<tlwoerner>
but bitbake won't switch the config files without the cleansstate
<RP>
tlwoerner: probably as meta/conf/bitbake.conf:FILESOVERRIDES = "${TRANSLATED_TARGET_ARCH}:${MACHINEOVERRIDES}:${DISTROOVERRIDES}"
<RP>
tlwoerner: only a subset of overrides affect files, I think we did that for performance reasons
<tlwoerner>
makes sense, and it's quite common to set machine-specific overrides, which is why i thought to try MACHINEOVERRIDES when OVERRIDES by itself didn't work
nemik has quit [Ping timeout: 272 seconds]
nemik has joined #yocto
<tlwoerner>
RP: the OVERRIDES vs MACHINEOVERRIDES piece of the puzzle is good, thanks for the pointer/explanation
<RP>
tlwoerner: are you saying with MACHINEOVERIDES it all works or you're still having the cleansstate issue?
<RP>
tlwoerner: I'd bet this is an annoying bug I have now I think about it
nemik has quit [Ping timeout: 246 seconds]
<RP>
tlwoerner: the one we've talked about where if a file is added to WORKDIR in a build, on a rebuild it isn't removed :/
<tlwoerner>
RP: with OVERRIDES nothing works, with MACHINEOVERRIDES it will half-work, but doesn't re-generate when the condition is added/removed from MACHINEOVERRIDES
nemik has joined #yocto
<RP>
tlwoerner: it is a fetcher bug, it doesn't know to remove the file form WORKDIR
<tlwoerner>
RP: true! true!
<tlwoerner>
RP: it simply keeps using the last one
<RP>
tlwoerner: I've tried to fix this one but my fixes are too invasive
<tlwoerner>
RP: lol, sorry to have stumbled upon this one again, i didn't recognize it as another version of that same issue
<RP>
tlwoerner: I didn't recognise it at first
<RP>
tlwoerner: we need to do something about that issue. So many different hard problems :/
polo is now known as money
money has quit [Quit: late]
Guest5904 has joined #yocto
money__ has joined #yocto
florian has quit [Ping timeout: 252 seconds]
Guest5904 has quit [Ping timeout: 256 seconds]
money__ has quit [Read error: Connection reset by peer]
Guest6868 has joined #yocto
vquicksilver has quit [Ping timeout: 268 seconds]
Guest6868 has quit [Client Quit]
vquicksilver has joined #yocto
tokamak[m] has joined #yocto
camus has quit [Remote host closed the connection]
camus has joined #yocto
camus1 has joined #yocto
camus has quit [Ping timeout: 260 seconds]
camus1 is now known as camus
Tokamak has joined #yocto
camus has quit [Remote host closed the connection]
camus has joined #yocto
invalidopcode has quit [Ping timeout: 246 seconds]
ric96 has quit [Quit: Connection closed for inactivity]
money has joined #yocto
money is now known as Guest7780
Guest7780 has quit [Client Quit]
aleksandarsimono has joined #yocto
<aleksandarsimono>
Hey, is there a way to automatically load kernel module but to also pass some arguments while loading? I know about KERNEL_MODULE_AUTOLOAD variable, but I need something that would do insmod kernel_module.ko param=value
florian has joined #yocto
money__ has joined #yocto
ptsneves has joined #yocto
money__ has quit [Changing host]
money__ has joined #yocto
money__ is now known as money
Algotech75 has quit [Remote host closed the connection]
amitk_ has joined #yocto
money__ has joined #yocto
money has quit [Ping timeout: 256 seconds]
vquicksilver has quit [Ping timeout: 246 seconds]
vquicksilver has joined #yocto
money__ has quit [Ping timeout: 256 seconds]
amitk_ has quit [Ping timeout: 268 seconds]
<JaMa>
aleksandarsimono: yes, see KERNEL_MODULE_PROBECONF
<aleksandarsimono>
I have but still only the last one gets executed
seninha_ has quit [Remote host closed the connection]
seninha_ has joined #yocto
<kergoth>
RP: I still wonder about unpacking directly into ${S} and having an option to strip off the leading directory when unpacking the primary archive, rather than unpacking to ${WORKDIR}. Tar has an option for this, for example.
<RP>
kergoth: it could be a lot cleaner but would require quite some changes to recipes
<JaMa>
that sounds interesting, maybe even unpacking to destsuffix (like git supports) with destsuffix default being BP
alessioigor has quit [Quit: alessioigor]
<JPEW>
That still keeps the file:// files in WORKDIR though, isn't that a problem also?
<RP>
JPEW: if we unpack to S and we drop support for S = WORKDIR it would work
<JPEW>
I was wondering if everything (patches, file://, archives, git, whatever) should just be unpacked to ${WORKDIR}/srcs or something; S would default to ${WORKDIR}/srcs/${BP} instead
<RP>
JPEW: there are a number of options, the hard part is working out which one is best and how we could migrate
<JPEW>
and do_unpack[cleandirs] = ${WORKDIR}/srcs just kinda does everything we want
<JPEW>
Sure
<RP>
people don't like lots of directories and I think we already struggle with paths that don't entirely make sense any more too
<kergoth>
I think unpacking to S isn't that different from the srcs approach except for the possible conflict of file:// items and the files within the unpacked source archive, but it also gets closer to aligning with devtool modify behavior, which I'd like to see. *shrug*
<RP>
kergoth: I think S would least surprise users
<RP>
I wonder if we make .patch SRC_URI entries have a magic subdir default. Or maybe all file entries?
