nerdboy has quit [Remote host closed the connection]
rfuentess has joined #yocto
<ak77>
morning
nerdboy has joined #yocto
nerdboy has quit [Changing host]
nerdboy has joined #yocto
g0hl1n has joined #yocto
frgo_ has joined #yocto
mulk has quit [Read error: Connection reset by peer]
frgo has quit [Ping timeout: 245 seconds]
frgo_ has quit [Ping timeout: 264 seconds]
frieder has joined #yocto
ptsneves has joined #yocto
wojci has quit [Remote host closed the connection]
CrazyGecko has joined #yocto
farmadupe has quit [Ping timeout: 248 seconds]
mulk has joined #yocto
frieder has quit [Ping timeout: 264 seconds]
ablu has quit [Ping timeout: 260 seconds]
farmadupe has joined #yocto
ablu has joined #yocto
prabhakalad has joined #yocto
frieder has joined #yocto
farmadupe has quit [Ping timeout: 276 seconds]
farmadupe has joined #yocto
frieder has quit [Ping timeout: 248 seconds]
dkl has quit [Quit: %quit%]
dkl has joined #yocto
mkusiak has joined #yocto
berton has joined #yocto
florian has joined #yocto
frieder has joined #yocto
g0hl1n has quit [Quit: Client closed]
mkusiak has quit [Changing host]
mkusiak has joined #yocto
mkusiak has quit [Quit: Client closed]
mkusiak has joined #yocto
<mkusiak>
Hello!
<RP>
mkusiak: hello!
<RP>
ak77: which distro and project release is that with?
<RP>
ak77: I've been away from computers, I wasn't ignoring you!
kanavin has quit [Ping timeout: 246 seconds]
kanavin has joined #yocto
<berton>
Hello all! I created a handler to use bb.event.BuildCompleted and the function that the handler calls needs opkg-utils-native, so I added this `do_my_task[depends] += "opkg-utils-native:do_populate_sysroot"`. I see that opkg-utils-native goes to sysroots-components, and if I remove the `do_my_task[depends]` opkg it is not there.
<berton>
Can I use the same varflags from an addtask when using the addhandler, like `depends` in this case?
mbulut has joined #yocto
<ak77>
RP: Scarthgap
<rburton>
berton: event handlers don't have sysroots etc afaik
<ak77>
RP: "nodistro"
g0hl1n has joined #yocto
mkusiak has quit [Quit: Client closed]
<berton>
rburton, What I'm trying to do is avoid calling package-index in a separate bitbake call, so I create a bbclass to run the same generate_index_files(d) from package-index.bb using the BuildCompleted event. And, I prepend sysroots-component to the PATH, so the opkg used is from sysroots-component. I don't know if this is a safe thing to do haha, but in my tests the result was the same compared to running `bitbake package-index`
<RP>
ak77: I wonder if we're missing some backport to scarthgap for py 3.13
<RP>
berton: sysroot-components are just the bare specific things with no dependencies, so no, that isn't safe. You may get lucky if the dependencies are minimal
<RP>
berton: as rburton says, events run in a different context to tasks/recipes and there is no WORKDIR/sysroot for them
<berton>
RP: Ok, thanks! And is there any safe way to avoid run package-index? We always need to workaround things like removing the git commit from buildhistory and there is also new directory in buildstats
<rburton>
that horrible thing can be inherited in an image class and it will regenerate the feed when the image is built
<kanavin>
someone was also working on an elaborate patchset for package indexing that would solve the same issues, they eventually abandoned the effort
<rburton>
i tried long ago with events but it was a bit annoying, probably the same problems you're hitting
<kanavin>
haha, that is not saying much, but you can find the iterations of that patchset in oe-core archives
<RP>
we don't really have a good way in bitbake to say "run this single target/task after everything else"
<RP>
it sounds simple until you realise people may have several of them, then you have to think about ordering and so on.
<RP>
Easier would be to turn off buildhistory/buildstats for your execution of package-index?
<berton>
RP: it is an option, but I need to remove the INHERIT from local.conf, right?
