init: discoverable root partitions, a.k.a. an omittable "root=" cmdline option

Documentation/admin-guide/discoverable-root.rst

@@ -0,0 +1,33 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+.. _discoverable_root:
+
+Discoverable root partitions
+============================
+
+On EFI systems using a supported architecture, the kernel can discover the root
+block device from GPT partition type UUID metadata on the disk containing the
+active EFI System Partition.
+
+This follows the `Discoverable Partitions Specification`_ which defines a list
+of architecture-specific root partition type UUIDs.
+
+Specifying ``root=`` on the kernel command line takes precedence and entirely
+disables this automatic root partition discovery.
+
+The disk to search is identified by the Boot Loader Interface
+``LoaderDevicePartUUID`` EFI variable. If multiple partitions on that disk match
+the architecture root partition type UUID, the kernel selects the first match in
+block device enumeration order. Systems should not expose multiple eligible root
+partitions unless that ordering is intended.
+
+Partitions marked with the DPS ``no-auto`` GPT attribute are skipped. This
+allows a partition with an otherwise discoverable type UUID to opt out from
+automatic discovery.
+
+The DPS read-only attribute is not enforced by kernel root discovery. The
+root filesystem is mounted read-only by default unless ``rw`` is specified,
+and user space remains responsible for later remount policy.
+
+.. _Discoverable Partitions Specification:
+   https://uapi-group.org/specifications/specs/discoverable_partitions_specification/

Documentation/admin-guide/index.rst

@@ -50,6 +50,7 @@ Booting the kernel
 
    bootconfig
    kernel-parameters
+   discoverable-root
    efi-stub
    initrd
 

Documentation/admin-guide/kernel-parameters.txt

@@ -6699,6 +6699,11 @@ Kernel parameters
 			ramdisk, "nfs" and "cifs" for root on a network file
 			system, or "mtd" and "ubi" for mounting from raw flash.
 
+			If this option is omitted, the kernel may try to
+			discover the root block device from the GPT partition
+			type UUID metadata when additional requirements are met.
+			See Documentation/admin-guide/discoverable-root.rst.
+
 	rootdelay=	[KNL] Delay (in seconds) to pause before attempting to
 			mount the root filesystem
 

arch/alpha/Kconfig

@@ -5,6 +5,7 @@ config ALPHA
 	select ARCH_32BIT_USTAT_F_TINODE
 	select ARCH_HAS_CURRENT_STACK_POINTER
 	select ARCH_HAS_DMA_OPS if PCI
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_MIGHT_HAVE_PC_PARPORT
 	select ARCH_MIGHT_HAVE_PC_SERIO
 	select ARCH_MODULE_NEEDS_WEAK_PER_CPU if SMP

arch/alpha/include/asm/dps_root.h

@@ -0,0 +1,8 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_ALPHA_DPS_ROOT_H
+#define _ASM_ALPHA_DPS_ROOT_H
+
+#define DPS_ROOT_PARTITION_TYPE_UUID "6523f8ae-3eb1-4e2a-a05a-18b695ae656f"
+
+#endif /* _ASM_ALPHA_DPS_ROOT_H */

arch/arc/Kconfig

@@ -10,6 +10,7 @@ config ARC
 	select ARCH_HAS_CACHE_LINE_SIZE
 	select ARCH_HAS_DEBUG_VM_PGTABLE
 	select ARCH_HAS_DMA_PREP_COHERENT
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_HAS_PTE_SPECIAL
 	select ARCH_HAS_SETUP_DMA_OPS
 	select ARCH_HAS_SYNC_DMA_FOR_CPU

arch/arc/include/asm/dps_root.h

@@ -0,0 +1,8 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_ARC_DPS_ROOT_H
+#define _ASM_ARC_DPS_ROOT_H
+
+#define DPS_ROOT_PARTITION_TYPE_UUID "d27f46ed-2919-4cb8-bd25-9531f3c16534"
+
+#endif /* _ASM_ARC_DPS_ROOT_H */

