lynxis has quit [Remote host closed the connection]
lynxis has joined #openwrt-devel
madwoota has quit [Ping timeout: 480 seconds]
madwoota has joined #openwrt-devel
tomn has quit [Remote host closed the connection]
tomn has joined #openwrt-devel
floof58 is now known as Guest6024
floof58 has joined #openwrt-devel
Guest6024 has quit [Ping timeout: 480 seconds]
goliath has joined #openwrt-devel
<dhewg>
nbd: master from 2 days ago still fixes the issue I've been seeing
<dhewg>
maybe there're multiple related issues at play? Anyway, will update to master
goliath has quit [Quit: SIGSEGV]
<mrkiko>
Morning all!!
mirko_ has joined #openwrt-devel
mirko has quit [Read error: Connection reset by peer]
KGB-2 has quit [Remote host closed the connection]
g___ has quit [Remote host closed the connection]
g___ has joined #openwrt-devel
indy has quit [Ping timeout: 480 seconds]
KGB-2 has joined #openwrt-devel
eloy__ has quit [Ping timeout: 480 seconds]
eloy_ has joined #openwrt-devel
indy has joined #openwrt-devel
bluew has quit [Ping timeout: 480 seconds]
goliath has joined #openwrt-devel
zarzarzar_ has joined #openwrt-devel
zarzarzar has quit [Read error: Connection reset by peer]
gladiac has joined #openwrt-devel
<nick[m]12>
djfe: I just compile my image by my own and change the mac80211 scripts to automatically enable wifi by default
rua has quit [Quit: Leaving.]
<nbd>
dhewg: i think the issue i was talking about is only reproducible if you don't get any broadcast/multicast traffic from the node you're trying to reach
srslypascal is now known as Guest6044
srslypascal has joined #openwrt-devel
Guest6044 has quit [Ping timeout: 480 seconds]
srslypascal has quit [Read error: No route to host]
srslypascal has joined #openwrt-devel
<dhewg>
okay
<dhewg>
I updates both my routers with current master, I'll let you know in case it breaks for my setup
<dhewg>
*updated
<dhewg>
rmilecki: I'm still getting the steering errors
<dhewg>
daemon.notice procd: /etc/rc.d/S25packet_steering: uci: Entry not found
<dhewg>
daemon.notice procd: /etc/rc.d/S25packet_steering: sh: write error: No such file or directory
<dhewg>
former with packet steering disabled, letter with it enabled
<dhewg>
*latter
danitool has joined #openwrt-devel
<dhewg>
I do have irqbalance on there, maybe that plays a role in the access error?
<rmilecki>
dhewg: i pinged nbd about that, but I don't think he replied with anything / any idea
<rmilecki>
dhewg: it should be harmless
<rmilecki>
dhewg: we should see why kernel refuses writing to that procfs file
<rmilecki>
i don't have time to look at it right now
<[Pokey]>
Hey \x, what box with an RGB LED in the front was it again that you were working on last year?
<[Pokey]>
Still waiting on my device to make it in xD
<\x>
it does work but i say youre bettter off with ipq807x
<\x>
but if you can get MR7350 cheap then maybe its fine, I got mine for 40$
<[Pokey]>
Not looking to get a new box, was just discussing OpenWRT with someone else to be honest
<[Pokey]>
I've been sat with the Pine64 guys for about 2 months I think now and learned a ton about embedded devel, written my first two drivers, so I will probably appear here again at some point to add support for OpenWRT to the new Ox64 once networking is stable
<[Pokey]>
I have you to thank for getting me stepped on that whole train though :P
valku has quit [Remote host closed the connection]
jenders has joined #openwrt-devel
gladiac has quit [Quit: k thx bye]
jenders has quit [Ping timeout: 480 seconds]
<dhewg>
that seems to be from openssl3/libcrypto? weird, I don't get that on another device
<dhewg>
oh wait
<dhewg>
lol, that's my own ustream-ssl debug spew that leaks into uhttpd
<dhewg>
it breaks the luci menu
<dhewg>
nvm
rua has quit [Quit: Leaving.]
