2025-12-01 01:17:55 <huajingyun> Good morning!
2025-12-01 01:18:07 <huajingyun> Hi ncopa, is the upgrade to kernrl 6.18 now in preparation for the release in v3.23?
2025-12-01 07:04:53 <mps> huajingyun: i think it is prepared https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/93886
2025-12-01 07:11:12 <huajingyun> Ok,thanks mps 
2025-12-01 07:13:12 <mps> right now I'm looking at new options and drivers for loongarch64 to build it and test
2025-12-01 08:23:15 <mps> Linux la 6.18.0-0-sp #1-Alpine SMP PREEMPT_DYNAMIC Mon, 01 Dec 2025 07:16:37 +0000 loongarch64 GNU/Linux
2025-12-01 08:24:38 <mps> boots and run, didn't noticed any (serious) problem
2025-12-01 09:13:08 <huajingyun> Thanks,I'm guessing you built  6.18 instead of 6.18-rc7, I checked https://www.kernel.org/ earlier today and it was still rc7, but it's been updated now.
2025-12-01 09:28:34 <mps> yes, built 6.18 stable and I think it will be LTS kernel
2025-12-01 09:34:52 <huajingyun> That's great!
2025-12-01 09:35:25 <huajingyun> ncopa:I remember initially rename vmlinuz /config/System.map to fix an error I encountered while creating an iso, but sorry I can't recall the specific error
2025-12-01 09:35:29 <huajingyun> I haven't tried deleting it now to see if it works, but hopefully it will
2025-12-01 09:38:49 <mps> I delete System.map in pkgs I build, not needed for 'normal' work. If it is needed maybe add subpackage or add System.map to -dev subpackage
2025-12-02 05:56:10 <huajingyun> Oops, I just upgraded my environment using "apk upgrade -a", and the reboot failed
2025-12-02 05:56:19 <huajingyun> If anyone wants to upgrade using 'apk upgrade -a', please be careful
2025-12-02 06:01:23 <huajingyun> Has anyone upgraded? any arches?  I mean upgrading the kernel to 6.18 via 'apk upgrade'
2025-12-02 06:01:36 <huajingyun> I think the biggest change in the last two days is linux-lts 6.18.0
2025-12-02 06:45:29 <ncopa> Ok. Sorry about that
2025-12-02 06:46:27 <mps> huajingyun: I will try in a hour or two to reboot with linux-lts 6.18.0 from alpine repository
2025-12-02 06:46:42 <ncopa> Almost sure it is 6.18 that broke. Does Linux-stable  work? The 6.17 
2025-12-02 06:47:25 <mps> my locally built 6.18.0 kernel works fine
2025-12-02 06:47:58 <ncopa> Probably the rename then. Can you what is different in the alpine kernel?
2025-12-02 06:48:20 <ncopa> Maybe list content of packages and run diff on that
2025-12-02 06:48:59 <mps> yes, I can but have first to do some morning task
2025-12-02 06:50:21 <ncopa> Same here :)
2025-12-02 06:50:33 <mps> but main diff is I have built ext4 and nvme drivers in kernel and not as a module
2025-12-02 06:51:09 <mps> ncopa: coffee? :)
2025-12-02 06:52:29 <DWwanghao> currently, it appears that the ext4 module is missing from modules.dep, which may be caused by filesystem configuration issues
2025-12-02 06:54:21 <huajingyun> In boot/config-6.18.0-0-lts, found the following # CONFIG_EXT4_FS is not set
2025-12-02 06:54:52 <huajingyun> Perhaps it is indeed related to this
2025-12-02 06:56:26 <mps> yes, this is 
2025-12-02 06:57:12 <mps> can't mount root FS
2025-12-02 06:57:26 <mps> can't find*
2025-12-02 06:58:57 <mps> I thought that limini bootloader could be problematic in my case
2025-12-02 07:10:31 <ncopa> ugh....
