[GRUB PARTUUID PATCH V9 0/5] Add PARTUUID detection support

[GRUB PARTUUID PATCH V9 0/5] Add PARTUUID detection support

[Nicholas Vinson]

Changes from Patch v8:
    - Renamed GRUB_ENABLE_LINUX_PARTUUID to GRUB_DISABLE_LINUX_PARTUUID
    - Updated the 10_linux logic so GRUB_ENABLE_LINUX_PARTUUID and
        GRUB_ENABLE_LINUX_UUID would behave more independently.
    - Documented interactions of GRUB_DISABLE_LINUX_UUID,
        GRUB_DISABLE_LINUX_PARTUUID, and initramfs detection in commit
        message.
    - Added optional patch that sets GRUB_DISABLE_LINUX_PARTUUID to true
        by default.
    - Fixed merge conflicts between this patchset and upstream's master
        branch.

Changes from Patch v7:
    - Changed checks in probe_partuuid() to use the variable 'p' instead
      of disk->partition
    - Moved 'disk->partition = p;' assignment to inside the
        'p & p->parent == NULL' statement
    - Moved 'disk->partition = p->parent;' assignment to before partmap
          name checks.
    - Fixed formatting issues.
    - Copied Daniel Kiper's reviewed-by line to unaltered patches.

Changes from Patch v6:
    - Corrected spelling and grammatical errors in description text for
      GRUB_ENABLE_LINUX_PARTUUID
    - Moved disk->partition save & restore logic to beginning and end of
      probe_partuuid()
    - Fixed formatting errors in probe_partuuid

Changes from Patch v5:
    - Added sign-off by lines
    - Fixed formatting errors found by Daniel Kiper

Changes from Patch v4:
    - Updated grub.texi to reflect new behavior for
      GRUB_ENABLE_LINUX_PARTUUID

    - Updated 10_linux.in logic to favor the PARTUUID when
      GRUB_ENABLE_LINUX_PARTUUID is enabled and GRUB_DISABLE_LINUX_UUID
      is disabled.

Changes from Patch v3:
    - Removed flex-2.6.3 compatibility patch

    - Removed Steve Kenton's patch

Changes from Patch v2:
    - Added flex-2.6.3 compatibility patch

    - Fixed a GPT partition read error

    - Added Steve Kenton's patch

    - Changed struct grub_part_gpt_type name to struct
      grub_part_gpt_part_guid

    - Changed grub_part_gpt_type_t typedef name to grub_part_gpt_guid_t

    - Added sprint_gpt_guid to Steve Kenton's patch

    - Updated v1 and Steve Kenton's patch to use similar methods when
      reading partition GUIDs.

Changes from Patch v1:
    - Added GRUB_ENABLE_LINUX_PARTUUID variable description to grub.texi

    - Removed added gpt_part_guid copy logic from
      grub_gpt_partition_map_iterate()

    - Removed added NT disk signature copy logic from
      grub_partition_msdos_iterate()

    - Removed modifications to partition number increment logic

    - Removed added guid union definition.

    - Added GRUB_ENABLE_LINUX_PARTUUID to grub-mkconfig.in export list

    - Moved PRINT_GPT_PARTTYPE printing logic to print_gpt_guid()
      function in grub-probe.c

    - Updated PRINT_GPT_PARTTYPE case to call print_gpt_guid() function
      in grub-probe.c.

    - Created probe_partuuid() function in grub-probe.c

    - Updated print == PRINT_PARTUUID check logic in probe() to call
      probe_partuuid().

    - Updated UUID logic in 10_linux.in to enable root=PARTUUID feature
      only if GRUB_DISABLE_LINUX_UUID is not set to true,
      and GRUB_DEVICE_PARTUUID is not empty, GRUB_ENABLE_LINUX_PARTUUID
      is set to true.

Hello,

This is a request to add PARTUUID detection support grub-probe for MBR
and GPT partition schemes.  The Linux kernel supports mounting the root
filesystem by Linux device name or by the Partition [GU]UID.  GRUB's
mkconfig, however, currently only supports specifying the rootfs in the
kernel command-line by Linux device name unless an initramfs is also
present.  When an initramfs is present GRUB's mkconfig will set the
kernel's root parameter value to either the Linux device name or to the
filesystem [GU]UID.

