Guest52 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Guest52 has joined #yocto
Guest52 has quit [Client Quit]
GillesM has joined #yocto
GillesM has quit [Client Quit]
GillesM has joined #yocto
mckoan|away has joined #yocto
GillesM has quit [Quit: Leaving]
vd has quit [Quit: Client closed]
vd has joined #yocto
xmn has quit [Quit: ZZZzzz…]
jmiehe1 has joined #yocto
jmiehe has quit [Ping timeout: 260 seconds]
jmiehe1 is now known as jmiehe
kpo has quit [Read error: Connection reset by peer]
kpo has joined #yocto
amitk has joined #yocto
tre has joined #yocto
vladest has quit [Ping timeout: 252 seconds]
camus has quit [Ping timeout: 265 seconds]
camus has joined #yocto
roussinm has quit [Quit: WeeChat 3.3-dev]
alessioigor has joined #yocto
alessioigor has quit [Remote host closed the connection]
Vonter has quit [Ping timeout: 252 seconds]
Vonter has joined #yocto
timbert[m] has joined #yocto
Guest31_niclas has quit [Quit: Client closed]
Emantor_ has joined #yocto
bantu_ has joined #yocto
rsalveti_ has joined #yocto
Emantor has quit [Ping timeout: 245 seconds]
Emantor_ is now known as Emantor
rsalveti_ is now known as rsalveti
<JosefHolzmayrThe>
yo dudX
Schlumpf has joined #yocto
<JosefHolzmayrThe>
Ch^W_: I'm no expert there, but shouldn't diffsigs help with that?
tnovotny has joined #yocto
zpfvo has joined #yocto
davidinux has joined #yocto
rfuentess has joined #yocto
agrue has joined #yocto
Guest44 has joined #yocto
<Guest44>
hello guys, I need to use unzip inside a recipe, so I have added DEPENDS += "unzip-native" but I'm facing to | unzip: not found | I have take a look inside core/meta/conf/bitbake.conf and I didn't find unzip :( any idea ?
<qschulz>
Ch^W_: yes, that's one thing that isn't supported actually AFAIR
<qschulz>
only the filename is used in the hashmap, not the path
<JosefHolzmayrThe>
qschulz: thats what I think too, but if the subdir is evalutated to a different value, shouldn't that change the hash of SRC_URI?
<qschulz>
I seem to recall you're not supposed to use absolute paths in subdir
<qschulz>
and ${S} is one
<qschulz>
but otherwise yeah, I don't know
Ad0 has quit [Ping timeout: 268 seconds]
<qschulz>
bitbake dumpsig should help identify what is watched
gsalazar_ has joined #yocto
<qschulz>
Guest44: this looks correct to me. Check in your recipe's sysroot-native that you have unzip binary there and that it's compiled for the arhitecture of your build machine (likely x64_64 I assume?)
<Guest44>
qschulz thx for support, i don't have this recipe (only unzip_6.0.bb) and yes it's on a basic 20.04 Ubuntu build machine
<Guest44>
more info, calling bitbake unzup-native doesn't show any error
<qschulz>
Guest44: you added DEPENDS += unzip-native to a recipe right?
grma has joined #yocto
<Guest44>
@qs
<Guest44>
oops
<Guest44>
qschulz yes and I have simply added do_unzip method ( addtask unzip before do_install after do_fetch) method is called but unzip not found
<RP>
Hmm, first failed "perf" build on the AB in over 300 builds
<qschulz>
Guest44: after do_fetch is incorrect
<Guest44>
qschulz Ha ? there is some rules to know about that ?
<qschulz>
it should be after do_prepare_recipe_sysroot
<qschulz>
Guest44: the sysroot is populated by this task
<qschulz>
so it is possible (and likely?) that your do_unzip task is run before do_prepare_recipe_sysroot which means unzip-native is not avaiulable to the recipe yet
<RP>
635 builds and 5 months ago was the last failure
<qschulz>
why do you need unzip ?
