2025-07-06 18:39:30 <js> So I'm guessing it is supposed to be booted via some UEFI shell incantation?
2025-07-06 18:39:34 <js> aha, apparently typing "fs0:\efi\boot\bootloongarch64.efi" in UEFI shell does the trick. Maybe that should be added to the wiki?
2025-07-06 18:39:36 <js> cd
2025-07-06 18:39:36 <js> ok, I got it booted, but the NIC driver seems fucked? I can see the ARP request coming from the loongson, I see the reply being sent to the loongson, but the loongson keeps sending ARP requests, so I guess incoming packets are broken?
2025-07-06 18:39:39 <js> unfortunately the install ISO doesn't have tcpduimp, so I cannot see what it sees
2025-07-06 18:39:39 <js> and the other way around, the loongson never responds to an ARP request
2025-07-06 18:39:40 <js> also only 2 of the NICs is detected at all, and adding a USB WiFi dongle isn't detected either - I guess the install ISO strips almost all NIC drivers?
2025-07-06 18:39:42 <js> *only one of the 2 NICs is detected
2025-07-06 18:39:43 <js> hm, nope, rt2800usb (which claims the device on x86_64) can be modprobed, but it won't take the device. So it's not a missing module but something being broken.
2025-07-06 18:39:46 <js> hm, firmware file is also in the ISO, but no attempts to load it in dmesg
2025-07-06 18:39:47 <js> ok, so with 3 NICs not working, no way to install Alpine then ☹️
2025-07-07 20:50:28 <Ariadne> js: what board
2025-07-07 20:50:50 <js> Oh wow my messages got through despite the bridge being out of disk space? Nice.
2025-07-07 20:50:59 <js> It’s the "Corerboard"
2025-07-07 20:51:38 <Ariadne> is it the reference board, the engineering silicon evaluation board, or one of the retail boards?
2025-07-07 20:52:57 <js> Corer board XC-LS3A6M
2025-07-07 20:53:02 <js> Is how it was listed on Aliexpress
2025-07-07 20:53:57 <Ariadne> is it edk2 firmware or?
2025-07-07 20:54:08 <Ariadne> it should be edk2 if xc-ls3a6m
2025-07-07 20:54:09 <js> firmware it came with
2025-07-07 20:54:19 <js> wait, I have a dump somewhere
2025-07-07 20:54:28 <Ariadne> i'm just trying to identify the actual board
2025-07-07 20:54:45 <Ariadne> also, what ISO?
2025-07-07 20:54:51 <js> Well, strings on the firmware dump says: "UEFI-EDK2 loongson Initializing..."
2025-07-07 20:55:03 <js> alpine-standard-3.22.0-loongarch64.iso
2025-07-07 20:56:02 <Ariadne> xc-ls3a6m should work fine with that ISO.  one NIC is realtek, the other NIC is provided by the southbridge
2025-07-07 20:56:26 <Ariadne> both NICs hate energy-efficient ethernet though
2025-07-07 20:56:31 <js> the weird part is that the USB WiFI also didn't work 😕
2025-07-07 20:56:45 <js> one of the NICs isn't found at all, I only have eth0
2025-07-07 20:57:05 <Ariadne> oh no
2025-07-07 20:57:21 <js> and if I plug in a USB WiFI, it enumerates the device in dmesg, but attaches no driver
2025-07-07 20:57:28 <js> so something is quite wrong
2025-07-07 20:57:52 <Ariadne> if you go into setup, is it a GUI-based setup with some hanzi in the upper left corner?
2025-07-07 20:57:58 <Ariadne> or is it the stock EDK2 setup
2025-07-07 20:58:15 <js> by setup you mean UEFI setup? It looks like a regular EDK2
2025-07-07 20:58:18 <js> even set it to English
2025-07-07 20:58:36 <js> 923800a0fcb2cbb90315afaf3f99aba8d139d3018a4594866f148254fcffc7b6
2025-07-07 20:58:39 <js> is the dumped UEFI
2025-07-07 20:58:43 <js> I believe you wrote on Mastodon that you had the same
2025-07-07 20:58:49 <Ariadne> ok
2025-07-07 20:59:11 <Ariadne> though my board with that firmware only has realtek NIC.
2025-07-07 20:59:17 <Ariadne> how delightfully odd!
2025-07-07 20:59:30 <js> wait. did they … fuck up and flash it with the wrong FW?
2025-07-07 20:59:54 <js> if you have dumped the same FW from the reference board, that is very ... odd
2025-07-07 21:00:11 <Ariadne> could be a different revision of the reference board
2025-07-07 21:00:23 <Ariadne> i have a few of htem but never have seen this issue
2025-07-07 21:00:31 <js> nah, they should definitely not have the same FW, as the "corerboard" was the one that had overclocking
2025-07-07 21:01:39 <js> as in, overclocking in UEFI
2025-07-07 21:01:47 <js> and I think I even saw the option there, I can check again tomorrow
2025-07-07 21:02:08 <Ariadne> this is weird, maybe it is a pre-production variant of the asus retail board
2025-07-07 21:02:34 <js> yeah, it's the "ASUS" board - as in, the one made by ASUS supposedly but not having written ASUS anywhere on it
2025-07-07 21:02:34 <Ariadne> the one i have with overclocking has a GUI setup
2025-07-07 21:03:05 <js> /say /home/luka_lian/xc-fusion/Build/Ls3a60007a/RELEASE_GCC5/LOONGARCH64/AsusPkg/LsRefCodeOvrdPkg/SampleCode/Desktop/Ls3aPlatformTable/Pei/PeiLs3aPlatformTableInit/DEBUG/PeiLs3aPlatformTableInit.dll
2025-07-07 21:03:12 <js> so yeah, definitely asus
2025-07-07 21:03:16 <js> that string is in the firmware
2025-07-07 21:03:30 <js> ok, maybe what you call GUI is what I don't 😄 It's a text UI, but not just a console
2025-07-07 21:03:39 <js> but it's none of those fancy fully graphical UEFIs with mouse support and all that
2025-07-07 21:06:43 <Ariadne> on the retail board, it's literally a graphical UEFI with mouse support, and two hanzi (the logo of the firmware company?) in the upper left corner.  trebuchet MS is the font when running in english mode.  chinese/japanese/english are the only languages in the firmware too.