Therefore, the only way to protect a Linux system from failing to boot
when its Linux storage device names change is to either manually edit
grub.cfg or /etc/default/grub and append root=PARTUUID=xxx to the
command-line or create an initramfs that understands how to mount
devices by filesystem [G]UID and let grub-mkconfig pass the filesystem
[GU]UID to the initramfs.

The goal of this patch set is to enable root=PARTUUID=xxx support in
grub-mkconfig, so that users don't have to manually edit
/etc/default/grub or grub.cfg, or create an initramfs for the sole
purpose of having a robust bootloader configuration for Linux.

Thanks,
Nicholas Vinson

Nicholas Vinson (5):
  Centralize guid prints
  Update grub_gpt_partentry
  Add PARTUUID detection support to grub-probe
  Update grub script template files
  Default to disabling partition UUID support

 docs/grub.texi               | 11 +++++++
 grub-core/disk/ldm.c         |  2 +-
 grub-core/partmap/gpt.c      |  4 +--
 include/grub/gpt_partition.h |  8 ++---
 util/grub-install.c          |  2 +-
 util/grub-mkconfig.in        |  3 ++
 util/grub-probe.c            | 76 +++++++++++++++++++++++++++++++++++---------
 util/grub.d/10_linux.in      | 22 +++++++++++--
 8 files changed, 102 insertions(+), 26 deletions(-)

-- 
2.16.3

[GRUB PARTUUID PATCH V9 1/5] Centralize guid prints

[Nicholas Vinson]

Define print_gpt_guid(), so there is a central function for printing
GUID strings.  This change is a precursor for later patches which rely
on this logic.

Signed-off-by: Nicholas Vinson <nvinson234@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>

---
 util/grub-probe.c | 30 +++++++++++++++---------------
 1 file changed, 15 insertions(+), 15 deletions(-)

diff --git a/util/grub-probe.c b/util/grub-probe.c
index 8ac527d2f..21cb80fbe 100644
--- a/util/grub-probe.c
+++ b/util/grub-probe.c
@@ -129,6 +129,20 @@ get_targets_string (void)
   return str;
 }
 
+static int
+print_gpt_guid (grub_gpt_part_guid_t guid)
+{
+  guid.data1 = grub_le_to_cpu32 (guid.data1);
+  guid.data2 = grub_le_to_cpu16 (guid.data2);
+  guid.data3 = grub_le_to_cpu16 (guid.data3);
+
+  return grub_printf ("%08x-%04x-%04x-%02x%02x-%02x%02x%02x%02x%02x%02x",
+		      guid.data1, guid.data2, guid.data3, guid.data4[0],
+		      guid.data4[1], guid.data4[2], guid.data4[3],
+		      guid.data4[4], guid.data4[5], guid.data4[6],
+		      guid.data4[7]);
+}
+
 static void
 do_print (const char *x, void *data)
 {
@@ -641,21 +655,7 @@ probe (const char *path, char **device_names, char delim)
 
               if (grub_disk_read (dev->disk, p->offset, p->index,
                                   sizeof (gptdata), &gptdata) == 0)
-                {
-                  grub_gpt_part_type_t gpttype;
-                  gpttype.data1 = grub_le_to_cpu32 (gptdata.type.data1);
-                  gpttype.data2 = grub_le_to_cpu16 (gptdata.type.data2);
-                  gpttype.data3 = grub_le_to_cpu16 (gptdata.type.data3);
-                  grub_memcpy (gpttype.data4, gptdata.type.data4, 8);
-
-                  grub_printf ("%08x-%04x-%04x-%02x%02x-%02x%02x%02x%02x%02x%02x",
-                               gpttype.data1, gpttype.data2,
-                               gpttype.data3, gpttype.data4[0], 
-                               gpttype.data4[1], gpttype.data4[2],
-                               gpttype.data4[3], gpttype.data4[4],
-                               gpttype.data4[5], gpttype.data4[6],
-                               gpttype.data4[7]);
-                }
+		print_gpt_guid(gptdata.type);
               dev->disk->partition = p;
             }
           putchar (delim);
-- 
2.16.3

[GRUB PARTUUID PATCH V9 2/5] Update grub_gpt_partentry

[Nicholas Vinson]

Rename grub_gpt_part_type to grub_gpt_part_guid and update
grub_gpt_partentry to use this type for both the partition type GUID
string and the partition GUID string entries.  This change ensures that
the two GUID fields are handled more consistently and helps to simplify
the changes needed to add Linux partition GUID support.

