danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
Rentong has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
_ppp has quit []
Borromini has joined #openwrt-devel
Rentong has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
_lore_- has joined #openwrt-devel
_lore_ has quit [Ping timeout: 480 seconds]
<aiyion>
Good morning.
<aiyion>
My build process produces what appears to be a legacy uimage and calls it squashfs sysupgrade; a factory image is not produced. Any ideas why the file is not a "firmware 470 v1 OpenWrt"?
danitool has joined #openwrt-devel
rsalvaterra has quit [Quit: Leaving...]
aleasto has joined #openwrt-devel
arifre has joined #openwrt-devel
dorf has quit [Remote host closed the connection]
dorf has joined #openwrt-devel
<PaulFertser>
aiyion: squashfs sysupgrade usually includes uimage prepended, correct
<PaulFertser>
aiyion: factory image is generated when Makefile specifies a special procedure to make vendor firmware happy accepting it.
Rentong has joined #openwrt-devel
StifflersMagic has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
<stintel>
mangix: don't we all :)
_lore_ has joined #openwrt-devel
_lore_- has quit [Ping timeout: 480 seconds]
Borromini has quit [Quit: Lost terminal]
<PaulFertser>
stintel: do you have any mt7621a hardware in use by any chance?
<PaulFertser>
stintel: it looks like gmac1 (the second integrated GMAC) can be connected "directly" either to port 0 or to port 4 which can improve typical wan-lan routing performance (as currently all the boards are pumping everything through GMAC0 alone).
<PaulFertser>
For anyone interested, the upstream Linux commit is 38f790a805609b255f0cacc097e2df4324a3d041 and basically it should be enough to enable gmac1 in board DTS and to add in there phy-handle = <&...>; to refer to port0 or port4 of the switch (define them on the mdio bus), make sure port5 is set to disabled.
<PaulFertser>
(it has a full example of using port 0 or port 4 as a "wan")
mwarning has joined #openwrt-devel
<mwarning>
hi, I like to start a program when some wifi interfaces come up. I think I need to add a netifd protocol for that. Is there an example around I can look at or documentation on how to do this?
<PaulFertser>
mwarning: you do not need a protocol, there's /etc/hotplug.d/ directory where you can add your hooks.
Rentong has quit [Remote host closed the connection]
rmilecki has joined #openwrt-devel
dedeckeh has joined #openwrt-devel
Rentong has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
aleasto has quit [Ping timeout: 480 seconds]
aleasto has joined #openwrt-devel
philipp64|work has quit [Quit: philipp64|work]
philipp64 has quit [Quit: philipp64]
<fda>
dangole_: a quick test to add a warning at top of overview: https://pastebin.com/k3npfYAC i just dont know how to get the pstore files. could it geb by "ubus"?
<fda>
in case of a crash / reboot (to recover) i did not know that i should check there
dedeckeh has quit [Remote host closed the connection]
<dangole_>
fda: you can use ubus 'file' object which is offered by rpcd and already used by other luci apps. ie. ubus call file list '{"path":"/sys/fs/pstore/"}'
<dangole_>
fda: nice, please submit to github.com/openwrt/luci.git and discuss details there. imho would be nice to be able to click on individual crashlogs to view (and maybe discard) them inside the web ui.
<fda>
if the content of the files are visible by luci, youll get only screenshots in the forum. maybe missing most important parts. and additional i dont know yet how i could show contents in a new windows etc :)
<dangole_>
fda: naiive me would say to just create another view which is for viewing pstore logs, just like the existing views to show system and kernel logs
<fda>
dangole_: ill check later. currently (since yesterday...) im trying to build autofs, but im a noob and dont get some includes correctly :((
<stintel>
interesting, it seems to be snmpd that triggers it
<stintel>
PaulFertser: not sure about mt7621a
PaulFertser has quit [Ping timeout: 480 seconds]
<stintel>
meh :)
<stintel>
timing!
<stintel>
ah, actually I do, GL-iNet GL-MT1300 seems to be MT7621A
PaulFertser has joined #openwrt-devel
<stintel>
PaulFertser: actually I do have an MT7621A device
shibboleth has quit [Quit: shibboleth]
<PaulFertser>
stintel: looks like for all the currently supported boards OpenWrt would limit routing to 500 Mbit/s because it uses just a single gmac.
<Slimey>
stintel no luck here booting the t1023 target from the m300 image just hung on loading kernel
<Slimey>
ill likely have to try from scratch :P
<mangix>
fda: patch it
<mangix>
anyone running ar71xx?
<stintel>
Slimey: m300 image wont boot on t1023
<stintel>
Slimey: m300 = t2081 = e6500
<stintel>
Slimey: t1023 is e5500
<Slimey>
heh maybe i miss read
<dangole_>
fda: regarding autofs: we got our own userspace implementation called 'blockd', I've just fixed a lot of things there and tested excessively on 32-bit and 64-bit targets, works fine (now).
<stintel>
if only that thing would support cifs
<mangix>
blockd doesn't support CIFS?
<stintel>
nope
<mangix>
that's unfortunate
<mangix>
would adapting https://git.openwrt.org/?p=project/fstools.git;a=blobdiff;f=block.c;h=2e3841ab15554b2b7980b54e3f7f62ddb51adb32;hp=b377429c61f5ab19d88a18fc25ebe2744eba6d80;hb=29e53af0b91954ca6a13bdd1f0e67c939ccb8e6d;hpb=dd02dad332958575cc52e60c78ef10ef2b5aaced to cifs work?
