LG gram 17 boot issues

[onomatopoeia]

Hello all,

I've been stuck trying to install Gentoo on my new LG gram 17 laptop (Intel) for the past few days, and looking for some inspiration on how to figure out the problem. I've installed and used Gentoo for many years now on previous laptops and desktops, so I'm not new to this. But I seem to be totally stuck right now with my custom kernel.

Where to start?

I have 2 partitions, 1 GB for a FAT32 EFI System Partition (ESP), and the rest of 2TB as an encrypted LUKS partition (with a BTRFS filesystem and several sub-volumes). I have never set up Gentoo with disk encryption or BTRFS before, but I really wanted to try this out this time. In my chroot environment, I've configured and built my kernel, installed GRUB on the ESP, created an initial RAMFS with dracut (including the btrfs and crypt modules), and created the GRUB configuration file.

The problem is consistently this: GRUB loads its menu fine, it loads the kernel and initramfs (echoing the relevant messages right before doing so), then says:

EFI stub: Loaded initrd from LINUX_EFI_INITRD_MEDIA_GUID device path
EFI Stub: Measured initrd data into PCR 9

...and then nothing after that, not even a visible kernel panic.

Initially I identified all the drivers and firmware I needed and built everything into the kernel. The display driver is xe and not i915 but I've tried enabling both for good measure. I have also tried to enable and disable various typical framebuffer devices with no luck. Finally, I tried building both xe and i915 as modules (installing the modules, re-running dracut and checking that they had been picked up), again with no luck.

At this point, I suspect this has nothing to do with the display driver but something more basic that I'm missing due to my unfamiliarity with initramfs loading (I've previously used EFI stub to load the kernel directly, which is not an option here because of the encrypted partition).

Some details:

• lspci -k output on LiveUSB https://pastebin.com/uVeSSfgv
• dmesg output on LiveUSB https://pastebin.com/GvnxQ1V3
• kernel config (/proc/config) on LiveUSB https://pastebin.com/v8eJFjff
• kernel config (/usr/share/linux-6.12.58-gentoo/.config) on disk https://pastebin.com/yGTmJrUC

My build process is roughly:

With the kernel configuration referenced above (last one), I perform a "make -j8" and copy over arch/x86_64/boot/bzImage over to /boot, then run dracut with the following configuration:

Code: Select all

add_dracutmodules+=" crypt btrfs resume "
add_drivers+=" xe i915 "
early_microcode="yes"

(Originally I only had xe but then later figured maybe I needed both, but nothing changed the outcome.)

My generated grub.cfg is here: https://pastebin.com/3Pu17FcG

I would appreciate some help in pinpointing what I'm doing wrong, or at least make it print something more useful.

[pietinger]

onomatopoeia,

Welcome to Gentoo Forums!

In general, I would recommend that you also use 6.18 because you have a very new device:

Code: Select all

[    0.000000] SMBIOS 3.8.0 present.
[    0.000000] DMI: LG Electronics 17Z90TL-H.AUB6U3/17Z90TL, BIOS L2ZJ3190 X64 04/21/2025
[    0.127923] smpboot: CPU0: Intel(R) Core(TM) Ultra 7 258V (family: 0x6, model: 0xbd, stepping: 0x1)

[...] and copy over arch/x86_64/boot/bzImage over to /boot [...]

When copying, you did rename the kernel correctly to "vmlinuz-6.12.58-gentoo" (and also before creating the initramfs with dracut)?

Code: Select all

### BEGIN /etc/grub.d/10_linux ###
menuentry 'Gentoo GNU/Linux, with Linux 6.12.58-gentoo' --class gentoo --class gnu-linux --class gnu --class os $menuentry_id_option 'gnulinux-6.12.58-gentoo-advanced-d19e8e17-0247-4aaa-aaff-95a98ceb5df1' {
	set gfxpayload=text
	insmod gzio
	insmod part_gpt
	insmod fat
	search --no-floppy --fs-uuid --set=root F140-0F30
	echo	'Loading Linux 6.12.58-gentoo ...'
	linux	/vmlinuz-6.12.58-gentoo root=/dev/mapper/cryptroot ro rootflags=subvol=@ rd.luks.name=14948c94-7a10-4902-91e0-d853952119d6=cryptroot rootfstype=btrfs earlyprintk=vga,keep resume=/dev/mapper/cryptroot resume_offset=533760 
	echo	'Loading initial ramdisk ...'
	initrd	/initramfs-6.12.58-gentoo.img

If nothing is displayed after loading the kernel (not even a kernel panic), then it is usually a problem with missing framebuffer device and console settings (we have seen this very often in our forum):
https://wiki.gentoo.org/wiki/User:Pieti ... nd_Console
You already have these settings... that's strange... that's why I looked at everything and see a difference in the resolution... maybe this isn't possible:

Code: Select all

(yours)
CONFIG_VGA_CONSOLE=y
CONFIG_DUMMY_CONSOLE=y
CONFIG_DUMMY_CONSOLE_COLUMNS=160
CONFIG_DUMMY_CONSOLE_ROWS=50
CONFIG_FRAMEBUFFER_CONSOLE=y
# CONFIG_FRAMEBUFFER_CONSOLE_LEGACY_ACCELERATION is not set
CONFIG_FRAMEBUFFER_CONSOLE_DETECT_PRIMARY=y
# CONFIG_FRAMEBUFFER_CONSOLE_ROTATION is not set
# CONFIG_FRAMEBUFFER_CONSOLE_DEFERRED_TAKEOVER is not set
<==>
(default)
CONFIG_VGA_CONSOLE=y
CONFIG_DUMMY_CONSOLE=y
CONFIG_DUMMY_CONSOLE_COLUMNS=80
CONFIG_DUMMY_CONSOLE_ROWS=25
CONFIG_FRAMEBUFFER_CONSOLE=y
# CONFIG_FRAMEBUFFER_CONSOLE_LEGACY_ACCELERATION is not set
CONFIG_FRAMEBUFFER_CONSOLE_DETECT_PRIMARY=y
CONFIG_FRAMEBUFFER_CONSOLE_ROTATION=y
CONFIG_FRAMEBUFFER_CONSOLE_DEFERRED_TAKEOVER=y

There are probably other problems as well:

1. This LG notebook does not appear to have SATA, so you have completely disabled ATA (which includes SATA). The problem with this is that ATA selects SCSI (*), which you need at least to access USB sticks.

Code: Select all

# CONFIG_BLK_DEV_SD is not set
# CONFIG_CHR_DEV_ST is not set
# CONFIG_BLK_DEV_SR is not set
# CONFIG_CHR_DEV_SG is not set
# CONFIG_BLK_DEV_BSG is not set -> Help says: "This option is required by UDEV to access device serial numbers, etc.)"
# CONFIG_CHR_DEV_SCH is not set

# CONFIG_ATA is not set

but you want it for this ->

CONFIG_USB_STORAGE=y
CONFIG_USB_UAS=y

2. Since it is an Intel machine, you must enable this in order to enable SPI-Intel in the menu.

Code: Select all

# CONFIG_SPI_MEM is not set

3. This should also be enabled urgently:

Code: Select all

# CONFIG_IRQ_REMAP is not set

*) My SCSI configuration:

Code: Select all

#
# SCSI device support
#
CONFIG_SCSI_MOD=y
# CONFIG_RAID_ATTRS is not set
CONFIG_SCSI_COMMON=y
CONFIG_SCSI=y
CONFIG_SCSI_DMA=y
CONFIG_SCSI_PROC_FS=y

#
# SCSI support type (disk, tape, CD-ROM)
#
CONFIG_BLK_DEV_SD=y
# CONFIG_CHR_DEV_ST is not set
# CONFIG_BLK_DEV_SR is not set
CONFIG_CHR_DEV_SG=y
CONFIG_BLK_DEV_BSG=y
# CONFIG_CHR_DEV_SCH is not set
CONFIG_SCSI_CONSTANTS=y
# CONFIG_SCSI_LOGGING is not set
# CONFIG_SCSI_SCAN_ASYNC is not set

#
# SCSI Transports
#
CONFIG_SCSI_SPI_ATTRS=y
# CONFIG_SCSI_FC_ATTRS is not set
# CONFIG_SCSI_ISCSI_ATTRS is not set
# CONFIG_SCSI_SAS_ATTRS is not set
# CONFIG_SCSI_SAS_LIBSAS is not set
# CONFIG_SCSI_SRP_ATTRS is not set
# end of SCSI Transports

# CONFIG_SCSI_LOWLEVEL is not set
# CONFIG_SCSI_DH is not set
# end of SCSI device support

CONFIG_ATA=y
CONFIG_SATA_HOST=y
CONFIG_PATA_TIMINGS=y
CONFIG_ATA_VERBOSE_ERROR=y
# CONFIG_ATA_FORCE is not set
CONFIG_ATA_ACPI=y
# CONFIG_SATA_ZPODD is not set
CONFIG_SATA_PMP=y

#
# Controllers with non-SFF native interface
#
CONFIG_SATA_AHCI=y
CONFIG_SATA_MOBILE_LPM_POLICY=0

Look also into the <Help> (in "make menuconfig") of CONFIG_BLK_DEV_BSG=y ->

This option is required by UDEV to access device serial numbers, etc.

(the initramfs starts UDEV very early)

[onomatopoeia]

pietinger,

I really appreciate your thoughtful and deeply considered reply! I can't stress this enough. It is now working! Well, almost - a few loose ends remain - but the system boots and is functional.

tl;dr version - I took a circuitous route but here's what I think finally worked for me:

• Using the newer 6.18.6-gentoo kernel from ~amd64 (I had been completely oblivious to the fact that the LiveUSB had a much newer kernel).
• Having DRM_XE as a module rather than built into the kernel (building into the kernel did not work, and it works fine with or without i915 (which I removed)).