Signed-off-by: Nicholas Vinson <nvinson234@gmail.com>

---
 grub-core/disk/ldm.c         | 2 +-
 grub-core/partmap/gpt.c      | 4 ++--
 include/grub/gpt_partition.h | 8 ++++----
 util/grub-install.c          | 2 +-
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/grub-core/disk/ldm.c b/grub-core/disk/ldm.c
index 0f978ad05..2a22d2d6c 100644
--- a/grub-core/disk/ldm.c
+++ b/grub-core/disk/ldm.c
@@ -135,7 +135,7 @@ msdos_has_ldm_partition (grub_disk_t dsk)
   return has_ldm;
 }
 
-static const grub_gpt_part_type_t ldm_type = GRUB_GPT_PARTITION_TYPE_LDM;
+static const grub_gpt_part_guid_t ldm_type = GRUB_GPT_PARTITION_TYPE_LDM;
 
 /* Helper for gpt_ldm_sector.  */
 static int
diff --git a/grub-core/partmap/gpt.c b/grub-core/partmap/gpt.c
index 83bcba779..103f6796f 100644
--- a/grub-core/partmap/gpt.c
+++ b/grub-core/partmap/gpt.c
@@ -33,10 +33,10 @@ static grub_uint8_t grub_gpt_magic[8] =
     0x45, 0x46, 0x49, 0x20, 0x50, 0x41, 0x52, 0x54
   };
 
-static const grub_gpt_part_type_t grub_gpt_partition_type_empty = GRUB_GPT_PARTITION_TYPE_EMPTY;
+static const grub_gpt_part_guid_t grub_gpt_partition_type_empty = GRUB_GPT_PARTITION_TYPE_EMPTY;
 
 #ifdef GRUB_UTIL
-static const grub_gpt_part_type_t grub_gpt_partition_type_bios_boot = GRUB_GPT_PARTITION_TYPE_BIOS_BOOT;
+static const grub_gpt_part_guid_t grub_gpt_partition_type_bios_boot = GRUB_GPT_PARTITION_TYPE_BIOS_BOOT;
 #endif
 
 /* 512 << 7 = 65536 byte sectors.  */
diff --git a/include/grub/gpt_partition.h b/include/grub/gpt_partition.h
index 9668a68c3..7a93f4329 100644
--- a/include/grub/gpt_partition.h
+++ b/include/grub/gpt_partition.h
@@ -22,14 +22,14 @@
 #include <grub/types.h>
 #include <grub/partition.h>
 
-struct grub_gpt_part_type
+struct grub_gpt_part_guid
 {
   grub_uint32_t data1;
   grub_uint16_t data2;
   grub_uint16_t data3;
   grub_uint8_t data4[8];
 } GRUB_PACKED;
-typedef struct grub_gpt_part_type grub_gpt_part_type_t;
+typedef struct grub_gpt_part_guid grub_gpt_part_guid_t;
 
 #define GRUB_GPT_PARTITION_TYPE_EMPTY \
   { 0x0, 0x0, 0x0, \
@@ -70,8 +70,8 @@ struct grub_gpt_header
 
 struct grub_gpt_partentry
 {
-  grub_gpt_part_type_t type;
-  grub_uint8_t guid[16];
+  grub_gpt_part_guid_t type;
+  grub_gpt_part_guid_t guid;
   grub_uint64_t start;
   grub_uint64_t end;
   grub_uint64_t attrib;
diff --git a/util/grub-install.c b/util/grub-install.c
index 690f180c5..78d0138cb 100644
--- a/util/grub-install.c
+++ b/util/grub-install.c
@@ -714,7 +714,7 @@ is_prep_partition (grub_device_t dev)
       if (grub_disk_read (dev->disk, p->offset, p->index,
 			  sizeof (gptdata), &gptdata) == 0)
 	{
-	  const grub_gpt_part_type_t template = {
+	  const grub_gpt_part_guid_t template = {
 	    grub_cpu_to_le32_compile_time (0x9e1a2d38),
 	    grub_cpu_to_le16_compile_time (0xc612),
 	    grub_cpu_to_le16_compile_time (0x4316),
-- 
2.16.3

[GRUB PARTUUID PATCH V9 3/5] Add PARTUUID detection support to grub-probe

[Nicholas Vinson]

Add PARTUUID detection support grub-probe for MBR and GPT partition
schemes.