2025-07-07 22:00:25 <js> Ah, no, none of that. Should I upgrade firmware? Do they even offer downloads?
2025-07-07 23:07:01 <Ariadne> not to my knowledge
2025-07-10 18:56:46 <ncopa> anyone available to help fix build of qt6-qtwebengine? https://build.alpinelinux.org/buildlogs/build-edge-loongarch64/community/qt6-qtwebengine/qt6-qtwebengine-6.9.1-r1.log
2025-07-11 01:38:56 <DWwanghao> I can try to fix the qt6-qtwebengine build. Let me take a look
2025-07-11 07:25:53 <ncopa> Thanks!
2025-07-11 09:06:01 <DWwanghao> You are welcome:)
2025-07-11 09:06:49 <DWwanghao> Found the issue - variable-length arrays in musl headers causing trouble
2025-07-11 09:07:31 <DWwanghao> and I`m attempting to fix it with some patches from GitHub
2025-07-11 09:09:06 <DWwanghao> Going AFK now,and  will provide updates next Monday when back
2025-07-14 19:38:21 <ncopa> cc1plus: fatal error: /home/buildozer/aports/community/pyside6/src/pyside-setup-everywhere-src-6.9.1/build/sources/pyside6/PySide6/QtWebEngineCore/PySide6/QtWebEngineCore/qwebengineprofilebuilder_wrapper.cpp: No such file or directory
2025-07-14 19:38:33 <ncopa> anyone has time to also have a look at pyside6?
2025-07-14 19:39:05 <ncopa> which also fails to build on loongarch64 now
2025-07-14 19:57:42 <achill> fwiw the qt6-qtwebengine build is just disabled and needs unfortunately a proper fix
2025-07-15 09:47:27 <achill> oh i think the pyside6 error is because it pulled the old qtwebengine
2025-07-15 09:47:43 <achill> this is all annoying
2025-07-15 09:48:06 <achill> but if qtwebengine is fixed, pyside6 should be too, but until then im gonna disable it aswell
2025-07-15 11:34:32 <DWwanghao> Agree to disable pyside6. Fixing qt6-qtwebengine is proving more difficult than I imagined - maybe I'm just a noob at this:'(
2025-07-15 11:36:23 <ncopa> qtwebengine is not the easiest source code
2025-07-15 11:36:50 <achill> yeah maybe later today i got some time to look at
2025-07-15 11:37:16 <achill> but chromium is very annoying to deal with
2025-07-15 11:39:54 <DWwanghao> Thanks for the help! Yeah, it's so annoying—some code runs fine on glibc but throws warnings or crashes on musl
2025-07-25 19:22:29 <Ariadne> hi.  i am working on upgrading alpine to gcc 15, but on alpine gcc 15 the following program segfaults:
2025-07-25 19:22:47 <Ariadne> __thread int a; int b; int main() { return a = b; }
2025-07-25 19:22:59 <Ariadne> this only happens on my loongarch machine
2025-07-25 20:03:23 <Ariadne> it seems only to happen with TLS, using normal constructor attributes are fine
2025-07-26 06:56:30 <Ariadne> seems this only happens with static PIE
2025-07-26 07:33:25 <Ariadne> i figured it out, it was my mistake :)
2025-07-27 14:38:11 <cely> Ariadne: If you're going to pkgrel-bump gcc again, there's a /usr/lib/libstdc++.modules.json in the main package now that probably could go into libstdc++-dev instead
2025-07-27 14:40:21 <cely> A while ago, i tried enabling libsanitizer for more architectures, and it seems besides the x86_64 currently enabled now, it also builds on aarch64 and loongarch64
2025-07-27 14:45:00 <cely> In case you're interested, also tried enabling Cobol and Modula-2 a while ago, rebased my changes onto current main/gcc but haven't tried it again: https://tpaste.us/Xv7J
2025-07-27 14:47:28 <cely> It seems gcobol also uses libquadmath on architectures where it is available (x86* and ppc64le), that's why i've removed the amove of libquadmath.a and libquadmath.so in gfortran(), so they stay in the main package
2025-07-27 14:49:09 <cely> The languages for which LANG_COBOL=true are those where both gcobol and libgcobol built for me, if i recall correctly, there were some architectures that built gcobol alright, but then failed to build libgcobol
2025-07-27 15:32:25 <cely> Oh right, building Modula-2 requires bash in makedepends, so maybe it is all just a distraction, feel free to ignore then
2025-07-27 16:38:54 <cely> Fixed up patch: https://tpaste.us/Lyxp
2025-07-27 19:24:14 <cely> The architectures for which LANG_COBOL=true*
2025-07-27 20:25:28 <Ariadne> cely: i was considering enabling modula-2, but i don't think gcobol is ready yet
2025-07-27 20:26:07 <Ariadne> probably will not bump again until next gcc snapshot
2025-07-30 18:31:28 <Ariadne> thoughts on adding -mlsx to loongarch CFLAGS?