2025-12-02 07:10:44 <ncopa> it means i messed up completely when updating the configs
2025-12-02 07:11:04 <ncopa> i missed the ext4 on x86_64 as well on my first try on 6.18-rc7
2025-12-02 07:11:41 <ncopa> thanks! I will try regen the config
2025-12-02 07:13:14 <ncopa> looks like it is also missing from armv7 aarch64, s390x, and ppc64le
2025-12-02 07:14:41 <DWwanghao> ncopa: 6.17.10-0-stable works fine
2025-12-02 07:14:51 <ncopa> ok. thanks
2025-12-02 07:15:21 <huajingyun> I just created a 3.23 iso, and here's an error log for reference: https://tpaste.us/yB6k
2025-12-02 07:15:55 <ncopa> i will push an emergency fix with ext4 enabled (as module). There was also missing CONFIG_XFS_ONLINE_REPAIR
2025-12-02 07:16:57 <huajingyun> ncopa: thanks!
2025-12-02 07:17:18 <DWwanghao> I saw Saijin_Naib's message. Thank you all:)
2025-12-02 07:21:01 <ncopa> sorry about that...
2025-12-02 07:23:19 <ncopa> i wonder if someone could help me with comparing the configs with diffconfig
2025-12-02 07:23:34 <ncopa> diffconfig is in linux source tree. ./script/diffconfig
2025-12-02 07:23:55 <ncopa> compare linux-stable 6.17.10 config with the linux-lts 6.18
2025-12-02 07:24:20 <ncopa> and compare linux-lts 6.18 (from alpine 3.22) with the 6.18.0 config
2025-12-02 07:24:43 <ncopa> there should be file in /boot/config-* in package
2025-12-02 07:27:10 <huajingyun> compare all arches? or just loongarch?
2025-12-02 07:29:09 <ncopa> well, i was thinking loongarch specifically
2025-12-02 07:29:20 <ncopa> but if you have time to do it on all arches....
2025-12-02 07:29:46 <ncopa> look if there is anything else important that got lost
2025-12-02 07:33:01 <huajingyun> Of course!
2025-12-02 07:42:36 <huajingyun> ncopa:Linux-LTS 6.18 (from Alpine 3.22) -> perhaps you meant 3.23? I see that the linux for 3.22 is 6.12.60-r1
2025-12-02 07:48:57 <ncopa> yeah 3.23 if you still have it
2025-12-02 07:49:00 <DWwanghao> ncopa: compare linux-stable 6.17.10 config with the linux-lts 6.18:https://tpaste.us/5QqE
2025-12-02 07:57:44 <ncopa> oh, you need the full config. from apk package in /boot/config-*
2025-12-02 07:58:03 <ncopa> the ext4 is not listed there, but it was missing
2025-12-02 08:01:41 <mps> CONFIG_MFD_LOONGSON_SE=m
2025-12-02 08:02:02 <mps> CONFIG_MFD_LS2K_BMC_CORE=y
2025-12-02 08:02:39 <mps> +CONFIG_SND_SOC_LOONGSON_CARD=m
2025-12-02 08:02:39 <mps> +CONFIG_SND_SOC_LOONGSON_I2S_PLATFORM=m
2025-12-02 08:02:39 <mps> +CONFIG_SND_SOC_LOONGSON_I2S_PCI=m
2025-12-02 08:03:23 <mps> CONFIG_EDAC_LOONGSON=m
2025-12-02 08:03:42 <mps> CONFIG_LOONGSON2_APB_DMA=y
2025-12-02 08:05:06 <mps> not sure for this but I enabled also CONFIG_LOONGSON_LAPTOP=y
2025-12-02 08:05:30 <mps> CONFIG_PWM_LOONGSON=m
2025-12-02 08:05:48 <mps> CONFIG_CRYPTO_DEV_LOONGSON_RNG=m
2025-12-02 08:06:19 <mps> and sorry for spam
2025-12-02 08:07:27 <ncopa> emergency fix pushed. -r1
2025-12-02 08:07:33 <ncopa> oncly change was ext4
2025-12-02 08:09:03 <mps> CONFIG_ACPI_VIDEO=y
