00:07 <farkuhar> although I didn't have any of the crux.nu stylesheets for a proper test of the pmwiki rendering on my local host, I did use a consistent syntax for any new bullet points or indented paragraphs. Broken formatting should be a simple matter of search-and-replace, once we determine where I went wrong in using the pmwiki syntax.

08:00 <SiFuh> jaeger: https://gitlab.com/SiFuh/Documentation/-/blob/master/IMG_20220808_155556_912.jpg

08:02 <SiFuh> Kernel shrunk by 256

12:45 <farkuhar> SiFuh: your screenshot shows that you got lsusb running successfully (no unresolved symbols from libatomic). You fixed it without injecting the entire gcc package?

12:52 <SiFuh> I did what jaeger suggested. I added on line 323 @cp /usr/lib/libatomic.so.1 $(ROOTFS_DIR)/usr/lib

12:52 <SiFuh> Took it straight from my system

13:10 <farkuhar> Aye, that's a simple fix. I wonder if some of the other big packages that fill up the rootfs of the install media can also be obviated, leaving only the necessary .so files

13:11 <farkuhar> Then you could install even 3.7 on that ancient machine of yours with only 1GB of RAM

13:15 <SiFuh> Revdep ;-)

14:06 <jaeger> I did it slightly differently... copied libatomic from the bootstrapped gcc package instead of the live system

15:30 <SiFuh> jaeger: '#/dev/#ESP# /boot/efi ...' shouldn't that be '#/dev/#ESP# /boot ...' ?

15:31 <jaeger> Not typically, no

15:31 <jaeger> Technically either works, though

15:32 <SiFuh> UEFI looks for a partition and searches for the contents of a folder named EFI

15:34 <SiFuh> So you would need to use efibootmgr to point to the new location then

15:34 <stenur> UEFI does not look into your fstab

15:35 <SiFuh> No but the user does. /boot/efi/EFI/BOOT/BOOX64.EFI is kind of strange don't you think?

15:36 <stenur> ..no? I have /dev/nvme0n1p1 /media/efi vfat noauto,defaults 0 0

15:36 <SiFuh> Did you modify?

15:36 <stenur> it is not even mounted until i replace the stage1 kernel

15:36 <stenur> boot is somewhere else entirely

15:37 <SiFuh> %partition%/EFI/BOOT/ is the standard location

15:37 <stenur> /boot is on an encrypted btrfs partition

15:38 <stenur> you confuse me. what is it that you want?

15:43 <SiFuh> You don't really need to mount the EFI partition unless you want to modify it. It was requested that jaeger add the commented out EFI partition to the fstab. However, the mount point is /boot/efi which is the part the user/system mounts. UEFI doesn't know nor care about /boot/efi it only cares about the partition then looks into the foler EFI/BOOT by default. So if you mount the EFI partition as

15:43 <SiFuh> /boot/efi/ that means you should have a folder after that EFI/BOOT/ so theoretically it would be /boot/efi/EFI/BOOT/. If you do not use the EFI/BOOT/ part you will need to inform UEFI of the location of the *.EFI image.

15:43 <SiFuh> stenur: understand now?

15:45 <stenur> it is the one file /media/efi/EFI/Boot/bootx64.efi, yes

15:45 <stenur> the kernels i have there are not under EFI/:

15:46 <SiFuh> Mine is mounted as /boot and under boot I have efi/boot/bootx64.efi

15:46 <stenur> oh shit! new efibootmgr does not longer print command line, it uses a hexdump for arguments!!!

15:47 <SiFuh> The reason I brought it is up is that in the manual that farkuhar is writing it may confuse a beginner when using SYSLINUX

15:47 <stenur> ah no; they do both; ok wait

15:47 <stenur> Boot0001* kent HD(1,GPT,5d6d756b-5de2-4e5d-b043-8d4ae1bb6eb0,0x800,0x82000)/File(\ideapad-stage1.efi)root=/dev/nvme0n1p1 rootfstype=vfat init=/kent.sh

15:47 <stenur> Boot0002* kent-direct HD(1,GPT,5d6d756b-5de2-4e5d-b043-8d4ae1bb6eb0,0x800,0x82000)/File(\ideapad-stage1.efi)root=/dev/nvme0n1p1 rootfstype=vfat init=/kent-direct.sh rtw88_pci.disable_aspm=1 rc.hostname=kent

15:47 <stenur> etc etc

15:48 <stenur> that is directly /media/efi/ideapad-stage1.efi thus, not under /media/efi/EFI/Boot/

15:48 <jaeger> For clarification, /boot/efi is just common in documentation. You can mount it anywhere you like

15:48 <stenur> yeah

15:49 <SiFuh> Yes I know, but shouldn't be in line with CRUX documentation?