<karlp>
oh, heh. yah, gotta be careful with printing in things that talk to luci. gotta be real careful to use stderr :)
maciekb has quit [Read error: Connection reset by peer]
maciekb has joined #openwrt-devel
Misanthropos has joined #openwrt-devel
rua has joined #openwrt-devel
<dhewg>
hehe yeah, my bad, but still a bit weird. I do get the same debug spew on two devices, yet it only breaks luci on one
floof58 is now known as Guest6060
floof58 has joined #openwrt-devel
Guest6060 has quit [Ping timeout: 480 seconds]
minimal has joined #openwrt-devel
<karlp>
I get fun luci errors when I have half it's files on an nfs share that's unmounted :)
<mrkiko>
Someone knows why on ipq807x there is this "stop spad before upgrading" need, not manifested in other targets? Plain technical curiosity.
<Znevna>
spad?
c0xc_ has joined #openwrt-devel
<oliv3r[m]>
So I have a I2C peripherial, that features 2 I2C master controllers, easy. But they 'share' on register address (for some global configuration). That shared register sits exactly in the middle of both addresses, so if I expand i2c0 by 1, and start i2c1 1 later, they can both cover all the needed registers. However, now we have 2 controllers accessing the same register, so I need locking. Can I have some global mutex variable in my I2C
<oliv3r[m]>
driver, or do I have to make the i2c driver a 'tree' (i2c-ctrl with 2 children, bus0 and bus1?
c0xc has quit [Ping timeout: 480 seconds]
djfe has joined #openwrt-devel
hurricos has joined #openwrt-devel
<neggles>
oliv3r[m]: that sounds like a "one i2c controller, two i2c buses" to me
<neggles>
yes
<oliv3r[m]>
one i2c controller, two i2c busses, 1 pinmux controller :)
<oliv3r[m]>
but the pinmux controller only tells you if a pin is GPIO or SDA; the drive strength etc, is in a different register (shared with all gpios)
<oliv3r[m]>
it's a really big piece of swiss-cheese
<neggles>
I mean, pinmux is irrelevant, pinmux is handled by the pinctrl driver
<oliv3r[m]>
but also i2c controller -> 2 busses -> 8 muxes per bus (ontop of that)
<neggles>
i2c doesn't care if the IO lines are actually routed anywhere
<oliv3r[m]>
yeah, but, the pinctrl driver will need access to 8 bits of one register of the i2c controller :)
<oliv3r[m]>
e.g. the i2c global controller register, also has 8 pinmux bits :(
<oliv3r[m]>
i'm sure it makes a lot of sense to a hardware engineer :D
<neggles>
I see no reason why any of those bits should be manipulated by the i2c master driver
<neggles>
that's all pinctrl
<oliv3r[m]>
yes, but, the pinctrl needs the `reg = <>;` property doesn't it :)
<neggles>
nope
<neggles>
`syscon`
<neggles>
that or you're adding a pinctrl subnode to your i2c master
<oliv3r[m]>
still, the syscon and i2c driver both access the same register
<neggles>
why does the i2c driver access this register?
<oliv3r[m]>
because you configure i2c delay timing there
<oliv3r[m]>
'data hold time', open-drain mode
<neggles>
...which is still a pinctrl thing
<neggles>
no?
<oliv3r[m]>
i2c sampling/holding time?