2025-12-02 08:09:34 <DWwanghao> ncopa: my local machine is still on 6.12.57-0-lts, and I have compared the differences between 6.12.57-0-lts and 6.18.0-0-lts.https://tpaste.us/QKgr
2025-12-02 08:09:44 <mps> CONFIG_LOONGSON3_CPUFREQ=m
2025-12-02 08:09:57 <huajingyun> diffconfig linux-lts-6.18.0-r0/boot/config-6.18.0-0-lts linux-stable-6.17.10-r0/boot/config-stable https://tpaste.us/oxWg
2025-12-02 08:11:57 <mps> afaik most loongson machines have amount of RAM so maybe CONFIG_PERSISTENT_HUGE_ZERO_FOLIO=y
2025-12-02 08:19:23 <mps> in all kernels: -CONFIG_LEGACY_PTYS=y and -CONFIG_LEGACY_TIOCSTI=y, i think this is not needed nowadays
2025-12-02 08:24:33 <mps> and a loot more but enough for now ;)
2025-12-02 08:25:46 <mps> imo currently alpine kernels should be 'better' configured
2025-12-02 08:26:59 <mps> because this I always build it locally with much changes
2025-12-02 08:28:53 <mps> one irritable thing with loongarch64 linux it is forced to have page_size of 16K only
2025-12-02 08:28:57 <huajingyun> thanks:-D
2025-12-02 08:42:56 <ncopa> Thanks
2025-12-02 08:43:15 <ncopa> Do we need debug info btf?
2025-12-02 08:43:31 <ncopa> its needed for ebpf progs 
2025-12-02 08:48:05 <mps> i think yes
2025-12-02 08:53:52 <ncopa> CONFIG_ACPI_VIDEO=y why?
2025-12-02 09:00:59 <DWwanghao> after upgrading to linux-lts-6.18.0-r1, the system rebooted successfully
2025-12-02 09:01:48 <ncopa> phew. thanks!
2025-12-02 09:03:06 <mps> CONFIG_ACPI_VIDEO=y is for amd/radeon card
2025-12-02 09:03:26 <ncopa> should only be needed if you dont have full GPU driver, which I think we have
2025-12-02 09:04:09 <mps> hm, you are right
2025-12-02 09:05:03 <mps> I forgot that it now works fine also on 3A6000
2025-12-02 09:08:06 <ncopa> $ grep _LOONG .config | tpaste
2025-12-02 09:08:06 <ncopa> https://tpaste.us/j1B1
2025-12-02 09:08:40 <ncopa> I suppose PWM would be good
2025-12-02 09:08:50 <ncopa> the other I dont know what they are or if they are needed
2025-12-02 09:08:55 <ncopa> eg CONFIG_MFD_LOONGSON_SE
2025-12-02 09:09:00 <ncopa> CONFIG_LOONGSON2_APB_DMA
2025-12-02 09:09:32 <ncopa> I dont see CONFIG_CRYPTO_DEV_LOONGSON_RNG, probably due to some other RNG related config missing
2025-12-02 09:11:31 <mps> MFD MultiFunction Device
2025-12-02 09:13:14 <mps> i don't have schematic of loongson but i think it is added to kernel because such device is in machines
2025-12-02 09:14:19 <ncopa> looks liek that is needed for crypto device
2025-12-02 09:19:49 <mps> well, serious analyzis what to enable/disable will need some time and iiuc you want to release stable soon
2025-12-02 09:21:32 <mps> and now, time for breakfast
2025-12-02 09:22:39 <ncopa> yes. it does require some time. I would appreciate help with whats good for loongsoon, since I also need to look over the other 8 architectures, and also virt flavors for other architectures
2025-12-02 09:22:52 <ncopa> you people here are the experts in loongarch
2025-12-02 09:25:30 <ncopa> does longarch support SMT? (hyper-threading?)