15:49 <stenur> just don't document syslinux, document efibootmgr for UEFI.

15:49 <jaeger> It isn't crux-specific and I'm pretty sure everything I've written with regards to UEFI uses /boot/efi

15:50 <SiFuh> Ayway, it is only a small thing.

15:50 <jaeger> It probably wouldn't hurt to be consistent in the crux docs about UEFI installation, though

15:52 <stenur> what he wants is that he mounts that VFAT thing into the filesystem plain, so that pkgadd for example simply stores early-ucode.cpio there, too

15:52 <stenur> since that stores in /boot/ there is a mismatch

15:52 <SiFuh> No, I don't use that ucode port

15:53 <jaeger> There's nothing that prevents you having stuff in /boot and /boot/efi

15:53 <stenur> i would not recommend users that need a handbook to use that VFAT thing as the real /boot

15:53 <jaeger> I keep kernels/system.map in /boot, bootloader still goes into /boot/efi

15:54 <SiFuh> I ran a test of another distro. I think it was Clear Linux 2 or 3 years ago, and I have never formated my EFI partition so I use the ucode that Clear Linux had installed.

15:54 <stenur> yes you can mount the same thing as /boot and /boot/efi, no? does that work from fstab? or must it be a --bind?

15:54 <jaeger> (except on an encrypted system where that doesn't work)

15:54 <jaeger> I would not mount the same partition to both places, there's no benefit to that in my mind

15:54 <stenur> and of secure boot i have no idea

15:54 <jaeger> Unless confusion is a benefit :P

15:55 <stenur> well you have to if you want the EFI VFAT as /boot?

15:55 <stenur> ..and the EFI VFAT as /boot/efi..

15:55 <jaeger> Why would I want that?

15:55 <stenur> i do not want it at all.

15:55 <SiFuh> ls

15:55 <jaeger> You are confusing me

15:56 <stenur> i also think we have reached the dead end of total confusion

15:56 <SiFuh> freestanding-00-intel-ucode.cpio freestanding-i915-firmware.cpio.xz <-- I am using these stenur

15:57 <stenur> ok; i do not know where these are located

15:57 <SiFuh> They were put in my efi partition but Clear Linux 3 years ago and I still use them, so no need to isntall the CRUX intel-ucode port

15:57 <jaeger> microcode may have been updated in those 3 years

15:58 <jaeger> possibly even for an old CPU

15:58 <SiFuh> I check every now and then.

16:01 <stenur> you are fine with OpenBSD, they print the necessary firmware IDs upon boot i think, at least for CPUs; these are the little things which help a lot, just last week i again compiled the ideapad kernel with a false firmware update for the other laptop which can use the same kernel, because i thought x-y-0 is a predecessor of x-y-4, 'had that once with -66 and -71. In both cases i felt as idiotic as i am.

18:13 <jaeger> SiFuh: looks like the LPSS stuff does indeed work on my zotac box

18:15 <SiFuh> jaeger: That's cool. I was reading in to it a little this morning. Seems to be quite important for a lot of low powered controllers

18:19 <SiFuh> https://www.kernelconfig.io/config_x86_intel_lpss?q=&kernelversion=5.15.55&arch=x86

18:20 <SiFuh> I also came accross this

18:20 <SiFuh> https://edc.intel.com/content/www/us/en/design/ipla/software-development-platforms/servers/platforms/intel-pentium-silver-and-intel-celeron-processors-datasheet-volume-1-of-2/005/intel-lpss-inter-integrated-circuit-i2c-controllers/

18:55 <jaeger> SiFuh: so which .config is the proper one for the ISO contrib dir? CD-5.15.55.config or the original -modular link?

19:03 <jaeger> I'm guessing that CD is for the CD config and -modular is intended to be a final installed system config, just want to make sure

19:05 <jaeger> (the same as .config vs .defconfig in iso.git)

20:35 <farkuhar> stenur: thanks for reminding us about intel-ucode and its footprint. How common is it for someone to pkgadd intel-ucode, when the fstab defines /boot itself as a mountpoint (not a subdir like /boot/efi) *and* that partition is not mounted at the time of running pkgadd? If that's not a common configuration, there's not much of a mismatch to worry about.