<berton>
Thanks for the answers. I will try other ways. If I find something that works and is a good solution, I will post it to the mailing list.
mbulut has quit [Quit: Leaving]
<ak77>
RP: I didn't think you were ignoring me.
g0hl1n has quit [Ping timeout: 240 seconds]
wojci has joined #yocto
Guest24 has joined #yocto
Guest24 has quit [Client Quit]
florian has quit [Quit: Ex-Chat]
farmadupe has quit [Quit: Lost terminal]
g0hl1n has joined #yocto
<ak77>
RP: looks like it works with that patch applied to scarthgap.
<ak77>
RP: Can we get it to scarthgap branch, maybe with my Acked-by :)
<RP>
ak77: you need to post the patch on list and copy Steve Sakoman (the stable maintainer)
<RP>
sakoman: ^^^
<sakoman>
ak77: I'll watch for your patch on the mailing list :-)
tgamblin has joined #yocto
rfuentess has quit [Remote host closed the connection]
rfuentess has joined #yocto
<ak77>
ah. no need. it's already there, I should pull more ofter :S
bwani54 has joined #yocto
cyxae has joined #yocto
<RP>
ak77: I was a little surprised we hadn't done this...
PresidentBiden51 has joined #yocto
<ak77>
good work guys, thank you. and sorry for the noise.
ListenHereJack35 has joined #yocto
ListenHereJack35 has quit [Remote host closed the connection]
PresidentBiden51 has quit [Remote host closed the connection]
KVCBuildBackBett has joined #yocto
KVCBuildBackBett has quit [Remote host closed the connection]
<SlimeyX>
biden spam lol?
BZ683678 has joined #yocto
<BZ683678>
Hey guys... Joe Biden here. I've decided to step down from the White House to focus on other projects. Billionaires are a threat to democracy, so check out https://BidenCash.st to put them in the bullseye. Keep an eye on the CNN inauguration for a promo code!
BZ683678 has quit [Read error: error:0A000119:SSL routines::decryption failed or bad record mac]
<SlimeyX>
stupid bots everywhere
florian has joined #yocto
wojci has quit [Ping timeout: 276 seconds]
goliath has quit [Quit: SIGSEGV]
tgamblin has quit [Remote host closed the connection]
tgamblin has joined #yocto
frgo has joined #yocto
florian has quit [Ping timeout: 276 seconds]
bhstalel has joined #yocto
frieder has quit [Remote host closed the connection]
berton_ has joined #yocto
berton_ has quit [Client Quit]
berton has quit [Ping timeout: 244 seconds]
<bhstalel>
Hello everyone, does wic tool modifes /etc/fstab and duplicates some entries ?
<bhstalel>
my rootfs 's /etc/fstab in the WORKDIR of the image recipe, is correct, after completing the .wic generation and flashing the SD card, the entries are duplicated, example:
<bhstalel>
Initial fstab:
<bhstalel>
After wic:
<bhstalel>
I cannot paste the full fstab, but the idea is clear I think.
<bhstalel>
anyone have an idea ?
leon-anavi has quit [Quit: Leaving]
starblue has quit [Quit: WeeChat 3.8]
starblue has joined #yocto
<bhstalel>
(WIC_CREATE_EXTRA_ARGS ?= "--no-fstab-update") should do the trick, being in a rush makes you forget about everything you learned
goliath has joined #yocto
<mckoan>
bhstalel: haste makes waste ;-)
rfuentess has quit [Remote host closed the connection]
tgamblin has quit [Ping timeout: 244 seconds]
paulg_ has quit [Quit: Leaving]
tgamblin has joined #yocto
jmd has joined #yocto
florian has joined #yocto
bhstalel has quit [Quit: Client closed]
<ak77>
Bhastel, yes it does change
mckoan is now known as mckoan|away
druppy has joined #yocto
Guest33 has joined #yocto
frgo has quit [Remote host closed the connection]
<fray>
sounds like there will be an oedam
<fray>
wrong channel
<Guest33>
do you have any personal OEM-layer-quality-ranking? Say, something like meta-raspberrypi, meta-intel, meta-nxp, meta-st-stm32mp, meta-ti ... ect ect ;)
druppy has quit [Ping timeout: 265 seconds]
frgo has joined #yocto
<Guest33>
by quality i mean, how hard is it to use it, how messy it is, how much effort would it take - to port from board X to Y
<fray>
layers that have Yocto Project compatible are often better quality. Layers that use Yocto Project kernel (vs vendor specific kernels) are often, but not always better as well..
