<AdrianTheFirst>
JaMa, thanks for the help yesterday. We decided to go the route of disabling the version-going-backwards globally for now, rather than taking the more complicated route of setting up a PR_service.
<AdrianTheFirst>
Our Yocto build on the build server start from scratch most of the time, so the version history is lost.
Guest98 has quit [Quit: Ping timeout (120 seconds)]
LocutusOfBorg has quit [Ping timeout: 240 seconds]
kpo has joined #yocto
LocutusOfBorg has joined #yocto
baeruhne[m] has quit [Remote host closed the connection]
Thorn has joined #yocto
ilunev has joined #yocto
Guest45 has joined #yocto
<Guest45>
is the git.yoctoproject.org server down? I'm getting fetcher errors and doing a: git clone git@git.yoctoproject.org:pseudo -b oe-core just freezes?
<mcfrisk>
Guest45: git access works for me, server responds on time
<Guest45>
mcfrisk: thanks for testing, strange, I'm not getting anything through the git protocol from two different servers, https works fine though...
<mcfrisk>
Guest45: add ";protocol=https" to the SRC_URI line. Looks like poky commit 1285f78ce861a92dbb99f476b49499574259e615 is not yet applied in dunfell branch.
<Guest45>
mcfrisk: ok, thanks for you help!
AdrianTheFirst has quit [Ping timeout: 245 seconds]
leon-anavi has joined #yocto
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
starblue has quit [Ping timeout: 265 seconds]
starblue has joined #yocto
Guest98 has quit [Quit: Client closed]
tlwoerner has joined #yocto
d-s-e has joined #yocto
Guest98 has joined #yocto
vladest has quit [Quit: vladest]
<Guest98>
i have a question about emmc write but it's related with yocto and emmc knowledge. may i ask? rookiee here O_O
vladest has joined #yocto
<rburton>
feel free to ask
vladest has quit [Client Quit]
<Guest98>
I'm trying to write the image created in imx8mq(coral dev board) board to emmc. By the way, using "meta-coral(zeus)". Encountering the following error:
<Guest98>
After investigating the problem, people said that these partitions should not be read only mode. I did the following step to get it out of read-only mode.
<Guest98>
When i reboot, the configurations are gone. How can i get an image installed on the sd card to be uploaded to emmc when i insert it into the device? emmc is working with mendel os but not working with yocto.
vladest has joined #yocto
Guest98 has quit [Quit: Client closed]
Guest98 has joined #yocto
seninha has joined #yocto
BobPungartnik has joined #yocto
<qschulz>
Guest98: this is something that needs to be done on each boot as this is an eMMC controller configuration AFAIR
<qschulz>
so do the echo before this copy is done
<qschulz>
highly recommend doing only one at a time
zpfvo has quit [Ping timeout: 264 seconds]
<qschulz>
and properly switch the eMMC boot partition to use for booting when flashing one
<qschulz>
so that if you have a power loss during a flash of boot partition 0, you boot from boot partition 1 next try
<qschulz>
at least less likely to be completely bricked
BobPungartnik has quit [Client Quit]
<Guest98>
qschulz It's hard to be a rookie :D. I will try to understand what you say and try to do something. Thank you!
<d-s-e>
Hey, I try to build a rust application with yocto, but it fails with "error running rustfmt: Os { code: 2, kind: NotFound, message: "No such file or directory" }". Any ideas?
xmn has quit [Ping timeout: 264 seconds]
kscherer has joined #yocto
<rburton>
d-s-e: looks like you need a dependency on whatever provides rustfmt? (note: i've never done rust)
<d-s-e>
maybe. rustfmt is available on the build host, but it seems not to be used. I also tried to add a a rustfmt recipe that fails with lots of errors (I'm not really into rust either).
<rburton>
tools on the build host are not visible unless they're in HOSTTOOLS
<rburton>
rustfmt appears to be part of the rust recipe, so dunno
camus has quit [Ping timeout: 240 seconds]
<d-s-e>
so I could simply append it to HOSTTOOLS?
<d-s-e>
part of the rust recipe?
vladest has quit [Ping timeout: 264 seconds]
Xagen has joined #yocto
<d-s-e>
WTH is canadian? I see several packages with that in the name!?