arch/arm/Kconfig

@@ -14,6 +14,7 @@ config ARM
 	select ARCH_HAS_DMA_ALLOC if MMU
 	select ARCH_HAS_DMA_OPS
 	select ARCH_HAS_DMA_WRITE_COMBINE if !ARM_DMA_MEM_BUFFERABLE
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_HAS_ELF_RANDOMIZE
 	select ARCH_HAS_FORTIFY_SOURCE
 	select ARCH_HAS_KEEPINITRD

arch/arm/include/asm/dps_root.h

@@ -0,0 +1,8 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_ARM_DPS_ROOT_H
+#define _ASM_ARM_DPS_ROOT_H
+
+#define DPS_ROOT_PARTITION_TYPE_UUID "69dad710-2ce4-4e3c-b16c-21a1d49abed3"
+
+#endif /* _ASM_ARM_DPS_ROOT_H */

arch/arm64/Kconfig

@@ -26,6 +26,7 @@ config ARM64
 	select ARCH_HAS_DEBUG_VM_PGTABLE
 	select ARCH_HAS_DMA_OPS if XEN
 	select ARCH_HAS_DMA_PREP_COHERENT
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_HAS_ACPI_TABLE_UPGRADE if ACPI
 	select ARCH_HAS_FAST_MULTIPLIER
 	select ARCH_HAS_FORTIFY_SOURCE

arch/arm64/include/asm/dps_root.h

@@ -0,0 +1,8 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_ARM64_DPS_ROOT_H
+#define _ASM_ARM64_DPS_ROOT_H
+
+#define DPS_ROOT_PARTITION_TYPE_UUID "b921b045-1df0-41c3-af44-4c6f280d3fae"
+
+#endif /* _ASM_ARM64_DPS_ROOT_H */

arch/loongarch/Kconfig

@@ -17,6 +17,7 @@ config LOONGARCH
 	select ARCH_HAS_CPU_FINALIZE_INIT
 	select ARCH_HAS_CURRENT_STACK_POINTER
 	select ARCH_HAS_DEBUG_VM_PGTABLE
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_HAS_FAST_MULTIPLIER
 	select ARCH_HAS_FORTIFY_SOURCE
 	select ARCH_HAS_KCOV

arch/loongarch/include/asm/dps_root.h

@@ -0,0 +1,8 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_LOONGARCH_DPS_ROOT_H
+#define _ASM_LOONGARCH_DPS_ROOT_H
+
+#define DPS_ROOT_PARTITION_TYPE_UUID "77055800-792c-4f94-b39a-98c91b762bb6"
+
+#endif /* _ASM_LOONGARCH_DPS_ROOT_H */

arch/mips/Kconfig

@@ -10,6 +10,7 @@ config MIPS
 	select ARCH_HAS_CURRENT_STACK_POINTER
 	select ARCH_HAS_DEBUG_VIRTUAL if !64BIT
 	select ARCH_HAS_DMA_OPS if MACH_JAZZ
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_HAS_FORTIFY_SOURCE
 	select ARCH_HAS_KCOV
 	select ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE if !EVA

arch/mips/include/asm/dps_root.h

@@ -0,0 +1,20 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_MIPS_DPS_ROOT_H
+#define _ASM_MIPS_DPS_ROOT_H
+
+#ifdef CONFIG_CPU_LITTLE_ENDIAN
+#ifdef CONFIG_64BIT
+#define DPS_ROOT_PARTITION_TYPE_UUID "700bda43-7a34-4507-b179-eeb93d7a7ca3"
+#else
+#define DPS_ROOT_PARTITION_TYPE_UUID "37c58c8a-d913-4156-a25f-48b1b64e07f0"
+#endif
+#else
+#ifdef CONFIG_64BIT
+#define DPS_ROOT_PARTITION_TYPE_UUID "d113af76-80ef-41b4-bdb6-0cff4d3d4a25"
+#else
+#define DPS_ROOT_PARTITION_TYPE_UUID "e9434544-6e2c-47cc-bae2-12d6deafb44c"
+#endif
+#endif
+
+#endif /* _ASM_MIPS_DPS_ROOT_H */