<fray>
often that question is silicon vendor specific. (I work for AMD formerly Xilinx) and we've done a lot to make porting easier then in the past. But it's never "easy"
<Guest33>
thanks, i am curious about hive mind answers, but happy to have yours too. Why do you think it is so, that manufacturers weren't able to come up with some reasonable standard or common approach, enabling people to change HW easier?
<fray>
there is little to no incentive to allow changing hardware.. in fact often there is this belief that allowing an easy change of hardware is BAD as you can lose customers. But that's not the real anwer.. the real answer is that there is no real standard for hardware in the embedded space.. Everyone developers their chips differently, everyone developers their boards separately, and everyone thinks their
<fray>
magic combination is the best possible way of doing it (no concensus)
<fray>
What we've done at AMD (Xilinx) for our FPGA projects is build a process, and a tool that automated the process to take our EDA tool output and convert it to a Yocto Project machine configuration (and associate files). This really helps OUR customers, I'd be happy to work with other silicon vendors (and competitors) on doing something similar, but the reality is the complexity of _our_ FPGAs is something
<fray>
that someone like NXP just doesn't have.. so there is no need to complicate things there.
<Guest33>
but then, maybe i don't fully get the problem. What real difference, from yocto perspective, is there? Firmware - sure. U-boot config, or whatever low level bootloaders, need usually similar input, yet i think it is always customized in such way, that makes it gibberish
<fray>
I personally have always had the opinion that the easier it is to swap hardware the better. As it allows different companies to compete on their strengths... but like I said, there is this belief that "lock-in" (explicit or implicit) is good.. it's not
<fray>
YP is built up for local configuration, distro, machine and recipes.. of those, in general, you can keep the local config, distro and recipes vendor neutral (more or less).. which leaves the machine confgiuration.
<Guest33>
hmm, i see. But fpga's are somewhat different breed than "just ARM"
<fray>
The machine configuration captures everything from firmware, to u-boot, to (required) kernel configuration (indirectly), and importantly device-trees..
<fray>
In the typical CPU world, it's "easier" for someone to hardcode and specify a device-tree, kernel configuration, u-boot configurationa and firmwares...
<fray>
So NXP will specify their config, TI will specify their config, etc..
<fray>
But moving from a 1 board to another, the implementation changes for the reference boards.. how memory is configured, timings, if things are conencted or not.. off-chip devices, etc.. So this means kernel configs, firmware, u-boot and device-trees change.. there is no "help" to the end user to do this, they just need people who are knowledgable to implement that stuff..
<Guest33>
well, indeed it might be easier for them. Also some layers just seem to span over bunch of (own) boards, using different dtb's so adding to complexity.
<fray>
Now move to the FPGA world (fpga w/ a traditional CPU in it).. In my case, lets take a ZynqMP part. It has a cortex-a53, a cortex-r5f, and a microblaze CPU that runs everything.. THEN you also have the FPGA which can implement custom devices.. All of that needs to be coordinated, and what would be 'firmware' on other platforms is now exposed. So something needs to manage that complexity and specify it..