Interestingly, I realized along the way that it wasn't a kernel panic after all (or had stopped being one at some point), which I inferred based on being able to Ctrl-Alt-Del to reboot the system. With the xe driver build into the kernel (not as a module), the display remained blank, but amazingly, I was able to type stuff in blindly and validate that I could decrypt the disk and login as root.

I had a few remaining issues:

• iwlwifi firmware (now fixed) - referencing the file in /lib/firmware/intel/iwlwifi/<file> did not work, I had to reference the symlink to the file at /lib/firmware/<link>.
• SoF firmware doesn't seem to be referenced in dmesg, I must have the wrong firmware identified or something (more research needed).
• I don't know why the system won't work with the xe driver built into the kernel - any suggestions or explanation? (I can live with the modules, but I like the simplicity of building it in.)

Also, thank you for the callouts on CONFIG_ATA, CONFIG_SPI_MEM, and CONFIG_IRQ_REMAP. I have fixed them all.

Many thanks for your help!

[NeddySeagoon]

onomatopoeia,

I suspect the xe driver needs firmware, which it loads from /lib/firmware when its built as <M>.
Root is mounted before modules are loaded so that works.

Built in xe starts before root is mounted, so the firmware must be built into the kernel binary, or firmware loading fails and you are left with a blank console. As you discovered, it still works but it's difficult to read :)

[pietinger]

It is now working! Well, almost - a few loose ends remain - but the system boots and is functional. [...]

Happy to hear that!

[...] I had a few remaining issues:

• iwlwifi firmware (now fixed) - referencing the file in /lib/firmware/intel/iwlwifi/<file> did not work, I had to reference the symlink to the file at /lib/firmware/<link>.[1]
• SoF firmware doesn't seem to be referenced in dmesg, I must have the wrong firmware identified or something (more research needed).[2]
• I don't know why the system won't work with the xe driver built into the kernel - any suggestions or explanation? (I can live with the modules, but I like the simplicity of building it in.)[3]

1) Yes, this is normal. You must point to the file.
2) Do you have installed the package sys-firmware/sof-firmware?
3) Thats strange. There was a time where you needed KUNIT to be able to configure XE as built-in ... but then it was changed, so this is/should be not necessary anymore ... maybe they changed it again? ... I dont know. TBH Your machine should work also with ONLY i915 (which should be no problem if you configure it statically; be aware you need some firmware files for it).
-> https://wiki.gentoo.org/wiki/User:Pieti ... hic_Kernel

[...] Many thanks for your help!

You are very Welcome!

Have fun with Gentoo!

[onomatopoeia]

Many thanks pietinger and NeddySeagoon!

I believe I have resolved these issues, though I am yet to install X/Wayland and test the sound.

Here's the summary of it, in case anyone runs into the same problem.

The problem turned out to be that CONFIG_DRM_XE_DISPLAY (and optionally CONFIG_DRM_XE_DP_TUNNEL) could not be set with CONFIG_DRM_XE set to 'y' instead of 'm' due to a dependency in drivers/gpu/drm/xe/Kconfig.

Code: Select all

config DRM_XE_DISPLAY
        bool "Enable display support"
        depends on DRM_XE && DRM_XE=m && HAS_IOPORT
        select FB_IOMEM_HELPERS if DRM_FBDEV_EMULATION
        select I2C
        select I2C_ALGOBIT
        default y
        help
          Disable this option only if you want to compile out display support.
config DRM_XE_DP_TUNNEL
        bool "Enable DP tunnel support"
        depends on DRM_XE_DISPLAY
        depends on USB4
        select DRM_DISPLAY_DP_TUNNEL
        default y
        help
          Choose this option to detect DP tunnels and enable the Bandwidth
          Allocation mode for such tunnels. This allows using the maximum
          resolution allowed by the link BW on all displays sharing the
          link BW, for instance on a Thunderbolt link.
          If in doubt say "Y".

I was able to work around this by dropping a patch in /etc/portage/patches and removing the DRM_XE=m constraint, and it boots up fine now with everything built-in.

The problem with the sound drivers turned out to be an incorrect topology file. I did have sof-firmware package installed, but the driver was trying to load intel/sof-ipc4-tplg/sof-hda-generic-idisp-2ch.tplg whereas I had initially specified intel/sof-ipc4-tplg/sof-hda-generic-2ch.tplg.

Cheers!

[pietinger]

onomatopoeia,

thank you also for your report!

It will surely help other users with the same machine (especially sound; which is almost always a problem with very new hardware).

[Hu]

That depends on ... DRM_XE=m has been there since Xe support was first added, and appears to still be present as of v6.19-rc6. I could not readily find an explanation for why that was done originally, and it seems strange to me to require it to be a module, particularly if it seems to work correctly when built in.

[pietinger]

FYI: https://www.phoronix.com/news/LG-Gram-S ... x-Speakers
(contains also some information about 17)