2025-12-02 09:36:51 <huajingyun> ncopa:as far as I know, not all models support SMT
2025-12-02 09:37:24 <huajingyun> The 3a6000 and 3c6000 （LA664）support , but other, such as the 5000 series, do not
2025-12-02 09:37:34 <ncopa> ok, so there are some that does support it
2025-12-02 09:37:48 <ncopa> make sense to enable it in kernel then
2025-12-02 09:42:33 <huajingyun> Yes
2025-12-02 09:50:23 <huajingyun> CONFIG_MFD_LOONGSON_SE is the SE module driver for Loongson chips,CONFIG_LOONGSON2_APB_DMA is the DMA driver for the Loongson 2 platform chip,CONFIG_CRYPTO_DEV_LOONGSON_RNG should be the Loongson encryption/decryption module driver
2025-12-02 09:50:24 <huajingyun> If there are requirements for accelerating encryption/decryption calculations or security modules, then those should be included.
2025-12-02 09:56:47 <ncopa> IU have this for now https://tpaste.us/BMyW
2025-12-02 09:56:55 <ncopa> let me know if there is anything else
2025-12-02 10:20:30 <mps> 
ACTION wonder why is kernel 'PE32+ executable for EFI' and not simply linux

2025-12-02 10:22:16 <mps> betrayal to microsoft
2025-12-02 10:31:08 <huajingyun> ncopa:thanks
2025-12-02 10:40:54 <huajingyun> ncopa,not sure if you still need this, this compares the config-* of linux-lts 6.18.0 and linux-stable 6.17.10 on several other arches
2025-12-02 10:41:56 <huajingyun> aarch64: https://tpaste.us/xDqQ ,  x86: https://tpaste.us/yB68 , ppc64le: https://tpaste.us/YqPe , x86_64: https://tpaste.us/5Qq7 , s390x: https://tpaste.us/vo8W , armv7: https://tpaste.us/QKg8
2025-12-02 10:42:17 <huajingyun> It appears that riscv64 has not yet completed syncing with Linux-lts 6.18.0
2025-12-02 10:42:35 <huajingyun> so hope this help:)
2025-12-02 10:51:19 <ncopa> thanks!
2025-12-02 11:07:46 <huajingyun> you're welcome
2025-12-02 11:47:49 <ncopa> i pushed new kernel with lots of improvements for loongarch64. thanks mps for good feedback!
2025-12-02 11:48:14 <ncopa> also thanks for the above diffs. they revelaed that erofs was missing
2025-12-02 11:49:27 <mps> ncopa: you are welcome. happy if i can be of any help
2025-12-02 11:50:23 <mps> though i'm far from expert on this
2025-12-02 11:53:12 <mps> now i'm fighting with smart monitors and wonder how the monitor could be smart :)
2025-12-02 11:53:51 <mps> or any machinery at the end
2025-12-02 11:54:46 <ncopa> ha. indeed. its a miracle anythign works at all
2025-12-02 11:55:07 <mps> 'smart phone' for me sounds very stupid term
2025-12-02 11:55:35 <mps> yes, you right
2025-12-02 12:01:28 <mps> btw, anyone know what is status on loongson gpu? last year i tried and it not worked. maybe it is improved in meantime
2025-12-02 12:08:52 <DWwanghao> to be precise, it should be GPGPU
2025-12-02 12:11:10 <DWwanghao> the 9A1000 has already taped out (haha, this isn't confidential, is it?), and its performance slightly outperforms the RX550
2025-12-02 12:11:53 <mps> aha, nice news
2025-12-02 12:33:28 <adrian> [    2.791913] loongson3_cpufreq loongson3_cpufreq: probe with driver loongson3_cpufreq failed with error -1
2025-12-02 12:33:29 <adrian> hmmm
2025-12-02 12:58:19 <mps> and this is on last series of stable kernels. 
