patches...
<karlp>
behaving differently with quilt sounds like a recipe for explosions that down the road honestly.
koniu has joined #openwrt-devel
<mangix>
karlp: The good news is, it's only three packages in base and one in packages
<mangix>
OpenSSL will be gone with the 3.0 version update I believe
<rsalvaterra>
mangix: Phew… don't kill my Tor, please. :P
<mangix>
cotequeiroz handles the version updates not me.
<guidosarducci>
aparcar[m]: not sure it's meaning, but also seeing: "Failed to get security report information. Please reload the page or try again later."
<mangix>
aparcar[m]: seems that script can be simplified
<mangix>
while read PKG; make package/$PKG/refresh ; end < packages.txt
rr123_ has quit [Remote host closed the connection]
<Hauke>
mangix: I think it is not a problem when something does not work with gcc 7 any more in master or 21.02
<Hauke>
it is not the default since some time and from my point of view only usefull if you see a problem which could be compiler related to check what happens with the older compiler
<rsalvaterra>
Hm… And what's keeping us from bumping both gcc and binutils in master? :)
<Hauke>
rsalvaterra: yes should probably be done
<rsalvaterra>
FWIW, I'm on both latest versions for months.
<rsalvaterra>
Hauke: Wait. Actually, we need to merge mangix's musl bump before, as it contains gcc 10 build fixes.
<rsalvaterra>
In any case, the musl bump is also good to go, all known issues are fixed.
<Hauke>
I would like to have some weks between them
<Hauke>
mangix: rsalvaterra: are umdns and umbim the only applications you are aware of that do not compile with gcc 10?
<rsalvaterra>
Hauke: As far as I know, yes (I don't use any of them, though).
nitroshift has quit [Quit: Gone that way --->]
<rsalvaterra>
So, I'd say we merge the musl bump, at least to give it more exposure. There are only so many systems we can test on (ath79, ramips, mvebu and x86-64, in my case).
<rsalvaterra>
In a couple of weeks, we can do the toolchain bump.
<Hauke>
I would prefer to do the gcc 10 update first
<Hauke>
at least I would expect there less problems
<rsalvaterra>
In that case, you need to cherry-pick the fixes in mangix's musl pull, no?
koniu has quit [Remote host closed the connection]
koniu has joined #openwrt-devel
koniu has quit [Ping timeout: 240 seconds]
koniu has joined #openwrt-devel
valku has joined #openwrt-devel
jlsalvador has quit [Quit: jlsalvador]
Guest88867 has quit [Quit: WeeChat 2.9]
vdl has joined #openwrt-devel
fblaese_ has quit [Remote host closed the connection]
christiandr has quit [Quit: No Ping reply in 180 seconds.]
Hauke has quit [Ping timeout: 240 seconds]
SimJoSt has quit [Ping timeout: 240 seconds]
manawyrm has quit [Ping timeout: 258 seconds]
magnus2 has quit [Ping timeout: 258 seconds]
fblaese_ has joined #openwrt-devel
christiandr has joined #openwrt-devel
SimJoSt has joined #openwrt-devel
Hauke has joined #openwrt-devel
magnus2 has joined #openwrt-devel
yanosz has joined #openwrt-devel
caiortp has quit [Ping timeout: 265 seconds]
vdl has quit [Quit: WeeChat 2.9]
jlsalvador has joined #openwrt-devel
manawyrm has joined #openwrt-devel
vdl has joined #openwrt-devel
Tost has quit [Ping timeout: 252 seconds]
eduardas has quit [Quit: Konversation terminated!]
gromero_ has joined #openwrt-devel
gromero has quit [Ping timeout: 250 seconds]
<philipp64>
aparcar[m]: thanks for fixing the dependency issue... well, "fixing". yeah, something better is needed.
<philipp64>
that's a vexxing one.
sbrown has quit [Ping timeout: 260 seconds]
sbrown has joined #openwrt-devel
papaja has quit [Ping timeout: 240 seconds]
lukedashjr has joined #openwrt-devel
luke-jr has quit [Ping timeout: 250 seconds]
lukedashjr is now known as luke-jr
lukedashjr has joined #openwrt-devel
luke-jr has quit [Ping timeout: 265 seconds]
lukedashjr is now known as luke-jr
<Pepe>
aparcar[m]: Would it be possible that you will obtain commit access for openwrt-routing? There is no CI and integration with packages feed is stalled. Would be good to have there Github Actions as it is in packages and telephony feeds.
