mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
<kerneltoast>
nbd: hi! i've got an interesting problem with mt7915: transmitting from mt7915 with MCS >9 @ 160 MHz bandwidth results in a near-dead connection, and i'm not seeing any packet mangling from a monitor capture
<kerneltoast>
when i force MCS 10 or 11 on mt7915, it suddenly starts sending BA requests and then the BA miss count rapidly goes up
<kerneltoast>
this even happens with mtk's official driver on my device's stock firmware, so i don't think it's a bug in mt76
noltari has quit [Quit: Bye ~ Happy Hacking!]
noltari has joined #openwrt-devel
noltari has quit [Read error: Connection reset by peer]
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Remote host closed the connection]
<Habbie>
Mangix, ah
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
srslypascal has joined #openwrt-devel
mzvd has quit [Remote host closed the connection]
mzvd has joined #openwrt-devel
goliath has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mangix has quit [Read error: Connection reset by peer]
mangix has joined #openwrt-devel
mzvd has joined #openwrt-devel
<mrkiko>
Mangix: something like libubox host build failure
mzvd has quit [Read error: Connection reset by peer]
valku has quit [Quit: valku]
mzvd has joined #openwrt-devel
goliath has quit [Quit: SIGSEGV]
<mangix>
mrkiko: I'd like more info.
<mrkiko>
Mangix: eh, I should restart building ... give me some time
<mrkiko>
Mangix: but bascically ld gives a reference error - arc4random
mzvd has quit [Ping timeout: 480 seconds]
mzvd has joined #openwrt-devel
<mangix>
mrkiko: incredible. are you building on macOS?
<mrkiko>
Mangix: no, archlinux
<mrkiko>
Mangix: I am distclean'ing things now- will rebuild from scratch
srslypascal is now known as Guest4144
srslypascal has joined #openwrt-devel
<mangix>
mrkiko: so arc4random is a function only found on BSDs. The only thing I can think of is that you have libbsd installed locally and json-c is picking up that.
<mrkiko>
Mangix: can confirm I have libbsd locally installed
<zorun>
f00b4r0: what is breaking exactly? it looks like both symlink or empty exec should be equivalent in 21.02?
<f00b4r0>
you can't run "make" without this patch
<f00b4r0>
due to the error listed in line 2 of commit msg
<zorun>
strange, we don't even have meson in 21.02
<f00b4r0>
the change is generic AFAICT
<f00b4r0>
and it happened for me on ath79
<f00b4r0>
well
<f00b4r0>
on anything in fact
<f00b4r0>
since you can't even run make menuconfig
<zorun>
ok, will test (for curiosity) & backport
<f00b4r0>
basically I went master -> 22.03 -> 21.02 (as I was backporting the same patch) and in 21.02 I hit that bug. Cherry picking fixed it immediately.
<mangix>
you need to run make package/libjson-c/host/clean first
<mangix>
wait a minute
<mangix>
this failure is in libblobmsg
<mangix>
you know what, just delete build_dir/hostpkg and try building again
<mrkiko>
Mangix: consider I started from a distclean state
<mrkiko>
Mangix: but effectively this was the second run since I did the first one in parallel - otherwise it would take much more time. Now deleted the folder and re-doing it
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
<mangix>
mrkiko: what I think is happening here is that libbsd is getting linked statically which makes libubox need to link it. tempted to call it an arch linux packaging bug.
mangix has quit [Remote host closed the connection]
Tapper has joined #openwrt-devel
mangix has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
<nick[m]1234>
Can someone with merge access may merge the json-c and f2fs-tools fixes?
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
<f00b4r0>
zorun: thanks!
<f00b4r0>
robimarko: indeed. One bit of info missing is the boot time impact. I suspect there's a decompression time penalty. Probably not as bad as lzma + lzma-loader, but you never know
<stintel>
I'd much rather see more work done on self-decompressing kernel
<stintel>
and ditch lzma-loader altogether
<stintel>
we have maybe 10+ different miniforks of it
<stintel>
it's a disaster
<f00b4r0>
true that
<robimarko>
f00b4r0: Yeah, I am still to test it myself, though I doubt that couple of seconds on the usually slow devices matters more than flash usage
<robimarko>
LZMA loader is a dumpster fire currently, every arch has its own cobbled together fork
<f00b4r0>
robimarko: sure, if it's the same order of magnitude. If the device takes 2 or 3x more time to boot, the evaluation may be different :)
<f00b4r0>
but I agree that we should strive to get rid of lzma-loader
<stintel>
get rid of, or at least merge it into a single one used by all targets that need it
<stintel>
that said, I did not manage to get the self-decompressing kernel to work on realtek
<robimarko>
I am not in the loop, does MIPS support zImage?
<f00b4r0>
i wonder if this could help us further repel going back to lzma-loader on mikrotik ath79 initramfs (see #9911)
<stintel>
I believe svanheule had it working on ath79
<robimarko>
Yeah, better get zboot working rather than lzma-loader
<karlp>
isn't one of the limiting factors bootloader requirements though?