2025-12-03 04:58:00 <adrian> regarding the cpufreq probe failures: https://lore.kernel.org/loongarch/20250623123321.5622-1-ziyao@disroot.org/
2025-12-03 04:58:22 <adrian> apparently the interface in the kernel is not implemented by anything released yet
2025-12-03 08:28:42 <mps> ncopa: !94087
2025-12-03 08:29:53 <mps> also !94086
2025-12-03 08:30:39 <mps> would be nice to have these in 3.23-stable release
2025-12-03 08:43:28 <ncopa> headers makes no sense
2025-12-03 08:45:59 <huajingyun> ncopa: just freed up a machine and tried to install https://dl-cdn.alpinelinux.org/alpine/v3.23/releases/loongarch64/alpine-standard-3.23.0_rc2-loongarch64.iso 
2025-12-03 08:46:14 <huajingyun> It ended up complaining about a "Segmentation fault"
2025-12-03 08:46:18 <huajingyun> https://tpaste.us/BMy4
2025-12-03 08:47:04 <ncopa> Can you find out what segfaulted?
2025-12-03 08:47:20 <ncopa> maybe dmesg can tell
2025-12-03 08:48:44 <ncopa> alternatively Run ulimit -c unlimited before so you get a core dump
2025-12-03 08:49:04 <ncopa> and use gdb to create a backtrace
2025-12-03 08:56:08 <huajingyun> Ok, I'll give it a try
2025-12-03 09:41:12 <huajingyun> forgot to send the dmesg log: https://tpaste.us/bQYJ
2025-12-03 09:41:31 <huajingyun> Core was generated by `/sbin/apk add --root /mnt --initdb --quiet --progress --update-cache --clean-protected --overlay-from-stdin --repository /media/sdb/apks --repository http://mirrors.ustc.edu.cn/alpine/v3.23/main alpine-base busybox-mdev-openrc dosfstools e2fsprogs grub-efi openssh openssl linux-lts alpine-base grub-efi linux-firmware-amdgpu linux-firmware-bnx2 linux-firmware-rtl_nic'.
2025-12-03 09:41:43 <huajingyun> Program terminated with signal SIGSEGV, Segmentation fault.
2025-12-03 09:41:44 <huajingyun> #0  0x00007ffff21572d0 in apk_pkg_uninstall () from /usr/lib/libapk.so.3.0.0
2025-12-03 10:01:47 <mps> something is bad with gitlab.a.o - i'm receiving multiple mails for single events
2025-12-03 10:06:08 <achill> known issue
2025-12-03 10:35:07 <ncopa> huajingyun: can you apk add apk-tools-dbg
2025-12-03 10:35:22 <ncopa> apk add apk-tools-dbg gdb
2025-12-03 10:35:46 <ncopa> gdb --core /path/to/core /sbin/apk
2025-12-03 10:35:56 <ncopa> and the do 'bt' to generate a backtrace
2025-12-03 10:36:05 <ncopa> create an issue for apk-tools
2025-12-03 10:36:43 <ncopa> https://gitlab.alpinelinux.org/alpine/apk-tools/-/issues
2025-12-03 10:45:37 <huajingyun> No problem, I'll create it as soon as possible
2025-12-03 11:05:02 <huajingyun> ncopa: here it is https://gitlab.alpinelinux.org/alpine/apk-tools/-/issues/11164
2025-12-03 11:06:19 <ncopa> thanks!
2025-12-03 11:06:25 <ncopa> its definitively a bug in apk-tools
2025-12-03 11:06:32 <ncopa> do we have valgrind for loongarch64?
2025-12-03 11:08:34 <DWwanghao> yes
2025-12-03 11:08:47 <huajingyun> alpine:~$ apk list valgrind-loongarch
2025-12-03 11:08:47 <huajingyun> valgrind-loongarch-3.25.0_git20250921-r0 loongarch64 {valgrind-loongarch} (GPL-2.0-or-later)
2025-12-03 11:08:48 <DWwanghao> community/valgrind-loongarch
2025-12-03 11:11:06 <ncopa> can you try run the apk command that segfaults in valgrind?