<fda>
autofs can also moutn more things like davfs2 (owncloud) or things like curlftpfs
<fda>
but my main reason: mount if there is a file is accessed, and unmount if no longer needed
<fda>
@dangole_ blockd does onyl mount at boot, or maybe some hotplug events if i read it correctly
<mangix>
fda: __WORDSIZE is definitely defined
<mangix>
probably a missing include to sys/reg.h
<mangix>
__WORDSIZE is a low level define though. I don't think it should be used.
<fda>
@mangix its used in autofs. i read somewhere it does not exist with musl. but i have not much experience with compiling
<mangix>
probably old info
<mangix>
git grep WORDSIZ | wc -l
<mangix>
42
<fda>
that is possible. but i get only 4 hits by "grep __WORDSIZE . -r" in a fresh clone
<fda>
cant unpack atm as i have hight cpu load...
<mangix>
for musl?
<mangix>
that result is from musl's git repo
<stintel>
looks like ppc64be does not like hardened usercopy
<fda>
@okay, 1st line in reg.h
<fda>
then the include is wrong or missing
rmilecki has quit [Ping timeout: 480 seconds]
<fda>
mangix: you know where _PATH_NSSWITCH_CONF is defined? this is also missing. (but not in musl)
<fda>
i just hardcoded a file -.-
<mangix>
never seen that before
aleasto has quit [Quit: Konversation terminated!]
<dangole_>
fda: block-mount does mount at boot, blockd does autofs-based automount as you described for local storage. support for network fs would be not very hard to add.
<dangole_>
fda: ie. blockd also does umount after inactivity timeout and mount on access. i use it heavily for 'uvol' which simplifies managing additional local storage volumes on devices with larger NAND flash or eMMC.
mangix_ has joined #openwrt-devel
mangix has quit [Ping timeout: 480 seconds]
<stintel>
Sun Aug 1 02:08:28 2021 daemon.debug dnsmasq-script[1]: /usr/lib/dnsmasq/dhcp-script.sh: .: line 5: can't open '/usr/share/libubox/jshn.sh': No such file or directory
<stintel>
what is this now
<stintel>
ujail again I guess
philipp64 has joined #openwrt-devel
<philipp64>
I just sysupgraded to master today and... it blew away all of my configuration. Hours of it.
<dangole_>
stintel: i guess so :/ script dependencies are not automatically detected by procd_add_jail_mount, so adding only $DHCPSCRIPT and not adding it's dependency probably never worked...
<philipp64>
is this a known issue?
<philipp64>
This was on a Xeon x86_64 platform.
<fda>
a known issue is that dir-premissions while restoreing are messed
<dangole_>
philipp64: i experienced a similar issue lately and the cause was kmod-fs-vfat not being installed
<fda>
eg 700 and then dnsmasq cant read from this dir anymore
<dangole_>
philipp64: because config backup on x86 gets stashed in a tarball on /boot during sysupgrade, and on EFI platforms that vfat...
<fda>
with vfat you dont have permission problems :)
<dangole_>
fda: it's a tarball, so there are permissions inside that and they should not get lost or randomly change.
<fda>
dangole_: restore by luci gets 700 permissions, by terminal 755
<fda>
for a directory which had 750
<fda>
the problem is that dir-permissions are not save in the tar!
<dangole_>
fda: if the problem you observe is on the e8450 mt7622 platform, this is most likely a flag to tar missing somewhere in the configuration saving codepath.
<fda>
dangole_: for a fix see on my localserver file "__backup_dirs.patch.txt"
<fda>
and commit if you like
<fda>
i test with my fritzbox 7320
<stintel>
philipp64: no backups?
<stintel>
philipp64: did you resize boot or root partition? in this case I've also lost configuration iirc
<dangole_>
fda: remind me the URL please ;)
<stintel>
dangole_: not doing anything special afaik
<fda>
"sysupgrde --list-backup" list the FILES for the tar - no dirs
<fda>
dangole_: autofs: if blockd works too its find and i dont need autofs. as it uses autofs module it should not hard to add nfs
<dangole_>
fda: most of the patches on that URL look like they should be merged. are you planning to submit them (and write patch descriptions...)?
<fda>
dangole_: yes, not all, i know. description: yes. merge: im still not sure about the signoff
<fda>
theme-things: i think i add a complete ne theme, with dark. most people dont like changes of current theme
<dangole_>
fda: why do you need __sysupgrade-losetup.patch.txt ? ie. i don't understand why it's needed (it might well be, so please enlighten me)
<fda>
@dangole_ for "mount -o bind" i have busybox applet "losetup"
<fda>
sysupgrade detects it and uses it
<fda>
but the applet does not know "-D"
<fda>
so add the full path to the binary version prevent detect of the applet
<fda>
you see then an error in console durin sysuograde (only my 7320 has a console)
<dangole_>
fda: ah ok, get it. that busybox applet isn't used in OpenWrt usually.
<fda>
for my usage the applet is enough
<fda>
my changed of theme so far: https://ibb.co/TBQ0zbW the bigger "bridgeboxes" may be good for devices like 8450 with 4Xlan + 2Xwlan + 1Xbridge
<fda>
dangole_:
<fda>
and i dont like the "description" of services and like more the name itself