<kevinrowland>
Am I right that there used to be a note in the mega manual (or the bitbake manual) about how bitbake treats variable dereference with curly braces different than variable dereference without? IIRC `${MYVAR}` is expanded by bitbake but `$MYVAR` is left to be expanded by the shell? I can't find anything about it in the docs right now.
<NilsH>
Hi there, i have a recipe that installs a header file with do_install into /usr/include, another recipe requires this file to build - this works fine. However unfortunately this header file does not make it into the sdk, what am i missing here? Id like to make it available in the sdk for manual cross compiling.
AdrianF has joined #yocto
invalidopcode1 has quit [Remote host closed the connection]
invalidopcode1 has quit [Remote host closed the connection]
invalidopcode1 has joined #yocto
prabhakarlad has joined #yocto
amitk_ has joined #yocto
PabloSaavedra[m] has quit [Quit: You have been kicked for being idle]
lukma has quit [Ping timeout: 260 seconds]
hays has quit [Read error: Connection reset by peer]
florian has joined #yocto
tunahan has joined #yocto
<tunahan>
Hi, I'm getting an error on bitbake. I think it has to do with not finding /sys/types.h but I have libc6-dev installed on my compiling host. I'm using linux-dummy in the image I created, could it have something to do with it? I also share the highlight and raw links of the log file. log-rawlink: https://www.paste.tc/raw/error-log log-link: https://www.paste.tc/error-log
<rburton>
tunahan: looks like luajit doesn't want to be cross-compiled
<rburton>
tunahan: the key line being gcc -m32 -Wall -I. -DLUAJIT_TARGET=LUAJIT_ARCH_arm -DLJ_ARCH_HASFPU=1 -DLJ_ABI_SOFTFP=0 -isystem/media/tunahan/f1cceee6-ac24-4956-9e80-7ce648ac31e4/yoctoproject-kirkstone/build/tmp/work/armv7vet2hf-neon-mepos-linux-gnueabi/luajit/2.1.0~beta3-210112-r0/recipe-sysroot-native/usr/include -O2 -pipe -c -o host/buildvm_fold.o host/buildvm_fold.c
<rburton>
where its looking in the native sysroot for system headers, which is wrong
nemik has quit [Ping timeout: 276 seconds]
nemik has joined #yocto
<tunahan>
rburton: How should I go about solving it?
<tunahan>
rburton: I examined the bb file a bit, but I did not understand how to edit it. Moreover, I think this package is installed because other packages want it (because they are dependent). I don't know if I can go on without this package.
nemik has quit [Ping timeout: 252 seconds]
nemik has joined #yocto
<rburton>
tunahan: there's not a lot that actually uses luajit so the first step would be determining if you do need it after all. in meta-oe all i can see is sysbench, sysdig, and mpv, and the last two its optional.
goliath has joined #yocto
<tunahan>
rburton: then let's try to exclude it from loading it and try it. will other things give an error? Since I use packagegroup, I don't know exactly what's in them in this compilation. So is there a command to exclude only this package? I'll add it to local.conf.
<tunahan>
building it*
<rburton>
tunahan: SKIP_RECIPE[luajit] = "skip" will make bitbake refuse to build it, assuming you're using a recent release
<tunahan>
rburton: yes, when I added that, I got an error: ERROR: Required build target 'core-image-minimal-xfce' has no buildable providers.
<rburton>
why is 'minimal' xfce pulling in the 'full' gui packagegroup
<rburton>
but mpv's dependency on luajit is optional, so just turn it off
<tunahan>
rburton: obviously i tried to create a full xfce desktop so i tried such a method. With minimal xfce, I wanted to build what is needed for xfce and full gui and other packages.
<rburton>
minimal and full in the same thing make confusion :)
<rburton>
but easily worked around until you want to dig into why luajit doesn't build
alessioigor has quit [Quit: alessioigor]
tunahan has quit [Quit: tunahan]
alessioigor has joined #yocto
tunahan has joined #yocto
<tunahan>
rburton: When you turn off the lua part in mpv, the build continues nicely now. Let's see if there will be any more mistakes from this mess in the future, with a total of 51% :D
<tunahan>
when i turn off*
DvorkinDmitry has joined #yocto
<tunahan>
sorry for my english i'm just learning i can spell it wrong sometimes
<rburton>
obviously you can fiddle the packagegroups and add/remove what you actually need.
invalidopcode1 has quit [Remote host closed the connection]
<tunahan>
rburton: Is there an easy way for these suggestions? I guess one by one, I need to go into layers.openembedded and look.
invalidopcode1 has joined #yocto
<DvorkinDmitry>
I have a lot of recipes in my layer. One of them is not mentioned in myimage.bb and hase no depeendency in the recipes required for myimage.bb. But if I run bitbake myimage the recipe GIT checked for new version because of AUTOREV. Sometimes its git is not available and gives annoying error. How to prevent downloading the recipe when it is not in dependency chain?