2025-12-03 11:11:18 <ncopa> valgrind apk add .....
2025-12-03 11:11:25 <ncopa> paste the output to the issue
2025-12-03 11:15:31 <huajingyun> add any packages?
2025-12-03 11:15:52 <huajingyun> eg: valgrind apk add vim?
2025-12-03 11:27:30 <mps> `apk add valgrind`
2025-12-03 11:28:20 <huajingyun> One fixed: https://gitlab.alpinelinux.org/alpine/apk-tools/-/commit/8aafa7ce1c8c774622eae7a5e1c776ee23076e49
2025-12-03 11:30:10 <huajingyun> ncopa:It seems that updating the issues is no longer necessary,@fabled has fixed it
2025-12-03 11:35:55 <ncopa> that was very quick fixed. thank you for a good bug report!
2025-12-03 16:42:07 <mps> 6.18 is officially lts https://www.kernel.org/category/releases.html
2025-12-09 21:33:55 <Ariadne> hi!  i am working on some CI for alpine core toolchain components i maintain (e.g. libucontext).  where is "official" alpine OCI image for loongarch64?
2025-12-09 21:35:41 <Ariadne> nevermind, found it :)
2025-12-10 01:11:32 <huajingyun> Ok, we can get from registry.a.o, ikke has already pushed them,eg: registry.alpinelinux.org/img/alpine:3.23 or other tags:)
2025-12-11 02:43:43 <Ariadne> hi loongson folks, is it intentional that glibc ucontext functions do not preserve FPU state?  i was hoping to enable hard-float contexts in libucontext
2025-12-11 03:12:37 <huajingyun> Hi Ariadne,i'm not sure,  need to ask for you
2025-12-11 03:15:18 <Ariadne> https://areweloongyet.com/en/docs/world-compat-details/ fascinating
2025-12-11 03:43:37 <Ariadne> according to https://loongson.github.io/LoongArch-Documentation/LoongArch-ELF-ABI-EN.html $fs0-$fs7 are supposed to be preserved, but glibc does not do it
2025-12-11 08:37:16 <xen0n> Ariadne: I'll also forward the question to the Loongson glibc maintainers for you
2025-12-11 09:32:31 <huajingyun> I took a look at glibc  loongarch ucontext.h, __extcontext field contains extension records, and the pointer points to the extension context, including some like fpu/lsx/lasx, etc. 
2025-12-11 09:32:46 <huajingyun> however, it is mutable, which means that some cpus may or may not support it, and it is not actually retained, it is just a 0-length array
2025-12-11 09:33:05 <huajingyun> it is inherited from Linux,https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/loongarch/include/uapi/asm/sigcontext.h
2025-12-11 09:33:11 <huajingyun> https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/loongarch/sys/ucontext.h
2025-12-11 09:36:23 <huajingyun> Ariadne: you may need to save them in other ways
2025-12-27 17:24:36 <mio> cross-posting a request from #alpine-devel for help with containerd build failing on loongarch64  https://gitlab.alpinelinux.org/alpine/aports/-/jobs/2152095
2025-12-27 17:25:11 <mio> from !94947
2025-12-29 01:17:58 <DWwanghao> mio: Thanks for forwarding, let me take a look
2025-12-29 01:28:59 <mio> DWwanghao: thanks for checking
2025-12-29 02:23:25 <DWwanghao> The reason is that the ebpf dependency used by containerd was previously specified as version 0.11.0 for the loongArch, which caused cgroup to be downgraded to version 3.0.3 in the vendor
2025-12-29 02:23:43 <DWwanghao> I have already verified this locally, and I will submit an mr shortly
2025-12-29 02:37:52 <DWwanghao> mio: !95177
2025-12-29 02:38:03 <DWwanghao> It looks good now. Thanks again for your feedback