arch/parisc/Kconfig

@@ -24,6 +24,7 @@ config PARISC
 	select ARCH_STACKWALK
 	select ARCH_HAS_CACHE_LINE_SIZE
 	select ARCH_HAS_DEBUG_VM_PGTABLE
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select HAVE_RELIABLE_STACKTRACE
 	select RTC_CLASS
 	select RTC_DRV_GENERIC

arch/parisc/include/asm/dps_root.h

@@ -0,0 +1,8 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_PARISC_DPS_ROOT_H
+#define _ASM_PARISC_DPS_ROOT_H
+
+#define DPS_ROOT_PARTITION_TYPE_UUID "1aacdb3b-5444-4138-bd9e-e5c2239b2346"
+
+#endif /* _ASM_PARISC_DPS_ROOT_H */

arch/powerpc/Kconfig

@@ -134,6 +134,7 @@ config PPC
 	select ARCH_HAS_DEVMEM_IS_ALLOWED
 	select ARCH_HAS_DMA_MAP_DIRECT 		if PPC_PSERIES
 	select ARCH_HAS_DMA_OPS			if PPC64
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_HAS_FORTIFY_SOURCE
 	select ARCH_HAS_GCOV_PROFILE_ALL
 	select ARCH_HAS_GIGANTIC_PAGE		if ARCH_SUPPORTS_HUGETLBFS

arch/powerpc/include/asm/dps_root.h

@@ -0,0 +1,16 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_POWERPC_DPS_ROOT_H
+#define _ASM_POWERPC_DPS_ROOT_H
+
+#ifdef CONFIG_PPC64
+#ifdef CONFIG_CPU_LITTLE_ENDIAN
+#define DPS_ROOT_PARTITION_TYPE_UUID "c31c45e6-3f39-412e-80fb-4809c4980599"
+#else
+#define DPS_ROOT_PARTITION_TYPE_UUID "912ade1d-a839-4913-8964-a10eee08fbd2"
+#endif
+#else
+#define DPS_ROOT_PARTITION_TYPE_UUID "1de3f1ef-fa98-47b5-8dcd-4a860a654d78"
+#endif
+
+#endif /* _ASM_POWERPC_DPS_ROOT_H */

arch/riscv/Kconfig

@@ -28,6 +28,7 @@ config RISCV
 	select ARCH_HAS_DEBUG_VIRTUAL if MMU
 	select ARCH_HAS_DEBUG_VM_PGTABLE
 	select ARCH_HAS_DEBUG_WX
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_HAS_ELF_CORE_EFLAGS if BINFMT_ELF && ELF_CORE
 	select ARCH_HAS_FAST_MULTIPLIER
 	select ARCH_HAS_FORTIFY_SOURCE

arch/riscv/include/asm/dps_root.h

@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_RISCV_DPS_ROOT_H
+#define _ASM_RISCV_DPS_ROOT_H
+
+#ifdef CONFIG_64BIT
+#define DPS_ROOT_PARTITION_TYPE_UUID "72ec70a6-cf74-40e6-bd49-4bda08e8f224"
+#else
+#define DPS_ROOT_PARTITION_TYPE_UUID "60d5a7fe-8e7d-435c-b714-3dd8162144e1"
+#endif
+
+#endif /* _ASM_RISCV_DPS_ROOT_H */