<Guest44>
qschulz ok so first thx for the link, sec that works now :) :)
<Guest44>
I'm using unzip to extarct Xilinx file
frieder has joined #yocto
bps has joined #yocto
bps has joined #yocto
pgowda_ has joined #yocto
<qschulz>
Guest44: that should be done automatically by Bitbake
<qschulz>
otherwise, I assume you could just do a rename of the file in a do_fetch_append
<qschulz>
to have a .zip extension so do_unpack knows it should unpack it
<Guest44>
qschulz that's a really good advice, i will try to follow this wy :)
<qschulz>
do_fetch is likely a python task, so you'll have to write the _append in python too
<qschulz>
or use do_fetch[postfunc] IIRC
<wCPO>
Shouldn't pseudo catch the getuid syscall? I'm having some trouble with a Go program (skopeo) where the syscall is returning the current user and not UID 0 when building.
Guest52 has joined #yocto
Guest52 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
florian has joined #yocto
leon-anavi has joined #yocto
florian has quit [Quit: Ex-Chat]
florian has joined #yocto
lucaceresoli has joined #yocto
<rburton>
wCPO: pseudo is a LD_PRELOAD that intercepts glibc calls. Go code typically hits the syscall interface directly, so you can't intercept it
<rburton>
build any go code you run at build time to use the libc, instead of hitting the syscalls directly
<Ch^W_>
qschulz: Thanks. That definitely makes it even harder than I thought to include source code in the same repository as recipes.
<wCPO>
rburton: make sense. Are you referring to CGO_ENABLED=1? because skopeo is already built with that
alessioigor has joined #yocto
alessioigor has quit [Client Quit]
<qschulz>
Ch^W_: that is something that is not recommended :)
<qschulz>
because you don't want to bind your software to a build system
Schlumpf has quit [Quit: Client closed]
Guest52 has joined #yocto
alessioigor has joined #yocto
<Ch^W_>
qschulz: Our software is exclusively used in our platform OS images, so we use bitbake recipes as a light wrapper around the usual make/cmake/ant/gradle/maven/etc. It generally works really well.
<qschulz>
Ch^W_: and for now you build your OS images with Yocto, but you could change in the future or add support for another buildsystem depending on client requirements. Hence why it is usually not recommended to mix both
<Ch^W_>
qschulz: That seems like a very poor reason to avoid a very simple check for a directory change in paths pointed to by file:// URLs.
<Ch^W_>
What about situations where there is a ton of metadata in a recipe that is not amenable to the expected flat layout/
<Ch^W_>
?
Ad0 has joined #yocto
alessioigor has quit [Quit: alessioigor]
Tokamak has quit [Read error: Connection reset by peer]
xmn has joined #yocto
Tokamak has joined #yocto
<RP>
Ch^W_: I just saw your file:// fetcher issue. It does sound like a bug but it may be harder to fix than you think :(
frieder has quit [Ping timeout: 252 seconds]
<wCPO>
Can I get the real user id from a recipe? (ex: the user running bitbake)
frieder has joined #yocto
CocoJoe has joined #yocto
<rburton>
it only pretends to be root inside pseudo tasks
jwillikers has joined #yocto
<qschulz>
Ch^W_: I think you should be able to work around this by using externalsrc pointing to your sources in-tree
<qschulz>
externalsrc uses git to detect changes
<qschulz>
Ch^W_: no, I was not giving a reason for not checking directory changes. I was just saying, regardless of what you encountered, that mixing sources and build systems is a bad idea.
<RP>
Ch^W_: also, which version of bitbake is that?
manuel1985 has quit [Ping timeout: 252 seconds]
Guest44 has quit [Quit: Client closed]
* RP
is sad the autobuilder didn't catch maximal eSDK build failures :(
Schlumpf has joined #yocto
CocoJoe has quit [Quit: Client closed]
xmn has quit [Ping timeout: 265 seconds]
<Guest52>
Hi everyone, i'm getting undefined reference to symbol 'ERR_reason_error_string@@OPENSSL_1_1_0' when compiling a package in a recipe. i'm using poky as a base for my image, which provides a recipe for openssl_1.1.1l . My best guess is that i need to provide a recipe with the exact version of openssl that the package uses. Is this right or is there a better alternative handling this issue?
Guest52 is now known as Pan5ky
<JosefHolzmayrThe>
RP: maximal esdk?
