Replace fdisk -l with sysfs-based disk enumeration (2tb+ drive support)

[tlaurion]

fdisk -l can deal with 2TB max size drives. Use linux sysfs instead.

Tested:

• qemu oem-factory reset with a canokey and public key exported to virt thumb drive's exclusive ext4 public partition
• qemu oem-facctory-reset with canokey and in-ram key generation and key to card, creating luks+exfat partition on virt thumb drive
• qemu-img 8tb qcow2 root attached: system information shows 8tb
• test on v540tu
• rootfs hash creation/hash checks on ubuntu/pureos/debian

Fixes #2034 (should have been the fix for #1884)

[Copilot]

Pull request overview

This PR replaces fdisk -l commands with sysfs-based disk enumeration to overcome fdisk's 2TB drive size limitation and improve busybox compatibility. The changes enable proper detection and display of large storage devices (e.g., 8TB drives).

Key changes:

• Introduces list_block_devices() function that reads from /sys/block/* instead of parsing fdisk -l output
• Replaces fdisk-based disk information gathering with sysfs-based size calculation
• Updates partition detection logic to use sysfs directory structure

Reviewed changes

Copilot reviewed 2 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
initrd/etc/gui_functions	Adds list_block_devices() function and updates show_system_info() to read disk sizes from sysfs
initrd/etc/functions	Adds list_block_devices() function, updates device_has_partitions() to check sysfs for partitions, updates is_gpt_bios_grub() to read partition types from sysfs, and updates detect_boot_device() to use new function
initrd/bin/root-hashes-gui.sh	Updates detect_root_device() to use list_block_devices() instead of fdisk
initrd/bin/oem-system-info-xx30	Updates disk information gathering to read from sysfs instead of fdisk
initrd/bin/config-gui.sh	Updates device listing to use list_block_devices() instead of fdisk

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 6 changed files in this pull request and generated 2 comments.

Comments suppressed due to low confidence (1)

initrd/etc/functions:875

• disk_info_sysfs() builds disk_info lines using "\n" and then echoes with echo -n, which produces literal backslash-n sequences unless the consumer interprets escapes. To make output consistent across callers, prefer emitting real newlines via printf (e.g., printf 'Disk /dev/%s: %s GB\n' ...) or use printf '%b' when printing an escape-containing buffer.

				local size_gb=$(((size_bytes * sector_size + 500000000) / 1000000000))
				disk_info="${disk_info}Disk /dev/${devname}: ${size_gb} GB\n"
				DEBUG "disk_info_sysfs: /dev/${devname} calculated as ${size_gb} GB (${size_bytes} sectors * ${sector_size} bytes/sector = $((size_bytes * sector_size)) bytes)"
			fi
		fi
	done
	echo -n "$disk_info"

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

is_gpt_bios_grub() checks for "PARTTYPENAME=BIOS boot" in sysfs uevent, but the kernel typically exports PARTTYPE (the type GUID) rather than a human-readable PARTTYPENAME (usually provided by udev). In initrd this will likely never match, so BIOS-boot partitions may no longer be filtered reliably. Suggest checking PARTTYPE against the BIOS boot GUID (21686148-6449-6E6F-744E-656564454649) and/or keeping the previous fdisk-based fallback; also update the comment that mentions a fallback that no longer exists.

Suggested change

	# Check partition type using sysfs if available, otherwise check for grub type
	# For GPT disks, check /sys/class/block/<dev>/<partition>/uevent for PARTTYPENAME
	local part_sys="/sys/class/block/$DEVICE/$(basename "$PART_DEV")"
	if [ -e "$part_sys/uevent" ]; then
	if grep -q "PARTTYPENAME=BIOS boot" "$part_sys/uevent"; then
	TRACE "$PART_DEV is a GPT BIOS grub partition"
	return 0
	fi
	fi
	# Fallback: try to detect using other sysfs attributes if available
	# For MBR disks, we would need to check partition type differently
	# Check partition type using sysfs if available.
	# For GPT disks, check /sys/class/block/<dev>/<partition>/uevent for PARTTYPE
	# matching the BIOS boot GUID (21686148-6449-6E6F-744E-656564454649). Some
	# environments may also expose PARTTYPENAME=BIOS boot via udev.
	local part_sys="/sys/class/block/$DEVICE/$(basename "$PART_DEV")"
	if [ -e "$part_sys/uevent" ]; then
	# Legacy/udev-based check: human-readable type name
	if grep -q "PARTTYPENAME=BIOS boot" "$part_sys/uevent"; then
	TRACE "$PART_DEV is a GPT BIOS grub partition (PARTTYPENAME)"
	return 0
	fi
	# Preferred check: kernel-exported type GUID
	if grep -qi "PARTTYPE=21686148-6449-6E6F-744E-656564454649" "$part_sys/uevent"; then
	TRACE "$PART_DEV is a GPT BIOS grub partition (PARTTYPE GUID)"
	return 0
	fi
	fi

[Copilot]

In disk_info_sysfs(), /sys/block//size is a count of 512-byte sectors (same units as blockdev --getsz), not “sectors of hw_sector_size”. Multiplying it by hw_sector_size/logical_block_size will over-report sizes (e.g., 4Kn devices) and can overflow sooner. Consider using bytes from blockdev --getsize64 "/dev/$devname" (still avoids busybox fdisk), or treat size as 512-byte sectors and multiply by 512 unconditionally.