<fray>
but in the end, 90% of the Yocto Project components are vender, silicon, arch, etc neutral.. its that last 10% you have to deal with and make work.. and most vendors have no incentive to do their work in a way to enable anything but what they are focusing on
<fray>
the process and tooling (for our fpgas) we built is designed to help coordinate all of that, generate device-trees, etc.
florian has quit [Ping timeout: 252 seconds]
<Guest33>
yet somehow, LK managed to organize them somehow =# so you personally, apart from obviously your own, do you have a subjective quality ranking? :)
<fray>
So back to my point, everyone (generally speaking) contributes to the 90% common code parts... the last 10% is architecture, cpu, implementation specific and short of best practices type behavior there is no reason for anyone to collaborate as the results are jsut 'different'. Companies could come together and try to do 'something', maybe along the lines of a CPU architecture, like ARM. But even between
<fray>
arm vendors the implementations are really different and there just isn't the incentive there
<fray>
LK doesn't manage it at all.. Look at how disjoint the components in the arch directory are.. but there is a clear incentive to people to get their changes into the LK -- but if you look at embedded _everyone_ has their own kernel tree and almost nobody uses the upstream stock LK as it isn't "complete" for a given part..
<fray>
As I said, layer quality.. does it pass YP Compatible? Does the layer have it's own customer kernel source or use common upstream or YP kernel source? Does it do things according to best practices? that's your quality metrics
<Guest33>
thanks. Not a lot on YP compatible, but a nice list indeed. I don't mind people having their own kernel, it is after all, the same one, though modified, either with patches or heavily modified and kept on a branch. In Yocto, being a distro builder, such freedom would mean that even the meta structure is different, which would be, for me, the
<Guest33>
equivalent of every OEM using different VCS for kernel + different directory structure :)
<fray>
more then just a different branch. With everyone havign a different kernel, you also get different versions, different patches, CVEs not being patched (or hell even being able to be tracked).. it's a nightmare..
<fray>
the keys in the questions are minimal behavior required for a layer to not break others, and there is a test script that verifies this.
ebassi has joined #yocto
<Guest33>
haha, i like how the money question is first "The submitter is a current Yocto Project member (Platinum, Gold, or Silver level)" :D
frgo has quit [Remote host closed the connection]
<fray>
it's a trademark
<fray>
the rest of the questions are best practice based
<fray>
by law you must protect trademarks, and in this case one of the YP requirements is you must be a member or sponsored by a member.. (a paying member). Thats why didn't say you HAD to be YP compatible, but it is a sign of better quality.. There are non-member layers that are just fine (quality wise) since they do conform to pretty mcuhe verything else there..
<Guest33>
funny, no one has yet written a simple script to check those? damn, i could have low hanging fruit first yocto contrbution :)
<fray>
and there are layers owned by member companies that are horrible quality (using these metrics)
<fray>
There _IS_ a script to check what can be checked
<Guest33>
oh. Bummer
<fray>
"All layers have successfully passed testing with the test script yocto-compat-layer.py"
<fray>
that is your script
frgo has joined #yocto
<fray>
(to be clear, I have layers I support which I would NOT consider to be 'good quality' using those metrics, but those layers are designed for specific demonstration purposes, and are thus not applicable for anyone but a very very limited use-case... thats why running the comptible test script and userstanding the purpose of the layer is required to make a quality judgement)
<Guest33>
Hmm so some questions on that form you linked could be just reduced to "layer passed the test(s) of yocto-compat-layer.py" :)
<fray>
things like does this file exist, do you not break certain things.. yes those are script testable
<Guest33>
on the other hand, if you have empty SECURITY file...
<fray>
other things are not explicitly testable, but are required for trademark reasons.. i.e. the last two questions
<fray>
and there are ways to "cheat" with any sort of automated script, which is why the questions are asked
<rburton>
Guest33: note that the first question is a member, or a member (including OpenEmbedded itself) endorses your layer. if you have a community layer then OE reps can endorse it.
savolla has joined #yocto
jmd has quit [Remote host closed the connection]
bwani54 has quit [Ping timeout: 264 seconds]
SlimeyX_ has joined #yocto
SlimeyX has quit [Ping timeout: 248 seconds]
SlimeyX_ is now known as SlimeyX
florian has quit [Ping timeout: 252 seconds]
savolla has quit [Quit: WeeChat 4.4.3]
rob_w has quit [Read error: Connection reset by peer]