<RP>
JosefHolzmayrThe: you can generate two versions, one with and one without the sstate built in
* RP
just fast tracked a patch into master to try and stop the perf builds failing
manuel1985 has joined #yocto
<JosefHolzmayrThe>
ah ok
jwillikers has quit [Quit: jwillikers]
jwillikers has joined #yocto
jwillikers has quit [Client Quit]
Pan5ky has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Guest52 has joined #yocto
Schlumpf has quit [Quit: Client closed]
cperon has joined #yocto
<cperon>
Hi, just a question I have generated an .uefiimg for x86 is there a quick way to generate an USB installer, (for the moment I boot on a LiveUSB and dd the img)? I look at the WIC format but it's intendend to install the EXT4 rootfs no a full Uefiimg or maybe I'm wrong...
<mcfrisk>
trying to send a patch to yocto@yoctoproject.org using git send-email and response from server is "lb01.groups.io[45.79.81.153] said: 500 Invalid request 1-0-1". What's that about? Overriding --from to match commit author no longer allowed or what?
<landgraf>
mcfrisk: yocto@lists.yoctoproject.org
<landgraf>
mcfrisk: been there yesterday :(
frieder has quit [Remote host closed the connection]
<rburton>
its been @lists. for quite a while now :)
<landgraf>
rburton: few places still points to @y.o (psplash's README for example)
<mcfrisk>
meta-selinux needs an update to MAINTAINERS
<rburton>
landgraf: psplash fixed. anywhere else you know of?
<JPEW>
RP: Hmm, so the idea is to add a /./ to the path which doesn't affect path resolution, but gives the taskhash code something to split on?
jwillikers has joined #yocto
<Ch^W_>
RP: Wow, thanks! I will take a look at it.
<Ch^W_>
RP: Currently using bitbake 1.50.0 from hardknott.
tnovotny has quit [Quit: Leaving]
<rburton>
RP: that is pretty horrible
ptsneves has joined #yocto
florian_kc has quit [Ping timeout: 252 seconds]
roussinm has joined #yocto
<qschulz>
can file-checksums have knowledge of the directory in which a task is run (do_func[dir] IIRC?)
ptsneves has quit [Ping timeout: 264 seconds]
<qschulz>
in which case, you know that relative paths are relative to the aforementioned dir and can derive the path you need to save in the taskhash?
<qschulz>
I'm not sure where I'm going with this, I think it's a bit too late on a Friday for the brain :|
florian_kc has joined #yocto
camus has quit [Quit: camus]
camus1 has joined #yocto
camus1 is now known as camus
<vd>
how do I get the first element of a variable? FIRST = "${@d.getVar('FOO').split()[0]}" ?
<rburton>
yes
<rburton>
get the variable, split on whitespace, first element
<rburton>
bitbake datastore is typically just strings. that's not true, but we'll ignore that.
<qschulz>
getVar expands by default now right?
<rburton>
yes
<rburton>
"no"
<rburton>
"now" even
<qschulz>
hehehe
<rburton>
that's been five years now?
<qschulz>
a long time for sure :)
<vd>
yes it does, only getVarFlags doesn't expand by default
eko has joined #yocto
eko has left #yocto [#yocto]
rpcme has joined #yocto
<rpcme>
Hi all, I'm working on master for virtually everything, and switched machines with fresh clones (ubuntu 18.04 to 20.04), and I am getting a ton of basehash errors "The metadata is not deterministic and this needs to be fixed"... never seen this one before ... any ideas?
<rpcme>
ps. I'm not cloning poky, I'm using openembedded-core and bitbake directly.
zyga has quit [Remote host closed the connection]
zyga has joined #yocto
rfuentess has quit [Remote host closed the connection]
Tokamak has quit [Read error: Connection reset by peer]
<rpcme>
Ok nevermind, it must have been some inconsistency between openembedded-core and bitbake. I made sure that the rev was the same on both systems and I'm not getting the error any longer.