arch/s390/Kconfig

@@ -91,6 +91,7 @@ config S390
 	select ARCH_HAS_DEBUG_WX
 	select ARCH_HAS_DEVMEM_IS_ALLOWED
 	select ARCH_HAS_DMA_OPS if PCI
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_HAS_ELF_RANDOMIZE
 	select ARCH_HAS_FORCE_DMA_UNENCRYPTED
 	select ARCH_HAS_FORTIFY_SOURCE

arch/s390/include/asm/dps_root.h

@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_S390_DPS_ROOT_H
+#define _ASM_S390_DPS_ROOT_H
+
+#ifdef CONFIG_64BIT
+#define DPS_ROOT_PARTITION_TYPE_UUID "5eead9a9-fe09-4a1e-a1d7-520d00531306"
+#else
+#define DPS_ROOT_PARTITION_TYPE_UUID "08a7acea-624c-4a20-91e8-6e0fa67d23f9"
+#endif
+
+#endif /* _ASM_S390_DPS_ROOT_H */

arch/x86/Kconfig

@@ -82,6 +82,7 @@ config X86
 	select ARCH_HAS_DEBUG_VM_PGTABLE	if !X86_PAE
 	select ARCH_HAS_DEVMEM_IS_ALLOWED
 	select ARCH_HAS_DMA_OPS			if GART_IOMMU || XEN
+	select ARCH_HAS_DPS_ROOT_PARTITION_TYPE_UUID
 	select ARCH_HAS_EARLY_DEBUG		if KGDB
 	select ARCH_HAS_ELF_RANDOMIZE
 	select ARCH_HAS_EXECMEM_ROX		if X86_64 && STRICT_MODULE_RWX

arch/x86/include/asm/dps_root.h

@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#ifndef _ASM_X86_DPS_ROOT_H
+#define _ASM_X86_DPS_ROOT_H
+
+#ifdef CONFIG_X86_64
+#define DPS_ROOT_PARTITION_TYPE_UUID "4f68bce3-e8cd-4db1-96e7-fbcaf984b709"
+#else
+#define DPS_ROOT_PARTITION_TYPE_UUID "44479540-f297-41b2-9af7-d131d5f0458a"
+#endif
+
+#endif /* _ASM_X86_DPS_ROOT_H */

block/blk.h

@@ -604,6 +604,7 @@ void blk_free_ext_minor(unsigned int minor);
 #define ADDPART_FLAG_RAID	1
 #define ADDPART_FLAG_WHOLEDISK	2
 #define ADDPART_FLAG_READONLY	4
+#define ADDPART_FLAG_NO_AUTO	8
 int bdev_add_partition(struct gendisk *disk, int partno, sector_t start,
 		sector_t length);
 int bdev_del_partition(struct gendisk *disk, int partno);

block/early-lookup.c

@@ -5,10 +5,12 @@
  */
 #include <linux/blkdev.h>
 #include <linux/ctype.h>
+#include <linux/uuid.h>
 
 struct uuidcmp {
 	const char *uuid;
 	int len;
+	struct gendisk *disk;
 };
 
 /**
@@ -45,13 +47,11 @@ static int __init match_dev_by_uuid(struct device *dev, const void *data)
  */
 static int __init devt_from_partuuid(const char *uuid_str, dev_t *devt)
 {
-	struct uuidcmp cmp;
+	struct uuidcmp cmp = { .uuid = uuid_str };
 	struct device *dev = NULL;
 	int offset = 0;
 	char *slash;
 
-	cmp.uuid = uuid_str;
-
 	slash = strchr(uuid_str, '/');
 	/* Check for optional partition number offset attributes. */
 	if (slash) {
@@ -252,6 +252,68 @@ int __init early_lookup_bdev(const char *name, dev_t *devt)
 	return devt_from_devnum(name, devt);
 }
 