<mangix>
Hauke: yeah it's just umdns and umbim. The former patch was rejected by ynezz. The latter's not a real fix.
<mangix>
gdb and bpftools patches are for musl 1.2.
Nick_Lowe has joined #openwrt-devel
* enyc
meows Hauke
<enyc>
Hauke: alas irl building-works and stuff all getting in way, not got to homehub v5a (lantix xrx200 with ath9k+ath10k) debugging, but can say current 21.02SNAPSHOT crashing and rebooting one awy or another....
<enyc>
there was that issue with ath10k CT firmware etc
<enyc>
tho that just made the ath10k reboot, not crash whole system iirc
Nick_Lowe has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
bernrad has joined #openwrt-devel
<rr123>
really like the dashboard luci
sbrown has quit [Read error: Connection reset by peer]
<aparcar[m]>
guidosarducci: ping
<aparcar[m]>
rr123: same
sbrown has joined #openwrt-devel
koniu has quit [Remote host closed the connection]
<mangix>
jow: how many CPU cores to the buildbots have?
luke-jr has quit [Ping timeout: 265 seconds]
lukedashjr has joined #openwrt-devel
<rr123>
is build.config supposed to be in the release image or not
lukedashjr is now known as luke-jr
opal has quit [Remote host closed the connection]
opal has joined #openwrt-devel
lukedashjr has joined #openwrt-devel
luke-jr has quit [Ping timeout: 240 seconds]
lukedashjr is now known as luke-jr
<Borromini>
rr123: if you're talking the image you build, not unless you configure the buildroot so
<Borromini>
but you can conigure it so
<Borromini>
* configure
<aparcar[m]>
mangix: ususally 8 cores
<aparcar[m]>
rr123: your choice, there is an option for that
Borromini has quit [Quit: Lost terminal]
<rr123>
found it, thanks
opal has quit [Remote host closed the connection]
opal has joined #openwrt-devel
<mangix>
aparcar[m]: cool. Ninja will speed it up then
<mangix>
AFAIK, it builds with -j 1
<aparcar[m]>
What,m
<aparcar[m]>
How will it then speedup anything?
<mangix>
ninja doesn't care about -j parameters. It always builds in parallel
ianc has joined #openwrt-devel
_lore_ has quit [Read error: Connection reset by peer]
ianc has quit [Client Quit]
<mangix>
oh no it seems I ran out of disk space
<mangix>
WSL begone!
ianccp has joined #openwrt-devel
<JyZyXEL>
should i replace a 8M GigaDevice SPI flash with the 16M variant or a cheaper WinBond alternative?
_lore_ has joined #openwrt-devel
<ianccp>
The recent update of iproute2 to version 5.11.0 in commit b048a30 breaks compilation where libintl-full is enabled. The check for the new functionality which checks for libbpf in the package's configure script fails due to an unresolved symbol in libintl-full and compilation aborts. Given that it's a crucial part of openwrt, I thought I'd mention
<ianccp>
it here. If I submit a patch to the mailing list, past experience tells me that it will take a month before someone looks at it. I have the necessary patches and can post them here if someone wants to take a look?
valku has quit [Quit: valku]
<ldir>
ianccp: my suggestion would be to raise a github pr and poke @guidosarducci about it - Tony is interested in iproute2 stuff and very thorough.
<ianccp>
can't raise a PR on the main openwrt repo
<ianccp>
i'll send guido an email and let him deal with it - I've done the debugging already, so he just needs to send the patches to the list...