Signed-off-by: Nicholas Vinson <nvinson234@gmail.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>

---
 util/grub-probe.c | 46 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)

diff --git a/util/grub-probe.c b/util/grub-probe.c
index 21cb80fbe..e91ae730e 100644
--- a/util/grub-probe.c
+++ b/util/grub-probe.c
@@ -28,6 +28,7 @@
 #include <grub/partition.h>
 #include <grub/msdos_partition.h>
 #include <grub/gpt_partition.h>
+#include <grub/i386/pc/boot.h>
 #include <grub/emu/hostdisk.h>
 #include <grub/emu/getroot.h>
 #include <grub/term.h>
@@ -62,6 +63,7 @@ enum {
   PRINT_DRIVE,
   PRINT_DEVICE,
   PRINT_PARTMAP,
+  PRINT_PARTUUID,
   PRINT_ABSTRACTION,
   PRINT_CRYPTODISK_UUID,
   PRINT_HINT_STR,
@@ -85,6 +87,7 @@ static const char *targets[] =
     [PRINT_DRIVE]              = "drive",
     [PRINT_DEVICE]             = "device",
     [PRINT_PARTMAP]            = "partmap",
+    [PRINT_PARTUUID]           = "partuuid",
     [PRINT_ABSTRACTION]        = "abstraction",
     [PRINT_CRYPTODISK_UUID]    = "cryptodisk_uuid",
     [PRINT_HINT_STR]           = "hints_string",
@@ -181,6 +184,43 @@ probe_partmap (grub_disk_t disk, char delim)
     }
 }
 
+static void
+probe_partuuid (grub_disk_t disk, char delim)
+{
+  grub_partition_t p = disk->partition;
+
+  /*
+   * Nested partitions not supported for now.
+   * Non-nested partitions must have disk->partition->parent == NULL
+   */
+  if (p && p->parent == NULL)
+    {
+      disk->partition = p->parent;
+      if (strcmp(p->partmap->name, "msdos") == 0)
+	{
+	    /*
+	     * The partition GUID for MSDOS is the partition number (starting
+	     * with 1) prepended with the NT disk signature.
+	     */
+	    grub_uint32_t nt_disk_sig;
+
+	    if (grub_disk_read (disk, 0, GRUB_BOOT_MACHINE_WINDOWS_NT_MAGIC,
+				sizeof(nt_disk_sig), &nt_disk_sig) == 0)
+	      grub_printf ("%08x-%02x",
+			   grub_le_to_cpu32(nt_disk_sig), 1 + p->number);
+	}
+      else if (strcmp(p->partmap->name, "gpt") == 0)
+	{
+	  struct grub_gpt_partentry gptdata;
+
+	  if (grub_disk_read (disk, p->offset, p->index,
+			      sizeof(gptdata), &gptdata) == 0)
+	    print_gpt_guid(gptdata.guid);
+	}
+      disk->partition = p;
+    }
+}
+
 static void
 probe_cryptodisk_uuid (grub_disk_t disk, char delim)
 {
@@ -635,6 +675,12 @@ probe (const char *path, char **device_names, char delim)
 	/* Check if dev->disk itself is contained in a partmap.  */
 	probe_partmap (dev->disk, delim);
 
+      else if (print == PRINT_PARTUUID)
+	{
+	  probe_partuuid (dev->disk, delim);
+	  putchar (delim);
+	}
+
       else if (print == PRINT_MSDOS_PARTTYPE)
 	{
 	  if (dev->disk->partition
-- 
2.16.3

[GRUB PARTUUID PATCH V9 4/5] Update grub script template files

[Nicholas Vinson]

Update grub-mkconfig.in and 10_linux.in to support grub-probe's new
partuuid target.  Update grub.texi documentation.  The following table
shows how GRUB_DISABLE_LINUX_UUID, GRUB_DISABLE_LINUX_PARTUUID, and
initramfs detection interact:

