tgamblin has quit [Remote host closed the connection]
tgamblin has joined #yocto
prabhakalad has quit [Ping timeout: 252 seconds]
prabhakalad has joined #yocto
tgamblin_ has joined #yocto
tgamblin has quit [Ping timeout: 252 seconds]
tgamblin_ is now known as tgamblin
sakman has joined #yocto
Daanct12 has joined #yocto
<sotaoverride>
why is my build still complaining about check sums if I my local.conf has this: BB_STRICT_CHECKSUM:forcevariable = "0"
<khem>
RP: the cross issue is commonly seen you dont see it with binutils-cross because it sets TOOLCHAIN = "gcc" explicitly
<khem>
if I repeat this with gdb-cross and go-cross it works there too, but :toolchain-clang override is wrongly used
<khem>
which is problematic
savolla has quit [Ping timeout: 252 seconds]
thomas_34 has joined #yocto
ardo has joined #yocto
aardo has quit [Ping timeout: 248 seconds]
thomas_34 has quit [Quit: Client closed]
pidge_ has joined #yocto
pidge has quit [Ping timeout: 276 seconds]
goliath has joined #yocto
savolla has joined #yocto
rfuentess has joined #yocto
ablu-linaro has quit [Read error: Connection reset by peer]
brgl has quit [Read error: Connection reset by peer]
jetm has quit [Read error: Connection reset by peer]
jetm has joined #yocto
rob_w has joined #yocto
brgl has joined #yocto
ablu-linaro has joined #yocto
Sreejith has joined #yocto
Sreejith has quit [Client Quit]
savolla has quit [Quit: WeeChat 4.6.3]
savolla has joined #yocto
zeemate has joined #yocto
thomas_34 has joined #yocto
<thomas_34>
Good morning, has someone similar issue? If I do bitbake imagename -c do_fetch , and THEN bitbake imagename it still executes a lot of do_fetch jobs of various packages. Why is that?
dgriego has quit [Ping timeout: 252 seconds]
dgriego has joined #yocto
<Fr4nk>
thomas_34: yes sometimes it happens, because the fetch command fails to satisfy all the dependencies required by the do_image. To fetch the entire dependency tree you can try bitbake imagename --runall=fetch
<thomas_34>
Thank you Fr4nk, I try that
dgriego_ has joined #yocto
dgriego has quit [Ping timeout: 252 seconds]
dturull has joined #yocto
ptsneves has joined #yocto
ptsneves has quit [Ping timeout: 276 seconds]
davidpartain has joined #yocto
florian_kc has joined #yocto
alperak has joined #yocto
olani_ has quit [Ping timeout: 276 seconds]
LainIwakura has joined #yocto
savolla has quit [Ping timeout: 252 seconds]
MarcWeDLM has joined #yocto
florian_kc is now known as florian
vthor has joined #yocto
vthor has quit [Changing host]
vthor has joined #yocto
vthor_ has quit [Ping timeout: 276 seconds]
davidpartain has quit [Quit: Client closed]
<RP>
mischief: that is a static lib and you're asking for dynamic linking (.so) ?
mihai has quit [Quit: Leaving]
<mischief>
there isn't one in the sysroot
<RP>
mischief: which might be the issue?
<mischief>
i dont know, not had to deal with this deep toolchain voodoo before
<mischief>
seems intentional there's no shared object?
savolla has joined #yocto
MarcWeDLM has quit [Quit: Client closed]
<mischief>
hm, it just seems confused about the right path to use. it definitely wants libgcc.a, since i passed --verbose to the linker
<mischief>
attempt to open /home/mischief/src/poky/build/tmp/work/riscv64-nativesdk-pokysdk-linux/nativesdk-glibc/2.41+git/recipe-sysroot-native/usr/bin/riscv64-pokysdk-linux/../../lib/riscv64-pokysdk-linux/gcc/riscv64-pokysdk-linux/14.2.0/libgcc.a failed
<mischief>
there is no libgcc.a in all of recipe-sysroot-native
<RP>
mischief: libgcc-initial is only used to build glibc
<RP>
mischief: is the build failing in glibc?