<ianccp>
thanks ldir
<ldir>
ianccp: yw - the trick is finding someone interested & with sufficient time/patience/skill - I'm interested but the last 3 are sorely lacking
swex has quit [Quit: swex]
rmilecki has quit [Ping timeout: 240 seconds]
netprince_ has quit [Read error: Connection reset by peer]
netprince_ has joined #openwrt-devel
swex has joined #openwrt-devel
zkrx has quit [Ping timeout: 240 seconds]
swex has quit [Quit: swex]
swex has joined #openwrt-devel
zkrx has joined #openwrt-devel
rsalvaterra1 has joined #openwrt-devel
rsalvaterra has quit [Ping timeout: 240 seconds]
ianccp has quit [Quit: Connection closed]
dedeckeh has quit [Quit: Connection closed]
black_an- has quit [Quit: simplicity does not kill]
<pkgadd>
philipp64: btw. I've been carrying this patch for strongswan for a long time https://paste.debian.net/hidden/8efb1017/ - just as a fyi, I haven't been using strongswan for almost a year now
<philipp64>
pkgadd: have you tried upstreaming it?
<pkgadd>
philipp64: no, I never got that far (having switched to a faster ipq8065 shortly after)
<Thermi>
ipsec starter is deprecated and getting removed in the 6.0.0 release
<Thermi>
together with ipsec stroke, ipsec.conf, and ipsec.secrets
<pkgadd>
o.k., that solves it as well ;)
<Thermi>
I'll see if I can spin up a different kind of loader-notifier for whatever openwrt is using
<Thermi>
shouldn't be too difficult
<karlp>
is there any "right" way of waiting for a network interface properly? bumping the START number like this person suggests doesnt seem safe/reliable: https://github.com/openwrt/packages/pull/15272
<Thermi>
The task is just to notify the init system when the daemon finished starting
<pkgadd>
but with many strongswan sub-packages installed, on a dead-slow router (well, ath79, so not thaaat slow) the timeout wasn't enough
<pkgadd>
the situation als became a lot better after I trimmed my installed strongswan package set to (almost-) the minimum I really used
<philipp64>
pkgadd: you could wait for PR 14708 to go it, and use swanctl instead... it groks *most* of the existing UCI settings (the ones that aren't wrong, anyway, i.e. not "left_firewall", etc.)
<Thermi>
pkgadd: looks like the issue is the router or its eMMC
<philipp64>
there are only two daemons that get started for strongswan, depending on whether you use ipsec or swanctl... so one really.
<pkgadd>
philipp64: I had to move over to wireguard, as I'm now on a cgNAT+IPv6 network - and the wireguard android app can deal with IPv6 better than the strongswan one (not really an issue with strongswan itself, but it nevertheless forced me to switch)
<Thermi>
pkgadd: Generally, you only need a very small set of plugins, so the amount of bytes read is quite limited
<Thermi>
pkgadd: with a newer Linux kernel, there's UDP encapsulation support for IPv6, so you could use that
<Thermi>
That was the primary issue with any roadwarriors
<pkgadd>
Thermi: interesting, thanks - I'll probably need to take another look
Radu-Mamy has joined #openwrt-devel
<Thermi>
pkgadd: using UDP encapsulation with IPv6 requires setting a check mark in the config for the VPN
<pkgadd>
VPN was the biggest question I had when I signed up with my current (cgNATed + IPv6) ISP - but fibre vs VDSL and using IPv6 for the incoming VPN connections helped me to settle
<jow>
karlp: uci managed network interface?
<Thermi>
Actually, we don't even need another wrapper around libcharon. We can just use a start-script entry to notify the daemon basically
<Thermi>
errr to notify the init system basically
<Thermi>
eh
<Thermi>
with procd, we can't even do that because the process needs to run in the foreground
ivanich has quit [Quit: Konversation terminated!]
<Thermi>
That makes it impossible to reliably chain other services to strongSwan on OpenWRT because it's not known when the daemon is ready
<Thermi>
That's a problem.
<Thermi>
There needs to be a readiness watch function or something. Like inotify for a pid file, or any file actually
<Thermi>
Like a socket
<Thermi>
A setting like "readinessobject" or something that can be a path to a file, or a socket address
<Thermi>
e.g. tcp://127.0.0.1:25 or something
<Thermi>
If it's a file, procd subscribes to an inotify for the directory is in and sets the service as having started once the file exists
<Thermi>
If it's a socket address, procd checks if the socket is connectable once a second