<jbo>
Hey guys, I am currently trying to get a custom board running. So far everything works. But I am struggling with the sound. I put an SSM2603 on the board. It's connected over I2C and to I2S0 of the SAMA5D. I am really lost on the DTS setup for this. Would anybody be willing to give me a hand?
<d-s-e>
rburton: thanks, very confusing. The rustfmt is part of the rust-tools-cross-canadian recipe, so does this mean, it will not be build to run natively on the build host?
Guest98 has quit [Quit: Client closed]
vladest has joined #yocto
<rburton>
you'd do better waiting for someone who understands how rust is built to get back to you. that recipe doesn't even exist in mickledore.
Guest98 has joined #yocto
<d-s-e>
thanks anyway, rust looks like a very strange half-baked thing to me so far. I'm on kirkstone.
sakoman has joined #yocto
<rburton>
yeah the rust recipes got rewritten post-kirkstone. there's a mixin layer with modern rust in if you want that instead.
<d-s-e>
ah ok, maybe that's why it's so hard :) I will have a look at that.
Minvera has joined #yocto
zpfvo has quit [Ping timeout: 240 seconds]
Peter[m]123 has joined #yocto
Guest98 has quit [Quit: Client closed]
rob_w has quit [Remote host closed the connection]
camus has joined #yocto
<Peter[m]123>
Hi all, I'm curious what the 2023 opinion is of the gold linker / "ld-is-gold" DISTRO_FEATURE. I'm currently evaluating it internally (at Arm) and seeing very promising results, particularly in terms of memory usage when linking large cpp binaries. Would you consider patches to exercise it on the autobuilder, or even enable it by default (for supported architectures)?
<JaMa>
Peter[m]123: I'm still using it for historical reasons in some of my builds, but lld seems to be much better alternative nowadays
d-s-e has quit [Ping timeout: 240 seconds]
<rburton>
lld would be if you've switch to clang too, right
<rburton>
someone should write a little lld recipe that builds just lld and test that
<rburton>
then again i've got a branch that adds and uses mold, switching linkers is never fun
<JaMa>
or mold :)
<fbre>
Hi, maybe a special question of the SoC I use, but is it possible to map parts of the L1 oder L2 cache as heap within Linux, just for having fast RAM?
<LetoThe2nd>
fbre: there are things like MLOCK, and cache locking when it comes to RT and nanoseconds - everything else is definitely premature overoptimization, and your problem is somewhere else.
<rburton>
reducing the amount of L1 cache for "performance" seems like the wrong thing to do :)
<LetoThe2nd>
rburton: there are rare circumstances where you can trade it for determinism. but yeah, generally such makes your system not faster, but slower - albeit more predictably slower :-)
<fray>
only devices I know of that use on-board cache as RAM, do it as a short-term quick-boot strategy, then they configure actual dram 'late' so they can stage device startup (which can take longer then ram config) earlier
<fray>
but that is really specialized stuff.. and as someone said, the wrong way to optimize..
<JaMa>
In my usecase I would prefer surprisingly fast over predictably slow :)
<rburton>
lol
<fbre>
the firmware I develop has an algorithm module which computes 10 times faster with such fast RAM. It works well on an DSP and its operationg system (SysBios) where the cache can be mapped as fast heap. Now we are porting to Linux and I come across this question.
<Peter[m]123>
JaMa/rburton: We already depend on meta-clang (for bindgen) in our project, so I'll give ld-is-lld a test drive. Thanks.
<fbre>
So I read your answers as "no, we don't know how to map the cache as another heap", right?