<JPEW>
RP: If we go that route, I'd say all of them
<JPEW>
IMHO if you keep anything in ${WORKDIR}, you haven't really solved the problem
<RP>
JPEW: well, I was thinking that combined with a default of S for unpack so you'd get S/extra-unpacked-files
<RP>
I suspect we need to decide what we'd like things to look like, then work to make it happen
<JPEW>
Ya, that might work. I'm leary of mixing the unpacked source and extra files though
<JPEW>
It's not ideal especially when you talk about git repos.... those end up being un-ignored files then
<JPEW>
.. which might confuse devtool? I don't know much about devtool though
<RP>
but if you put them elsewhere like WORKDIR/extra-unpacked-files, you can't just cleandirs S
<RP>
JPEW: until someone overrides the dest subdir for the file://
<JPEW>
Is that allowed for files? That would also break with S/extra-unpacked-files correct?
vquicksilver has joined #yocto
<RP>
JPEW: it is allowed for file urls iirc and it depends how strongly you enforce subdir of S
<RP>
the fetcher parameters give a lot of freedom
<RP>
which is partly why I concluded manifest :/
<JPEW>
I'd rather avoid a manifest; I don't think it's a good play in the long run
<JPEW>
but I probably need more of a think about it
<RP>
JPEW: SPDX users are going to require one anyway and we use manifests for a lot of the rest of the system
<RP>
much as I dislike them
<JPEW>
RP: Right, I don't think that manifests are all awful, but using a manifest to try and correctly clean something rather than just blowing it all away based on structure seem like its going to just be wrong in a lot of edge cases
<RP>
JPEW: I've not thought of an edge case which does break but perhaps I'm not being inventive enough
<JPEW>
I think it's easier to derive a manifest from structure that makes sense than than try to use a manifest to compensate for a less simple structure
<RP>
JPEW: I've been finding corner case after corner case with this bitbake threading change btw :(
<JPEW>
RP: I was afraid of that
<RP>
JPEW: right, I would prefer to have a better structure but I'm not sure we can cover all use cases with a structure and I'm not sure users will like the pain of migrating all their recipes to one either
<RP>
JPEW: It is really close I think, the code was mostly designed to be able to do something like this. There are just a few races and ordering challenges
camus has quit [Remote host closed the connection]
camus has joined #yocto
gho has quit [Quit: Leaving.]
ptsneves has joined #yocto
florian has joined #yocto
aleksandarsimono has quit [Quit: Client closed]
florian_kc has joined #yocto
florian has quit [Ping timeout: 252 seconds]
vquicksilver has quit [Ping timeout: 265 seconds]
ptsneves has quit [Quit: ptsneves]
invalidopcode has quit [Remote host closed the connection]
invalidopcode has joined #yocto
money__ has joined #yocto
goliath has joined #yocto
vladest has quit [Remote host closed the connection]
vladest has joined #yocto
yashraj466 has joined #yocto
money__ has quit [Changing host]
money__ has joined #yocto
money__ is now known as money
vquicksilver has joined #yocto
<kergoth>
We could add the extra files directory to the local git ignore with .git/info/exclude or whatever that path is, without mucking with the .gitignore, yeah?
money has quit [Quit: late]
vquicksilver has quit [Ping timeout: 260 seconds]
<RP>
kergoth: yes, I think devtool already manipulates that a bit?
<RP>
or perhaps it messes with gitignore, I can't remember offhand, I know it does something
camus has quit [Ping timeout: 255 seconds]
camus has joined #yocto
Guest9 has joined #yocto
xmn has joined #yocto
<Guest9>
Hello everybody, I have a problem building Xorg, I keep getting an error "Glamor for Xorg requires gbm >= 10.2.0" and have difficulties to find the correct libgbm version, https://layers.openembedded.org/ semms to only carry 10.0.0
Guest9 is now known as Xorg
Lihis has quit [Remote host closed the connection]
Lihis has joined #yocto
olani has quit [Ping timeout: 246 seconds]
uniqdom has joined #yocto
<uniqdom>
Hello, I have deleted the output files in tmp/deploy. I wanted to rebuild the output image files again, but I'm facing this 2 errors: https://paste.debian.net/1265637/
<uniqdom>
I have tried with "bitbake -c clean core-image-base" and "bitbake -c cleansstate core-image-base", but both 2 errors are there
_lore_ has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
<uniqdom>
I was building dey-image-qt, and after that, I built core-image-base. It went fine. Then I deleted the files in tmp/deploy and I couldn't rebuild core-image-base, because of those 2 error
vquicksilver has joined #yocto
_lore_ has joined #yocto
seninha_ has quit [Quit: Leaving]
seninha has joined #yocto
Guest16 has joined #yocto
Guest16 has quit [Client Quit]
<uniqdom>
should I just cleanall?