<ptsneves>
Hey people! Today i would like to thank the community for the tinfoil API. Very much appreciated.
azcraft has joined #yocto
<rburton>
DvorkinDmitry: autorev means it has to evaluate the version on _parse_ which is why autorev is not recommended. what do you mean by git not being available? the binary, or the git server? potentially, a bug needs to be filed.
hays has joined #yocto
<DvorkinDmitry>
rburton, from time to time they git server is glitching and stops handling the connections. althrough the recipe is not required by anybody the bitbake is trying to access its git and fails. It is usually the short period of time, when the git is glitching, but when you run the build for the night and find that if failed for stupid reason in the morning it is annoying
<RP>
DvorkinDmitry: until it parses, bitbake can't know myimage.bb doesn't reference it since the recipe in question could override some of your others. Disable autorev and that recipe in those builds?
<rburton>
yeah i'd definitely make the recipe have fixed SHAs and let people opt-in to autorev if they want
<DvorkinDmitry>
RP, I disabled AUTOREV and it runs. I've been using the certain branch and I need the most fresh version of this branch if I'm building the image with this recipe. But seems, che check is done at parse stage and there are no way not to access git before it creates the dependency tree?
<RP>
DvorkinDmitry: If you include that recipe in the build, bitbake *has* to parse it to know whether anything in it should override something in your targets
<RP>
DvorkinDmitry: you could add a BBMASK to however you're triggering one image vs the other?
<RP>
if you're trying to build both images in a given build, it just won't work I'm afraid
<RP>
it can't
<DvorkinDmitry>
RP, BBMASK? oh! I'm building ` image at one build. How to use it then?
<RP>
DvorkinDmitry: or just disable the autorev in that build?
amitk__ has joined #yocto
<DvorkinDmitry>
RP, bad idea to disable autorev. need to remember about it and it will be also annoying :)
amitk has quit [Ping timeout: 268 seconds]
<ernstp>
was there something to help with python rdepends, ie. python requirements, for making and updating recipes?
rfuentess has quit [Remote host closed the connection]
<rburton>
nothing that exists, afaik
<rburton>
should be easy to write though :)
brazuca has quit [Quit: Client closed]
<ernstp>
rburton: right, yes! :-)
rob_w has quit [Quit: Leaving]
seninha has joined #yocto
seninha has quit [Remote host closed the connection]
seninha has joined #yocto
xmn has joined #yocto
vladest1 has joined #yocto
vladest has quit [Ping timeout: 240 seconds]
vladest1 is now known as vladest
kscherer has joined #yocto
eFfeM has joined #yocto
tunahan has quit [Quit: tunahan]
tunahan has joined #yocto
<eFfeM>
Hi, I got a question if it was possible to fetch from git using only the commit id without specifying the branch. Is that possible?
<eFfeM>
(idea is ot just checkout using a hash)
tunahan has quit [Quit: tunahan]
tunahan has joined #yocto
thomasd13 has quit [Ping timeout: 240 seconds]
eFfeM has quit [Quit: Client closed]
nemik has quit [Ping timeout: 248 seconds]
nemik has joined #yocto
camus has quit [Remote host closed the connection]
nemik has quit [Ping timeout: 255 seconds]
nemik has joined #yocto
Spinola has joined #yocto
eFfeM has joined #yocto
<eFfeM>
oops did I get a reply on my question? somehow I got disconnected
<Spinola>
Hi Guys, back again with still the same basic problem. I'm not able to build an out of three kernel module (kirkstone branch) Used the hello-mod skeleton. And just added a few lines to include the time.h. But this include file can't be found by the build system! What is going on?
<Spinola>
Can anybody try this? and verify the behaviour I'm getting?
amitk_ has quit [Ping timeout: 265 seconds]
tunahan has quit [Quit: tunahan]
tunahan has joined #yocto
prabhakarlad has quit [Quit: Client closed]
<mckoan>
Spinola: #include <linux/time.h>
prabhakarlad has joined #yocto
sakoman has joined #yocto
Xagen has joined #yocto
invalidopcode1 has quit [Remote host closed the connection]
invalidopcode1 has joined #yocto
<Chocobo>
I think something like this will work for me (where TAG is a defined variable): SRCREV = "${@ TAG if not bb.utils.to_boolean(d.getVar('BB_NO_NETWORK')) else "ad0131f49db7391c8af026172b3b809560cee6ec" }"
<Spinola>
mckoan: Using #include <linux/time.h> gives the same errors!
davidinux has quit [Ping timeout: 255 seconds]
davidinux has joined #yocto
prabhakarlad has quit [Quit: Client closed]
<Spinola>
mckoan: I think time.h is part of the standard c library, why isn't it being found?
<Spinola>
mckoan: The file "time.h" is present in the recipe-sysroot (usr/include)
<JaMa>
chicken and egg issue? as kernel doesn't depend on glibc?
<LetoThe2nd>
Spinola: inside the kernel, you don't have the standard library.
<LetoThe2nd>
Spinola: only kernel-internal APIs are available.