Initramfs  GRUB_DISABLE_LINUX_PARTUUID  GRUB_DISABLE_LINUX_UUID  Linux Root
detected   Set                          Set                      ID Method

False      False                        False                    part UUID
False      False                        True                     part UUID
False      True                         False                    dev name
False      True                         True                     dev name
True       False                        False                    fs UUID
True       False                        True                     part UUID
True       True                         False                    fs UUID
True       True                         True                     dev name

Signed-off-by: Nicholas Vinson <nvinson234@gmail.com>
---
 docs/grub.texi          | 10 ++++++++++
 util/grub-mkconfig.in   |  3 +++
 util/grub.d/10_linux.in | 18 +++++++++++++++---
 3 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/docs/grub.texi b/docs/grub.texi
index 65b4bbeda..06f0afe45 100644
--- a/docs/grub.texi
+++ b/docs/grub.texi
@@ -1424,6 +1424,16 @@ the Linux kernel, using a @samp{root=UUID=...} kernel parameter.  This is
 usually more reliable, but in some cases it may not be appropriate.  To
 disable the use of UUIDs, set this option to @samp{true}.
 
+@item GRUB_DISABLE_LINUX_PARTUUID
+If @command{grub-mkconfig} cannot identify the root filesystem via its
+universally-unique indentifier (UUID), @command{grub-mkconfig} will use the UUID
+of the partition containing the filesystem to identify the root filesystem to
+the Linux kernel via a @samp{root=PARTUUID=...} kernel parameter.  This is not
+as reliable as using the filesystem UUID, but is more reliable than using the
+Linux device names.  When enabled, this option requires the Linux kernel version
+to be greater than 2.6.37 (3.10 for systems using the MSDOS partition scheme) or
+newer.  To disable the use of partition UUIDs, set this option to @samp{true}.
+
 @item GRUB_DISABLE_RECOVERY
 If this option is set to @samp{true}, disable the generation of recovery
 mode menu entries.
diff --git a/util/grub-mkconfig.in b/util/grub-mkconfig.in
index 35ef583b0..33332360e 100644
--- a/util/grub-mkconfig.in
+++ b/util/grub-mkconfig.in
@@ -134,6 +134,7 @@ fi
 # Device containing our userland.  Typically used for root= parameter.
 GRUB_DEVICE="`${grub_probe} --target=device /`"
 GRUB_DEVICE_UUID="`${grub_probe} --device ${GRUB_DEVICE} --target=fs_uuid 2> /dev/null`" || true
+GRUB_DEVICE_PARTUUID="`${grub_probe} --device ${GRUB_DEVICE} --target=partuuid 2> /dev/null`" || true
 
 # Device containing our /boot partition.  Usually the same as GRUB_DEVICE.
 GRUB_DEVICE_BOOT="`${grub_probe} --target=device /boot`"
@@ -188,6 +189,7 @@ if [ "x${GRUB_ACTUAL_DEFAULT}" = "xsaved" ] ; then GRUB_ACTUAL_DEFAULT="`"${grub
 # override them.
 export GRUB_DEVICE \
   GRUB_DEVICE_UUID \
+  GRUB_DEVICE_PARTUUID \
   GRUB_DEVICE_BOOT \
   GRUB_DEVICE_BOOT_UUID \
   GRUB_FS \
@@ -223,6 +225,7 @@ export GRUB_DEFAULT \
   GRUB_TERMINAL_OUTPUT \
   GRUB_SERIAL_COMMAND \
   GRUB_DISABLE_LINUX_UUID \
+  GRUB_DISABLE_LINUX_PARTUUID \
   GRUB_DISABLE_RECOVERY \
   GRUB_VIDEO_BACKEND \
   GRUB_GFXMODE \
diff --git a/util/grub.d/10_linux.in b/util/grub.d/10_linux.in
index faedf74e1..6916f1841 100644
--- a/util/grub.d/10_linux.in
+++ b/util/grub.d/10_linux.in
@@ -45,10 +45,16 @@ esac
 