<f00b4r0>
robimarko: fully agreed
mzvd has joined #openwrt-devel
<stintel>
I was trying it on the Edgecore ECS4100-12PH because the default u-boot env bootcmd: bootcmd=mtdparts default;sf probe 0;sf read 0x81000000 0x00200000 0x300000;bootm
<robimarko>
karlp: AFAIK, the whole point of zboot is that it is self-extracting, thus not requiring bootloader support
<stintel>
I can change the u-boot env but it's nicer if we can just boot with the default env
<robimarko>
stintel: I am seeing more and more crazy stock bootcmd-s
<robimarko>
They really like hardcoding the kernel size there
<stintel>
yeah :(
<stintel>
and 3M is not much in 5.10 or 5.15 times
<robimarko>
3M is great, D-Link hardcoded the kernel partition to 1.3M
<stintel>
pfff
<f00b4r0>
yikes
<robimarko>
And to sweeten the deal, they use dual FW on the 32MB of NOR
<robimarko>
And enforce checksum verification
<f00b4r0>
ugh
<robimarko>
And that only works within the partition bounds
<robimarko>
Had to modify the bootloader to get rid of that
<robimarko>
OpenWrt currently works by luck on DGS-1210 series as it fails checksum verification but the implementation is so dumb that it just cares that one of the FW-s passes
<robimarko>
And just boots from the initially requested offset
<stintel>
pfff I cancelled my order of the dual 10G NICs. ordered on amazon.de with expedited delivery on thursday, they were still not dispatched so they wouldn't arrive before I'm travelling again
<stintel>
expedited delivery my ass
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
minimal has joined #openwrt-devel
KGB-2 has quit [Remote host closed the connection]
KGB-2 has joined #openwrt-devel
danitool has quit [Ping timeout: 480 seconds]
mzvd has joined #openwrt-devel
Borromini has joined #openwrt-devel
zx2c4_ has quit []
zx2c4 has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
minimal has quit [Quit: Leaving]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Ping timeout: 480 seconds]
Borromini has quit [Quit: Lost terminal]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
fda has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
<hauke>
nick[m]1234: we have a github action to build test the tools for linux and macos
AtomiclyCursed has quit [Quit: ZNC 1.8.2 - https://znc.in]
AtomiclyCursed has joined #openwrt-devel
<rsalvaterra>
SwedeMike: In the context of the buffer bloat chat, the other day, it seems like there's a Portuguese ISP that gets it…! https://www.dslreports.com/speedtest/71094711
<SwedeMike>
very nice!
<rsalvaterra>
This is with the provided CPE. Quite remarkable.
cbeznea has quit [Quit: Leaving.]
danitool has quit [Remote host closed the connection]
danitool has joined #openwrt-devel
Ansuel has joined #openwrt-devel
<Ansuel>
i wonder... any reason tools is not part of the toolchain ?
<Ansuel>
or any problem if for example i cache them in github ci ?
<Ansuel>
i'm having fun with external toolchain trying to reduce github ci actions but tools compile time is still 20 min
<Habbie>
i guess the SDK is not for you in this case?
<Ansuel>
sdk is good for packages. with openwrt repo we need actions that runs on a full buildroot
<Ansuel>
so i'm trying to reduce all the wasted time with things not related to the target
<Ansuel>
currently to arrive at the target build 20min of "setup" are needed.... 10-15 min of them are tools compiling
<Ansuel>
github.com/Ansuel/openwrt/runs/7186547625 this is the current runner
<Ansuel>
the user was a bit slow to answer to i fixed the title and merged. should i stop doing that?
<Ansuel>
(honest question)
<mangix>
another thing is that the driver is a shitshow. it needs a cleanup badly. also hardware checksum support so that GRO can work properly.
<mangix>
there's also the issue that there's an upstream ag71xx driver which is not this one.
<Ansuel>
it's a dsa driver
<Ansuel>
but perf are shit
<mangix>
qca8k is the dsa driver, no?
<Ansuel>
fot that soc it's mixed... some are qca8k based other are based on another driver and still doesn't actually work due to changes in the tagger
<Ansuel>
but main problem is that perf drop from 1gig to 100-200
<robimarko>
If its the internal switch then AR9331 DSA exists
<robimarko>
I think somebody was working on AR934x support as well
<Ansuel>
robimarko there was one guy that tested it
<Ansuel>
it didn't work
<Ansuel>
by checking the documentation the tagger used for ar9331 is wrong
<Ansuel>
(but it was probably ar934x)
<robimarko>
Well, you gotta also remember that old atheros docs were hit and miss with the "facts"
<Ansuel>
true but we found 2 diferent documenation that had the same tagger documentation just one had more info than the other
<Ansuel>
it's similar to qca8k one with the difference of a different handling for the ports dest/source
<Ansuel>
../src/drivers/driver_nl80211.c:2939:31: warning: passing argument 1 of 'nl80211_bss_msg' from incompatible pointer type [-Wincompatible-pointer-types]
<Ansuel>
../src/drivers/driver_nl80211.c: In function 'wpa_driver_nl80211_del_beacon':