<oliv3r[m]>
delay, hold and wait times are the only things that are less 'pinctrllery' so if we can wing that under pinctrl it solves that problem nicely
<neggles>
I mean, plenty of dumber stuff hides in pinctrl
<oliv3r[m]>
that's deff. a better way forward
<neggles>
this is probably a stupid question but do any of those values need to be changed when you change clockspeed
<neggles>
or are these all going to be "set up once when the chip's fired up"
<oliv3r[m]>
we're not touching them right now, so probably can even ignore it, in theory
<neggles>
welp
<oliv3r[m]>
but pushing it all under pinctrl makes a lot of sense; if that's 'legal' to do
<neggles>
it's also feasible for both drivers to access the same range as long as you hide it all behind regmap stuff
<neggles>
`regmap.h` is 👌
<neggles>
hmm
<neggles>
looking at the INTERFACE block, I think the "correct" way would be to have a parent DT node which contains a tiny pinctrl and then the two i2c master nodes
<neggles>
from the look of it, spi controller would fall under that as well
<neggles>
yeesh realtek sure didn't do a lot of organization here did they
<oliv3r[m]>
yeah, we do intend to refactor all drivers to use regmaps; but then you still have to somehow make it work nicely int he devicetree (which should be possible, with mfd)
<oliv3r[m]>
yeah, it's a huge pile of stinking mess
<oliv3r[m]>
<neggles> "looking at the INTERFACE block..." <- any pointers to some examples where they make use of this?
<oliv3r[m]>
in any case, to set the pin io strength, that's a different register, shared with different pins, and I'm not sure you can (easily) have two pinctrollers for a single pin?
<oliv3r[m]>
so in that case, I'd leave the entire pinctrl bit out of the equation here, and just mush it with the other pin controller bits
<oliv3r[m]>
Anyway, thanks so far neggles was good to brainstorm this a bit
<Kufat>
nebster, your first line was truncated after "looking in the"
<nebster>
looking in the config.gz, it says "# CONFIG_PERF_EVENTS is not set"
<schmars[m]>
do you mean the vendor's config.gz? PERF_EVENTS is enablement for the perf tracing tool, i think it doesn't have much to do with device support
<nebster>
So, I am using the snapshot from downloads.openwrt.org. So it is the config.gz from there.
<nebster>
I'm assuming that means buildbot just uses the default config from the openwrt repo?
spyrral has joined #openwrt-devel
spyrral has left #openwrt-devel [#openwrt-devel]
<slh>
mrkiko: it's been discussed in detail in the xiaomi ax3600 thread, but in short, there are situations when ath11k won't terminate as quickly as procd wants it to (because it has pending stuff in its queues), this means it keeps wpad running and sysupgrade gives up too early. a sleep 30 (just adding a random -untested- figure here) in sysupgrade after killprocesses, before giving up would paper over the
<slh>
issue already
<Kufat>
nebster, it might be one of those confusing interactions between the OpenWrt builtroot and the kerel build system, but I'm not certain
goliath has quit [Quit: SIGSEGV]
<Kufat>
I just got a BPi R3 and haven't flashed my first image yet, heh
<jow>
nebster: CONFIG_PERF_EVENTS depends on CONFIG_PROFILING to be effective and that appears to be off in the generic kernel config
<nebster>
jow: That's so weird that a config flag can be on without its dependencies and there is no error. I guess I will start building a custom image then. Thanks so much!
<jow>
nebster: the config fragments are sometimes hand edited
<nebster>
jow: I mean I'm surprised the configs aren't checked programatically for things like this, the options are just silently disabled. Do you think a PR would be accepted to set the CONFIG_PROFILING flag to "y" in that file?
<nebster>
Kufat: It seems like a fun little router! I have two and I'm going to try to upgrade my current mesh that works awfully (random ping spikes, maybe even packet loss, I don't remember).
<nebster>
Not sure if that is just because the devices are underpowered or because I have so many wifi-enabled microcontrollers chattering over the network
djfe has joined #openwrt-devel
cmonroe has joined #openwrt-devel
<jow>
nebster: depends on the runtime performance and store footprint implications of CONFIG_PROFILING
Tapper has quit [Quit: Tapper]
cmonroe_ has quit [Ping timeout: 480 seconds]
<Kufat>
nebster: the question I'm asking myself at the moment is how I want to configure the filesystems. Options are rootfs+overlay in flash, emmc mounted to /mnt or similar; rootfs in flash and emmc as overlay; rootfs and overlay on emmc; or something else, like throwing extroot in there
<Kufat>
but I suppose that's more on-topic for #openwrt