<mischief>
correct. ERROR: Task (virtual:nativesdk:/home/mischief/src/poky/meta/recipes-core/glibc/glibc_2.41.bb:do_compile) failed with exit code '1'
<RP>
ah, nativesdk-glibc
<RP>
it sounds like the compiler search paths may be messed up. I'd also check it is trying to link statically
<mischief>
it seems very much like it wants the static library. the above is from adding -Wl,--verbose to the cc invocation, and it also prints.. "| /home/mischief/src/poky/build/tmp/work/riscv64-nativesdk-pokysdk-linux/nativesdk-glibc/2.41+git/recipe-sysroot-native/usr/bin/riscv64-pokysdk-linux/../../libexec/riscv64-pokysdk-linux/gcc/riscv64-pokysdk-linux/1
<mischief>
4.2.0/ld.bfd: have you installed the static version of the gcc library ?
<RP>
did it look through the right paths for it?
<RP>
mischief: did glibc itself build ok? Is this just the sdk glibc?
<RP>
I don't know of anyone who has built a riscv sdk on riscv before
<mischief>
i can start a normal non-sdk build and see what happens
<mischief>
i was just trying to build uninative-tarball since that was the first thing that broke when i first started this, lack of uninative
<mischief>
there is no libgcc.a in recipe-sysroot-native at all, only in recipe-sysroot.. not clear if thats the one it needs to use
<RP>
mischief: the build should be able to work without uninative. We could build one on a different arch
<mischief>
fired off bitbake glibc, we will see if it works. going to take a while..
<RP>
mischief: sounds good. I'm just trying to narrow down the problem
<RP>
mischief: what SDKMACHINE did you set btw?
<RP>
khem: still around? I tried to reproduce this but if I set TOOLCHAIN ?= "clang", it will try and build gdb-cross-XXX with clang, as expected
<mischief>
RP: default value? i didn't change anything
<RP>
khem: where are you setting it to gcc?
<RP>
mischief: there is a meta/conf/machine-sdk/riscv64.conf so I guess it used that. I hadn't realised we had one!
dturull has quit [Ping timeout: 272 seconds]
<thomas_34>
Another question about bitbake and gitlab CI: Bitbake runs within a docker container which is executed by gitlab pipeline. Within the container, I setup git this way, that I'm able to access my own git repos during the build process: git config --global url."https://gitlab-ci-token:${CI_JOB_TOKEN}@mygitlabserver.com/".insteadOf
<thomas_34>
"ssh://g...@mygitlabserver.com:2289/"
<thomas_34>
However, bitbake fails to fetch my own private repos. I think its because bitbake uses a "clean environment" which does not use the insteadOf solution to access.
<thomas_34>
Has anyone done something like this and knows, how I tell git, which bitbake uses, to use that git config which replaces the url with the CI_JOB_TOKEN?
<RP>
thomas_34: I thought you might be right but looking at bitbake/lib/bb/fetch2/git.py at the bit where it looks at FETCHCMD_git, it doesn't look like it sets anything that would cause it to ignore the global git config
<thomas_34>
RP, interesting. Thanks for pointing me there. In worst case I can add there debug output to trace it further down. Googling lead me to this: https://groups.google.com/g/kas-devel/c/F8w09z5l7zk , which I dont understand fully, but I think they are passing some ENV-Variables to the bitbake process.
<RP>
thomas_34: ah, what is probably happening is that CI_JOB_TOKEN is being cleared out the environment
<thomas_34>
ahhhh :) nice RP! Sharp thought
<thomas_34>
Thank you - I bet thats it
<RP>
thomas_34: see bitbake/lib/bb/fetch2/__init__.py and FETCH_EXPORT_VARS for some that we do pass through
<thomas_34>
Thanks man! I think I will get it working from here
<mischief>
you can get around this with .netrc
<thomas_34>
mischief, thanks I dont know anything about .netrc. Will google it
LainIwakura has quit [Quit: Client closed]
ablu has quit [Ping timeout: 252 seconds]
ablu has joined #yocto
<thomas_34>
Hmm RP, I think the the guess with cleared out CI_JOB_TOKEN is not the issue. The git config command replaces the ENV-variable with the actual value, which happens before I start the bitbake call. I'll debug into this with some print output. I tried also different variations with .netrc beside of that. Also did not work.