<Spinola>
LetoThe2nd: I'm investigating a problem concerning a kernel module that did build fine on Rocko, but fails to build on Kirkstone. In the sources of this module #include <time.h> is present?!?
<LetoThe2nd>
Spinola: maybe the path changed? check out elixier.bootlin.com for the matching kernel version. Those APIs change occasionally.
<qschulz>
wasn't rocko using a shared sysroot?
<Spinola>
LetoThe2nd: Is the link correct?
<qschulz>
Spinola: elixir.bootlin.com
<LetoThe2nd>
qschulz: i think not, should have arrived with pyro
<qschulz>
LetoThe2nd: ok, because otherwise could be a case of a recipe not migrated to per-recipe sysroot and looking at the wrong path
<LetoThe2nd>
qschulz: yup
<Spinola>
LetoThe2nd: in both 4.9.87 and 6.1.14 the file time.h is located at the same path include/uapi/linux/time.h
<zeddii>
sakoman. I just sent a decent size queue of -stable updates your way. We may need lttng-module bumps as well, but I wanted to at least get the linux-yocto bumps out and visible.
alessioigor has quit [Quit: alessioigor]
<sakoman>
zeddii: thanks much! Will test shortly
alessioigor has joined #yocto
seninha has quit [Remote host closed the connection]
nemik has quit [Ping timeout: 240 seconds]
nemik has joined #yocto
Spinola has quit [Quit: Leaving]
<mckoan>
Spinola: building with #include <linux/time.h> was successful on my Qemuarm linux 5.4.230
<zwelch>
It seems that systemd uses a hard-coded home directory for the root user (/root), but Yocto defines ROOT_HOME in bitbake.conf differently (/home/root). Is there a good reason why the Yocto systemd recipe does not include a patch to set this path from ROOT_HOME?
<rburton>
zwelch: feel free to write a test a replacement if its easy to change systemd in a way that doesn't involve 100 seds
risca has quit [Ping timeout: 276 seconds]
seninha has joined #yocto
seninha has quit [Remote host closed the connection]
seninha has joined #yocto
diego_r has quit [Quit: Konversation terminated!]
prabhakarlad has quit [Quit: Client closed]
<zwelch>
lol... I knew that I couldn't be the first one to run across this. My workaround here was to symlink `/root` to `$ROOT_HOME`, but that only works for us because both are on the same volume (i.e. it's not a general solution). Since the symlink appears to solve the problems that we were seeing, I am disinclined to get my hands dirty with systemd, since this seems like a great example of their hostility to anything that does not mesh with
<zwelch>
their narrow world view. I have been pretty ambivalent about systemd, but this seems to demonstrate why some people really hate it.
<zwelch>
rburton, anyway, thanks for the reference. :)
brazuca has joined #yocto
risca has joined #yocto
florian has quit [Quit: Ex-Chat]
ecdhe has quit [Read error: Connection reset by peer]
tunahan has quit [Quit: tunahan]
tunahan has joined #yocto
ecdhe has joined #yocto
brazuca has quit [Quit: Client closed]
invalidopcode1 has quit [Remote host closed the connection]
invalidopcode1 has joined #yocto
prabhakarlad has joined #yocto
florian_kc has joined #yocto
tangofoxtrot has quit [Remote host closed the connection]
tangofoxtrot has joined #yocto
Guest18 has joined #yocto
kevinrowland has joined #yocto
florian_kc has quit [Ping timeout: 276 seconds]
AdrianF has quit [Quit: Leaving]
florian_kc has joined #yocto
<fury>
how do i look up the current PACKAGECONFIG for bluez and maybe have it tell me where it's getting altered? i suspect something somewhere is screwing it up and removing things that need to be there :/
<fury>
e.g. the default bluez PACKAGECONFIG includes a2dp-profiles and avrcp-profiles, but i seem to have a copy of bluez in my image that does not even have the .Media1 endpoint necessary for registering a2dp endpoints
<zeddii>
fury. it's just a variable, so bitbake -e is one way to see all modifications to the variable.
<fury>
ah ok, cool. and it shows me which lines of which files are responsible for the fully expanded value. nice. thanks!
tlwoerner has joined #yocto
<fury>
sadly that was not the answer, it does have --enable-a2dp and --enable-avrcp
<fury>
was hoping it would be as simple as adding a2dp-profiles and avrcp-profiles back in, but it has to be trickier than that :D
invalidopcode1 has quit [Remote host closed the connection]
invalidopcode1 has joined #yocto
kscherer has quit [Quit: Konversation terminated!]
Guest18 has quit [Ping timeout: 260 seconds]
Vonter has quit [Ping timeout: 265 seconds]
<halstead>
LetoThe2nd: I see you accepted. If you finish confirmation I can proceed with getting sponsorship. :)
<halstead>
LetoThe2nd: Perfect. You're an owner now.
sakoman has quit [Quit: Leaving.]
brazuca has joined #yocto
sakoman has joined #yocto
invalidopcode1 has quit [Remote host closed the connection]