 # btrfs may reside on multiple devices. We cannot pass them as value of root= parameter
 # and mounting btrfs requires user space scanning, so force UUID in this case.
-if [ "x${GRUB_DEVICE_UUID}" = "x" ] || [ "x${GRUB_DISABLE_LINUX_UUID}" = "xtrue" ] \
-    || ! test -e "/dev/disk/by-uuid/${GRUB_DEVICE_UUID}" \
+if ( [ "x${GRUB_DEVICE_UUID}" = "x" ] && [ "x${GRUB_DEVICE_PARTUUID}" = "x" ) \
+    || ( [ "x${GRUB_DISABLE_LINUX_UUID}" = "xtrue" ] \
+	&& [ "x${GRUB_DISABLE_LINUX_PARTUUID}" = "xtrue" ] ) \
+    || ( ! test -e "/dev/disk/by-uuid/${GRUB_DEVICE_UUID}" \
+	&& ! test -e "/dev/disk/by-partuuid/${GRUB_DEVICE_PARTUUID}" ) \
     || ( test -e "${GRUB_DEVICE}" && uses_abstraction "${GRUB_DEVICE}" lvm ); then
   LINUX_ROOT_DEVICE=${GRUB_DEVICE}
+elif [ "x${GRUB_DEVICE_UUID}" = "x" ] \
+    || [ "x${GRUB_DISABLE_LINUX_UUID}" = "xtrue" ]; then
+  LINUX_ROOT_DEVICE=PARTUUID=${GRUB_DEVICE_PARTUUID}
 else
   LINUX_ROOT_DEVICE=UUID=${GRUB_DEVICE_UUID}
 fi
@@ -242,7 +248,13 @@ while [ "x$list" != "x" ] ; do
   if test -z "${initramfs}" && test -z "${initrd_real}" ; then
     # "UUID=" and "ZFS=" magic is parsed by initrd or initramfs.  Since there's
     # no initrd or builtin initramfs, it can't work here.
-    linux_root_device_thisversion=${GRUB_DEVICE}
+    if [ "x${GRUB_DEVICE_PARTUUID}" = "x" ] \
+	|| [ "x${GRUB_DISABLE_LINUX_PARTUUID}" = "xtrue" ]; then
+
+	linux_root_device_thisversion=${GRUB_DEVICE}
+    else
+	linux_root_device_thisversion=PARTUUID=${GRUB_DEVICE_PARTUUID}
+    fi
   fi
 
   if [ "x$is_top_level" = xtrue ] && [ "x${GRUB_DISABLE_SUBMENU}" != xy ]; then
-- 
2.16.3

[GRUB PARTUUID PATCH V9 5/5] Default to disabling partition UUID support

[Nicholas Vinson]

The addition of GRUB_DISABLE_LINUX_PARTUUID and having it default to
'false' is a breaking change for systems that use GPT partition schemes
and Linux kernel version 2.6.37 or older and a breaking change for
systems that use MBR partition schemes and Linux kernel version 3.10 or
older.

This patch updates grub.d/10_linux so that GRUB_DISABLE_LINUX_PARTUUID
is set to 'true' if the system has not specified otherwise.

Signed-off-by: Nicholas Vinson <nvinson234@gmail.com>
---
 docs/grub.texi          | 5 +++--
 util/grub.d/10_linux.in | 4 ++++
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/docs/grub.texi b/docs/grub.texi
index 06f0afe45..3fcc42e74 100644
--- a/docs/grub.texi
+++ b/docs/grub.texi
@@ -1426,13 +1426,14 @@ disable the use of UUIDs, set this option to @samp{true}.
 
 @item GRUB_DISABLE_LINUX_PARTUUID
 If @command{grub-mkconfig} cannot identify the root filesystem via its
-universally-unique indentifier (UUID), @command{grub-mkconfig} will use the UUID
+universally-unique indentifier (UUID), @command{grub-mkconfig} can use the UUID
 of the partition containing the filesystem to identify the root filesystem to
 the Linux kernel via a @samp{root=PARTUUID=...} kernel parameter.  This is not
 as reliable as using the filesystem UUID, but is more reliable than using the
 Linux device names.  When enabled, this option requires the Linux kernel version
 to be greater than 2.6.37 (3.10 for systems using the MSDOS partition scheme) or
-newer.  To disable the use of partition UUIDs, set this option to @samp{true}.
+newer.  This option defaults to @samp{true}.  To enable the use of partition
+UUIDs, set this option to @samp{false}.
 