<RP>
thomas_34: fair enough, I was just trying to give pointers on where to look
jclsn has joined #yocto
leon-anavi has joined #yocto
davidpartain has joined #yocto
LainIwakura has joined #yocto
dturull has joined #yocto
<Fr4nk>
Is there an automatic recipe to generate the third-part licence list, as on Android "third-party menu"?
chep has quit [Read error: Connection reset by peer]
chep` is now known as chep
zeddii has quit [Ping timeout: 244 seconds]
wooosaiiii has quit [Remote host closed the connection]
wooosaiiii has joined #yocto
savolla has quit [Ping timeout: 248 seconds]
savolla has joined #yocto
rfuentess has quit [Remote host closed the connection]
wooosaiiii has quit [Remote host closed the connection]
wooosaiiii has joined #yocto
rfuentess has joined #yocto
<thomas_34>
RP, I got it. The git configuration about insteadOf was a little bit off. I had to add a second entry... No other change was necessary
belsirk has joined #yocto
rfuentess has quit [Ping timeout: 276 seconds]
Daanct12 has quit [Quit: WeeChat 4.6.3]
<RP>
glad you got it sorted
goliath has quit [Quit: SIGSEGV]
rob_w has quit [Remote host closed the connection]
dvergatal has quit [Ping timeout: 244 seconds]
belsirk has quit [Remote host closed the connection]
Tyaku has quit [Ping timeout: 248 seconds]
frgo_ has quit [Read error: Connection reset by peer]
frgo has joined #yocto
goliath has joined #yocto
dturull has quit [Quit: Client closed]
<paulg>
I see the "before vs. after host dmesg comparison" is now a reality - nice.
<paulg>
Now someone just needs to ensure a couple of the AB bots are using it.
florian has quit [Quit: Ex-Chat]
<tlwoerner>
AdrianF: i guess i have one more comment regarding the fitimage changes... in meta-rockchip, for example, some MACHINEs are using a fitimage and others aren't
<tlwoerner>
an easy way to figure out which MACHINEs were and which weren't was to check the KERNEL_IMAGETYPE variable
<tlwoerner>
but that trick is now gone. i guess i'll have to create my own variable to track which ones do and which don't
jmd has joined #yocto
<tlwoerner>
AdrianF: is there any variable being set that i could use to track which builds are using a fitimage and which aren't? if not that's fine, i just don't want to be making variables unnecessarily
thomas_34 has quit [Quit: Client closed]
zeddii has joined #yocto
ptsneves has joined #yocto
zeddii has quit [Ping timeout: 244 seconds]
<tlwoerner>
i'm stumped on how to get this new fitimage thing to work
<tlwoerner>
if i set PREFERRED_PROVIDER_virtual/kernel = "linux-yocto-fitimage"
<tlwoerner>
then it doesn't build because linux-yocto-fitimage doesn't provide PREFERRED_PROVIDER_virtual/kernel
leon-anavi has quit [Quit: Leaving]
frgo has quit [Quit: Leaving...]
frgo has joined #yocto
tgamblin has quit [Ping timeout: 248 seconds]
tgamblin has joined #yocto
zeddii has joined #yocto
<AdrianF>
tlwoerner: PREFERRED_PROVIDER_virtual/kernel = "linux-yocto-fitimage" is not how it works. PREFERRED_PROVIDER_virtual/kernel = "linux-yocto" remains.
<AdrianF>
I'm just going to test PREFERRED_PROVIDER_virtual/kernel = "linux-yocto-tiny" which should work as well.
<RP>
mischief: nice. That narrows the scope of the problem a bit I guess and means sdk should be close too
<RP>
khem: around?
sizzop has quit [Remote host closed the connection]
tgamblin_ has joined #yocto
tgamblin has quit [Ping timeout: 252 seconds]
Kubu_work has quit [Quit: Leaving.]
tgamblin_ is now known as tgamblin
<rburton>
khem: how hard/easy do you think it would be to enable libgdiagnostic in gcc? it depends on --enable-host-shared which gcc suggests is not a great idea for the normal build, so i guess that means another split-out recipe like libgcc etc?
florian has joined #yocto
jmd has quit [Remote host closed the connection]
<RP>
rburton: have a look and see if you could configure gcc-runtime to enable it
<RP>
rburton: that code is able to iterate compiler libs
<mischief>
i started up core-image-minimal again. 57%!
<mischief>
something is causing single bit errors, making GCC ICE and other fun stuff like Exception: KeyError: "Unknown property 'htpps://rdf.openembedded.opg/spdx/3.0/alias'"