+#ifdef CONFIG_DPS_ROOT_AUTO_DISCOVERY
+/**
+ * match_dev_by_type_uuid - callback for finding a partition using its type UUID
+ * @dev:	device passed in by the caller
+ * @data:	opaque pointer to the desired struct uuidcmp to match
+ *
+ * Returns: 1 if the device matches, and 0 otherwise.
+ */
+static int __init match_dev_by_type_uuid(struct device *dev, const void *data)
+{
+	struct block_device *bdev = dev_to_bdev(dev);
+	const struct uuidcmp *cmp = data;
+
+	return bdev->bd_disk == cmp->disk && bdev->bd_meta_info &&
+	       !bdev_test_flag(bdev, BD_NO_AUTO_DISCOVERY) &&
+	       !strcasecmp(cmp->uuid, bdev->bd_meta_info->type_uuid);
+}
+
+/**
+ * early_lookup_bdev_by_type_uuid - look up a partition by its type UUID
+ * @type_uuid:		partition type UUID to search for
+ * @efi_partuuid:	partition UUID identifying the active EFI partition
+ * @devt:		matching dev_t result
+ *
+ * This helper follows the Discoverable Partitions Specification rules. It uses
+ * @efi_partuuid to find the disk containing the active EFI System Partition,
+ * then searches only partitions on that disk for the partition type UUID
+ * specified by @type_uuid.
+ *
+ * Returns: 0 on success or a negative error code on failure.
+ */
+int __init early_lookup_bdev_by_type_uuid(const char *type_uuid,
+					  const char *efi_partuuid, dev_t *devt)
+{
+	struct uuidcmp efi_cmp = {
+		.uuid = efi_partuuid,
+		.len = UUID_STRING_LEN,
+	};
+	struct uuidcmp type_cmp = {
+		.uuid = type_uuid,
+	};
+	struct device *efi_dev;
+	struct device *type_dev;
+
+	efi_dev = class_find_device(&block_class, NULL, &efi_cmp,
+				    &match_dev_by_uuid);
+	if (!efi_dev)
+		return -ENODEV;
+
+	type_cmp.disk = dev_to_disk(efi_dev);
+	type_dev = class_find_device(&block_class, NULL, &type_cmp,
+				     &match_dev_by_type_uuid);
+	put_device(efi_dev);
+	if (!type_dev)
+		return -ENODEV;
+
+	*devt = type_dev->devt;
+	put_device(type_dev);
+	return 0;
+}
+#endif
+
 static char __init *bdevt_str(dev_t devt, char *buf)
 {
 	if (MAJOR(devt) <= 0xff && MINOR(devt) <= 0xff) {

block/partitions/core.c

@@ -382,6 +382,8 @@ static struct block_device *add_partition(struct gendisk *disk, int partno,
 
 	if (flags & ADDPART_FLAG_READONLY)
 		bdev_set_flag(bdev, BD_READ_ONLY);
+	if (flags & ADDPART_FLAG_NO_AUTO)
+		bdev_set_flag(bdev, BD_NO_AUTO_DISCOVERY);
 
 	/* everything is up and running, commence */
 	err = xa_insert(&disk->part_tbl, partno, bdev, GFP_KERNEL);

block/partitions/efi.c

@@ -739,9 +739,12 @@ int efi_partition(struct parsed_partitions *state)
 		/* If this is a RAID volume, tell md */
 		if (!efi_guidcmp(ptes[i].partition_type_guid, PARTITION_LINUX_RAID_GUID))
 			state->parts[i + 1].flags = ADDPART_FLAG_RAID;
+		if (le64_to_cpu(ptes[i].attributes) & GPT_ATTRIBUTE_NO_AUTO)
+			state->parts[i + 1].flags |= ADDPART_FLAG_NO_AUTO;
 
 		info = &state->parts[i + 1].info;
 		efi_guid_to_str(&ptes[i].unique_partition_guid, info->uuid);
+		efi_guid_to_str(&ptes[i].partition_type_guid, info->type_uuid);
 
 		/* Naively convert UTF16-LE to 7 bits. */
 		label_max = min(ARRAY_SIZE(info->volname) - 1,