rmilecki has quit [Quit: Konversation terminated!]
rmilecki has joined #openwrt-devel
rmilecki has quit []
rmilecki has joined #openwrt-devel
robimarko has joined #openwrt-devel
<mrkiko>
hasn't PR 13087 already been merged via commit vf70ee53b08466f612546f699c556cbdaa39e1466 ? Or am I missing something?
Tapper has joined #openwrt-devel
schwicht has joined #openwrt-devel
schwicht has quit []
<mrkiko>
rmilecki: I've seen commit c7655e207f2e6a89db9835b2304dcbf17d6623b5 ... was there any controversial element about it? It seemed quite small as a patch
<rmilecki>
mrkiko: i asked for upstreaming it, but author couldn't do that and asked for dropping it instead
<rmilecki>
i'm ok with downstream patches that can't be upstramed because they are somehow hacky
<rmilecki>
this one is perfectly fine and it's zero-effort to upstream it
<mrkiko>
rmilecki: got it; seemed strange looking at it, it didn't bring a small gain after all. But OK
<dwfreed>
no idea, never used ynezz's action or that one
* jow
smells a can of worms
<Ansuel>
uses: openwrt/gh-action-sdk@main
<Ansuel>
env:
<Ansuel>
ARCH: ${{ matrix.arch }}
<dwfreed>
some make target early in the build goes through and updates the symlinks in staging_dir/host/bin to the first matching item in $PATH; if you've got staging_dir/host/bin at the start of your PATH, it ends up making symlinks that point back to themselves
<dwfreed>
you don't actually need staging_dir/host/bin in $PATH for the SDK to work, the buildsystem handles that for you
<jow>
so the gh-action-sdk will simply build the repo using it as OpenWrt package?
<dwfreed>
yeah
<jow>
that means the package needs to ship an OpenWrt makefile?
<jow>
*the repo needs to ship an OpenWrt Makefile, referencing itself
<jow>
is there some implementation example somewhere?
<jow>
hm yeah, so the repo needs to be a proper feed
<jow>
ynezz' CI so far was just a "compile this cmake project using the SDK gcc"
<jow>
adding an OpenWrt makefile is doable, just implies a bit of code duplication (having a CI specific self-referencing Makefile and another copy somewhere in the packages feed or base)
<dwfreed>
you probably don't need to duplicate the makefile if you symlink it to the one in the right openwrt feed
<jow>
but then I'm not testing the latest version, but whatever is set in the feed makefile
<dwfreed>
can't you override the source location ?
<jow>
ah right, git-src
<jow>
or rather source tree override
<dwfreed>
horrible hack, duplicating the makefile might be better, but it would likely work :D
<jow>
but I guess it's not abstracted in those gh-actions-sdk
<jow>
so needs manual commands
<jow>
at that point I could also imply invoke make
<jow>
Ill go with the duplicated makefile for now
<dwfreed>
and this way you know if you've got to update your CI makefile, when you go to land the release in OpenWrt proper, you can copy those changes to the OpenWrt makefile
valku has joined #openwrt-devel
minimal has joined #openwrt-devel
slh has quit [Ping timeout: 480 seconds]
slh64 has quit [Remote host closed the connection]
slh_ has joined #openwrt-devel
<Slimey>
Ansuel what do i need to do now heh
<jow>
dwfreed: hm another issue, the shipped repo makefile does not take precedence over the one in the openwrt feeds
<jow>
guess I could rename it, that makes the generated .ipk artifacts less useful though
<dwfreed>
jow: you have to change the feeds.conf order
<dwfreed>
which might not be possible with the action
<dwfreed>
ideally the action would make the local feed have precedence over the standard feeds
<jow>
can of worms is getting bigger :D
<dwfreed>
perhaps this is why I did not use the action
<dwfreed>
the lack of flexibility to do things how I wanted/needed
<jow>
yeah, noticing the same here
<jow>
for example I would like to only pull the base feed
<dwfreed>
it's more geared toward shipping a third party package, but can probably be easily stripped down to your needs
<jow>
thanks
<jow>
will try to fumble with NO_DEFAULT_FEEDS and EXTRA_FEEDS before going into the other rabbit hole
gladiac has quit [Quit: k thx bye]
<Slimey>
is there such as thing as using a USB-A to A cable with some routers and accessing them some how?
<Habbie>
there is such a thing, but i haven't seen it very often
<dwfreed>
a proper serial port is significantly more common, along with the necessary bootloader config to use it
<Slimey>
adtran has a netgear wax218 clone with no console port and i never could figure out the serial pins on the board
<Slimey>
bsap-6040 4x4 ax ap
<tmn505>
Slimey: what's so difficult about finding them? The https://fccid.io/2AL6XWAP380/Internal-Photos/Internal-photos-4872069.pdf clearly shows UART pin holes. The BT one is probably for bluetooth while the 2nd one is probably routed to SoC. Check the voltage of pins, set Your UART adapter to matching voltage, connect GND and check with RX cable from the adapter on which hole You'll get some output.
<tmn505>
Offcourse don't probe voltage pin. Very unlikely serial output is completely disabled.
goliath has quit [Quit: SIGSEGV]
<Slimey>
yeah i was never able to get any console output from them
<tmn505>
Slimey: so dump contents of SPI flash if there's no output. BTW did You check J1 pins (or J7 can't see clearly), what are they?
<Slimey>
might hack to some how :P jtag pins are populated tho but i broke my jtag :\ im not sure ill have to take one completely apart and see if i can find anything
<philipp64>
mcbridematt: thanks for getting the ARM platform normalized. btw, if you have any ten64's running around that you need to rehome, I could use one for benchmarking DPDK and VPP on... hint hint...