Xagen has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<rburton>
fbre: you're asking an incredibly specific question on the general yocto project support channel
<fbre>
yes, it's true, I should ask in the SoC forum, but I just wanted to know if Linux allows such mapping of hardware components to own heaps with own address ranges
<fbre>
...which is a general question independent of the hardware
<fbre>
as I said, the TI operating system (Sys BIOS) has that feature of mapping the L1 and L2 cache hardware to special heap address ranges. This makes things so fast on a DSP. World is turning and so we have to get that working on Linux too
d-s-e has joined #yocto
ahrimen has quit [Ping timeout: 240 seconds]
<fbre>
... on the DSP you separate the actual algorithm firmware module from the maintenance system
ahrimen has joined #yocto
<fbre>
...for which controlling fast heap makes much sense that way
d-s-e has quit [Client Quit]
<fbre>
Thanx anyway, I'll ask in the SoC forum
silbe has quit [Ping timeout: 250 seconds]
goliath has quit [Quit: SIGSEGV]
ahrimen has quit [Ping timeout: 240 seconds]
ahrimen has joined #yocto
Xagen has joined #yocto
thomasd13 has quit [Ping timeout: 268 seconds]
bps2 has quit [Ping timeout: 240 seconds]
<khem>
rburton: lld would need llvm backend to be available for build, technically we can repurpose the llvm recipe in oe-core to provide that and then have a recipe for lld, however if you use meta-clang then it lld comes by default as part of tool suite and all you need is ls-is-lld in DISTRO_FEATURES 🙂
<khem>
err ld-is-lld
<khem>
I have been using it in yoe distro for a long time now
fbre has quit [Quit: fbre]
silbe has joined #yocto
NishanthMenon[m] has joined #yocto
geoffhp has quit [Remote host closed the connection]
<khem>
mold is inder AGPL-3 hmmm interesting
<jbo>
Hey guys, I'm running into an issue where I do "bitbake -c menuconfig virtual/kernel". change settings, then same them. then do a build and the settings are reverted next time. I did not have this problem before my meta-custom layer added a patch for a DTS. is there something obvious I am missing?
<smurray>
khem: heh, I saw your discussion about lld and was wondering about the current state of mold myself
zpfvo has quit [Ping timeout: 265 seconds]
nemik has quit [Ping timeout: 268 seconds]
nemik has joined #yocto
florian has quit [Quit: Ex-Chat]
nemik has quit [Ping timeout: 240 seconds]
nemik has joined #yocto
<rburton>
mold is 'interesting'
<rburton>
in my lame testing, it wasn't hugely faster
rfuentess has quit [Remote host closed the connection]
xmn_ has quit [Read error: Connection reset by peer]
florian__ has quit [Ping timeout: 240 seconds]
Thorn has quit [Ping timeout: 240 seconds]
xmn has joined #yocto
frieder has quit [Remote host closed the connection]
nemik has quit [Ping timeout: 268 seconds]
nemik has joined #yocto
<jbo>
Guys, is there any magic to having a persistent custom kernel config in my own layer? I have meta-custom/recipes-kernel/linux/linux-mchp_6.1.22.bbappend. In there, I have an SRC_URI:append = "file://defconfig" and in the same directory I have "defconfig". However, changes do not seem to apply.
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
<dario>
do you have a line like FILESEXTRAPATHS:prepend := "${THISDIR}:" or similar?
<dario>
iirc if anything, only the files/ subdir is automatically added to the files search path
<jbo>
yes, I have that
<jbo>
I am still investigating
<jbo>
the documentation mentions ${THISDIR}/${PN}: instead. I am not sure yet what ${PN} actually is.
Thorn has joined #yocto
<khem>
rburton: mold has a commercial offering called sold 🙂 so I guess thats the play
<dario>
jbo: the package name, i guess it's a matter of taste/repo organization which one you choose
<jbo>
aye, I think I got it working now. I removed some stuff, compiled the image and stuff seems to be gone. also doing a bitbake -c menuconfig virtual/kernel now shows those options disabled
<dario>
i guess in general bitbake -e [recipename] | less lets you search whether it includes all the files you expect it to
<jbo>
I think I never struggled with something as much as getting this SSM2603 codec to show up as a device under ALSA. I still don't have it working. I don't understand the DTS :D
kscherer has quit [Quit: Konversation terminated!]
ahrimen has quit [Ping timeout: 264 seconds]
ahrimen has joined #yocto
bps2 has joined #yocto
dmoseley_ has joined #yocto
dmoseley has quit [Ping timeout: 240 seconds]
<kergoth>
Ugh, getting a no provider message with no info on what depended on it, so no idea what's pulling it into the build with the wrong provider dep