<qschulz>
should have said that there is a mapping between versions of both projects and that's what I meant by "kept in sync"
<kergoth>
Hmm, that sysroot_stage_all hang we saw at mentor with dunfell only occurs with the cpio from the native sysroot, not the host cpio. interesting
<rpcme>
qschulz: not sure how that's telling be which bitbake rev to take in conjunction with openembedded-core ... you must be saying "use the revs with the yocto release or proceed at your own peril" :D
hpsy has joined #yocto
<vd>
Since latest masters I get "usermod: prefix must be an absolute path" error from EXTRA_USERS_PARAMS = "usermod -P password root;" and indeed the manpage says that -P is for prefix and -p is for the password, but I don't get it, it used to work until now and even the documentation states -P...
<kergoth>
rpcme: as long as you're using a matching bitbake release *branch* it'll be compatible, don't necessarily have to match the rev, and the link he gave shows the branch, so..
<vd>
qschulz: kergoth: this link is missing a column for the openembedded-core branch, as well as the bitbake branch name for the honister row.
<kergoth>
vd: honister is brand new, most likely just master. and the oe-core branch name *is* the yocto release name, i.e. dunfell
zpfvo has quit [Remote host closed the connection]
<kergoth>
you're right though, could certainly be clearer
<kergoth>
not sure when that gets updated
manuel1985 has quit [Ping timeout: 260 seconds]
kiran_ has quit [Remote host closed the connection]
* vd
currently uses 'master' for all layers as well as bitbake and openembedded-core, because it this new syntax thing and latest fixes.
<rpcme>
just fyi ... I have added perl-module-locale and perl-modules but got build break in perl and needed to manually add back in the dbm packageconfig. If this is known that's ok - the change to remove dbm probably happened about a month ago
<RP>
qschulz: in summary, no, only the fetcher code knows what the correct prefix point is :/
<RP>
JPEW: that was the idea, yes
kanavin_ has joined #yocto
Guest52 has quit [Ping timeout: 246 seconds]
florian has quit [Quit: Ex-Chat]
tre has joined #yocto
florian_kc has quit [Ping timeout: 252 seconds]
ecdhe has joined #yocto
amitk has quit [Quit: leaving]
gsalazar_ has quit [Ping timeout: 264 seconds]
vladest has joined #yocto
drewfustini has joined #yocto
<drewfustini>
hello, anyone have suggestion about how to disable perl in the git client build?
<drewfustini>
to remove '--with-perl' but I am not sure what the best way is to accomplish that
hpsy has quit [Quit: Client closed]
lucaceresoli has quit [Ping timeout: 265 seconds]
<RP>
drewfustini: are you trying to do that from a bbappend?
<RP>
drewfustini: we'd likely take a patch to make perl a PACKAGECONFIG option
bps has quit [Ping timeout: 268 seconds]
dmoseley has joined #yocto
pgowda_ has quit [Quit: Connection closed for inactivity]
florian_kc has joined #yocto
angolini has joined #yocto
<angolini>
I've been using EXTRA_USERS_PARAMS use useradd -P and now I'm facing an error that -P is not a valid argument any more
<angolini>
but I could not spot what has been changed. So I think I'm missing some piece
<angolini>
I get this: useradd: unrecognized option '--clear-password'
jmiehe has quit [Ping timeout: 260 seconds]
vladest has quit [Quit: vladest]
kiran has joined #yocto
florian_kc has quit [Ping timeout: 264 seconds]
* paulg
wonders what gio-module-cache is and how to kill it with fire.
<paulg>
kinda sad when I can do git whatchanged on meta/classes/gio-module-cache.bbclass and still have NFI what it is/does.
* paulg
does what he always does - deletes it and sees what breaks
otavio has quit [Remote host closed the connection]
<jsbronder>
Is there any effort to backport changes from bitbake master to 1.46 for compatibility with newer python versions? And if not, would patches to that effect be be welcome?
<jsbronder>
I have bitbake-selftest passing on 3.10.
<vd>
angolini: I literally wrote about this 2 hours ago ^^ I don't understand neither, you must use -p for the password instead of -P now...
<angolini>
crazy! Thanks @vd
<angolini>
but when I user -p, I cannot log in :(
<vd>
angolini: Ha. That being said I just fix the compilation like this, but I didn't tested it...
<angolini>
crying
<angolini>
I understand the parsing for the arguments are one, and the adduser binary are another and they don't match. However, I don't get how the adduser bin got changed
<vd>
me neither
florian_kc has joined #yocto
dev1990 has quit [Quit: Konversation terminated!]