nemik has quit [Ping timeout: 246 seconds]
nemik has joined #yocto
nemik has quit [Ping timeout: 255 seconds]
nemik has joined #yocto
Xorg54 has joined #yocto
Xorg has quit [Quit: Client closed]
Xorg54 is now known as Xorg
dmoseley has quit [Ping timeout: 260 seconds]
dmoseley has joined #yocto
Xorg has quit [Ping timeout: 260 seconds]
Mistyron has joined #yocto
<Mistyron>
Hi, I want to add X11 to my build but keep running into an issue, I keep getting "error: Glamor for Xorg requires gbm >= 10.2.0", the complete message can be seen here: https://pastebin.com/kVePqPGe
dmoseley_ has joined #yocto
dmoseley has quit [Ping timeout: 246 seconds]
yashraj466 has quit [Quit: Client closed]
uniqdom has quit [Ping timeout: 260 seconds]
dmoseley has joined #yocto
uniqdom has joined #yocto
dmoseley_ has quit [Ping timeout: 260 seconds]
dmoseley_ has joined #yocto
dmoseley has quit [Ping timeout: 265 seconds]
uniqdom has quit [Read error: Connection reset by peer]
dmoseley_ has quit [Ping timeout: 260 seconds]
uniqdom has joined #yocto
dmoseley has joined #yocto
dmoseley_ has joined #yocto
dmoseley has quit [Ping timeout: 260 seconds]
dmoseley_ has quit [Ping timeout: 255 seconds]
dmoseley has joined #yocto
dmoseley_ has joined #yocto
dmoseley has quit [Ping timeout: 252 seconds]
olani has joined #yocto
dmoseley has joined #yocto
dmoseley_ has quit [Ping timeout: 246 seconds]
uniqdom has quit [Quit: Leaving]
<jclsn>
Why is populate_sdk command trying to build qtwayland although I have Qt set to run with egl?
<jclsn>
Getting this message
<jclsn>
Feature 'xcomposite-glx' was enabled, but the pre-condition 'features.wayland-client && features.opengl && !features.opengles2 && features.xlib && libs.xcomposite && libs.glx' failed
tangofoxtrot has quit [Remote host closed the connection]
seninha has quit [Quit: Leaving]
sakoman has quit [Quit: Leaving.]
mvlad has quit [Remote host closed the connection]
tangofoxtrot has joined #yocto
sakoman has joined #yocto
starblue1 has joined #yocto
azcraft has quit [Remote host closed the connection]
invalidopcode has quit [Read error: Connection reset by peer]
invalidopcode has joined #yocto
florian_kc has quit [Ping timeout: 252 seconds]
florian_kc has joined #yocto
tangofoxtrot has quit [Remote host closed the connection]
ecdhe has quit [Read error: Connection reset by peer]
ecdhe has joined #yocto
tangofoxtrot has joined #yocto
seninha has joined #yocto
<tlwoerner>
RP: after more experimenting i'm left with something i can't explain
<tlwoerner>
using MACHINEOVERRIDES or DISTROOVERRIDES, if i set or unset the override and build the full image (core-image-base) the config file contains the correct contents
<tlwoerner>
but the contents of $TMPDIR/work/$MACHINE/<package>/<version>/packages-split/<package> don't change based on the overrides
<RP>
tlwoerner: hashequivalence playing a part maybe?
<tlwoerner>
RP: ok. let me try with a clean build and fresh sstate
<RP>
tlwoerner: sometimes these can make interesting experiments in lib/oeqa/selftest/cases/ since you can run them as tests that way
camus has quit [Remote host closed the connection]
florian_kc has quit [Ping timeout: 260 seconds]
camus has joined #yocto
DvorkinDmitry has joined #yocto
<DvorkinDmitry>
I have weak set for IMAGE_ROOTFS_EXTRA_SPACE ?= in my image file. but it doesn't work case bitbake.conf has ?= 0 for this var. I want to have WEAK declaration in my image file that will work and to let user redefine it in his local.conf
<RP>
DvorkinDmitry: it is hard to get this to work nicely :(
<DvorkinDmitry>
thanks, RP! what is the most beautiful solution?
money has joined #yocto
money is now known as Guest1026
Guest1026 has quit [Changing host]
Guest1026 has joined #yocto
Guest1026 is now known as money
<RP>
JPEW: I'm now confused. It seems there is something I'm missing about python dicts and threading. https://autobuilder.yocto.io/pub/failed-builds-data/bitbake-cookerdaemon5.log shows it looping in idle. It shows an idle handler was added so the count (idle functions) should be 2 but it isn't and it stalls :(
<RP>
DvorkinDmitry: I'm not sure you can make it work like that easily without changing the default in bitbake.conf to be set differently, probably by making that a ??=
<RP>
DvorkinDmitry: I've sent a few emails to the arch list about the problems we have with defaults, I still don't know what a good solution which covers all the needs looks like