<drewfustini>
RP: sorry was away, thanks for the question. I have not created a bbappend yet but I was thinking that would be the correct way?
vladest has joined #yocto
dlan has joined #yocto
<abelloni>
drewfustini: I guess you'd need to add something like:
<moto-timo>
and then remove the existing --with-perl line from the EXTRA_OECONF
<abelloni>
I'm concernend about what is done in EXTRA_OEMAKE though
<moto-timo>
in a bbappend, you might be able to get away with EXTRA_OECONF:remove = "--with-perl=${STAGING_BINDIR_NATIVE}/perl-native/perl"
<smurray>
angolini vd: the -P option was being added by an OE-specific patch to shadow that was dropped when it was last upgraded. If you want to set a password you need to provide a hashed one with -p now AIUI
<smurray>
hrm, there was mention on the ml of that maybe needing to go in the migration guide for 3.4, but it seems to not have made it
<moto-timo>
abelloni: good point, the EXTRA_OEMAKE is going to still pull in perl
<moto-timo>
(or just fail gloriously)
dtometzki has quit [Read error: Connection reset by peer]
<angolini>
thanks @smurray
dtometzki has joined #yocto
camus has quit [Quit: camus]
<drewfustini>
moto-timo: thanks, I will try EXTRA_OECONF:remove. BTW, the ":" is the new syntax right?
<moto-timo>
angolini: this is what I use in my phosh image to set the "password" to 1234
<moto-timo>
where the user is "phosh", so modify as needed for your use case
<angolini>
great! THanks a lot @moto-timo! I will need this <3
<moto-timo>
I've tried other ways to create the password hash, but this is the one that worked for me (mkpasswd -m sha512crypt)
<moto-timo>
drewfustini: correct, that's the "new override" syntax, so replace with _ for older releases
<moto-timo>
drewfustini: you might also have to EXTRA_OEMAKE:remove = "'PERL_PATH=/usr/bin/env perl'"
<moto-timo>
no idea if it will blow up or not
<Tokamak>
anyone ever had an issue where sstate isn't picking up changes to PACKAGECONFIG for a package (specifically iptables recipie)? Trying to enable the nftables hybrid interface by doing: PACKAGECONFIG_append_iptables = " libnftnl libnfnetlink". Those PACKAGECONFIG are propagating as seen by bitbake -e iptables. But what's puzzling is bitbake is not triggering a rebuild of iptables for my image with changes in PACKAGECOFNIG for my image.
<Tokamak>
i haven't tried a manual clean - my next step. but wanted to ask how to debug/determine if sstate has a problem or not?
<vd>
smurray: this definitely need to go in the 3.4 migration guide, as the error isn't obvious at all, especially when it worked for a long time.
<vd>
moto-timo: thanks for the snippet. I guess EXTRA_USER and EXTRA_USER_PASSWORD_PLAIN would be a nice addition to the extrausers.bbclass to compute this automatically from the class itself.
<smurray>
vd: can you file a bug in the Yocto bugzilla for that?
<vd>
smurray: sure.
<smurray>
vd: okay, thanks
rpcme has quit [Quit: Client closed]
<moto-timo>
DISTRO_FEAURES:remove = "x11" doesn't work without the T LOL
<moto-timo>
TGIF
<moto-timo>
Corollary: don't look for a complicated solution until you have ruled out typos
<vd>
moo-imo did a ypo
<moto-timo>
indeed
<moto-timo>
also, put "bitbake-getvar" into your list of reasons to convince your boss or customers to move to a newer release...
<moto-timo>
I should go create a sh**post twitter account named moo-imo
<moto-timo>
too bad I'm not really a sh**poster
<moto-timo>
troll yes
florian_kc has quit [Ping timeout: 264 seconds]
atril has quit [Remote host closed the connection]
atril has joined #yocto
<sakoman>
moo-imo is a roll
rcw has joined #yocto
tre has quit [Remote host closed the connection]
lucaceresoli has joined #yocto
<moto-timo>
\o/
<moto-timo>
vd: sakoman: thank you for making Friday fun!