 @item GRUB_DISABLE_RECOVERY
 If this option is set to @samp{true}, disable the generation of recovery
diff --git a/util/grub.d/10_linux.in b/util/grub.d/10_linux.in
index 6916f1841..146499f22 100644
--- a/util/grub.d/10_linux.in
+++ b/util/grub.d/10_linux.in
@@ -43,6 +43,10 @@ case ${GRUB_DEVICE} in
   ;;
 esac
 
+# Default to disabling partition uuid support to maintian compatibility with
+# older kernels.
+GRUB_DISABLE_LINUX_PARTUUID=${GRUB_DISABLE_LINUX_PARTUUID-true}
+
 # btrfs may reside on multiple devices. We cannot pass them as value of root= parameter
 # and mounting btrfs requires user space scanning, so force UUID in this case.
 if ( [ "x${GRUB_DEVICE_UUID}" = "x" ] && [ "x${GRUB_DEVICE_PARTUUID}" = "x" ) \
-- 
2.16.3

Re: [GRUB PARTUUID PATCH V9 4/5] Update grub script template files

[Daniel Kiper]

On Sat, Apr 07, 2018 at 04:28:13PM -0700, Nicholas Vinson wrote:
> Update grub-mkconfig.in and 10_linux.in to support grub-probe's new
> partuuid target.  Update grub.texi documentation.  The following table
> shows how GRUB_DISABLE_LINUX_UUID, GRUB_DISABLE_LINUX_PARTUUID, and
> initramfs detection interact:
>
> Initramfs  GRUB_DISABLE_LINUX_PARTUUID  GRUB_DISABLE_LINUX_UUID  Linux Root
> detected   Set                          Set                      ID Method
>
> False      False                        False                    part UUID
> False      False                        True                     part UUID
> False      True                         False                    dev name
> False      True                         True                     dev name
> True       False                        False                    fs UUID
> True       False                        True                     part UUID
> True       True                         False                    fs UUID
> True       True                         True                     dev name

What will happen if GRUB_DISABLE_LINUX_PARTUUID and/or GRUB_DISABLE_LINUX_UUID
are not set? I think that you can avoid that by setting defaults. You do that
for GRUB_DISABLE_LINUX_PARTUUID in next patch but GRUB_DISABLE_LINUX_UUID
does not have any default.

> Signed-off-by: Nicholas Vinson <nvinson234@gmail.com>
> ---
>  docs/grub.texi          | 10 ++++++++++
>  util/grub-mkconfig.in   |  3 +++
>  util/grub.d/10_linux.in | 18 +++++++++++++++---

Could you update util/grub.d/20_linux_xen.in too?

>  3 files changed, 28 insertions(+), 3 deletions(-)
>
> diff --git a/docs/grub.texi b/docs/grub.texi
> index 65b4bbeda..06f0afe45 100644
> --- a/docs/grub.texi
> +++ b/docs/grub.texi
> @@ -1424,6 +1424,16 @@ the Linux kernel, using a @samp{root=UUID=...} kernel parameter.  This is
>  usually more reliable, but in some cases it may not be appropriate.  To
>  disable the use of UUIDs, set this option to @samp{true}.
>
> +@item GRUB_DISABLE_LINUX_PARTUUID
> +If @command{grub-mkconfig} cannot identify the root filesystem via its
> +universally-unique indentifier (UUID), @command{grub-mkconfig} will use the UUID
> +of the partition containing the filesystem to identify the root filesystem to
> +the Linux kernel via a @samp{root=PARTUUID=...} kernel parameter.  This is not
> +as reliable as using the filesystem UUID, but is more reliable than using the
> +Linux device names.  When enabled, this option requires the Linux kernel version

s/When enabled, /When enabled, GRUB_DISABLE_LINUX_PARTUUID set to false, /
or something like that. Otherwise it is unclear.

Daniel

Re: [GRUB PARTUUID PATCH V9 0/5] Add PARTUUID detection support

[Daniel Kiper]

On Sat, Apr 07, 2018 at 04:28:09PM -0700, Nicholas Vinson wrote:
> Changes from Patch v8:
>     - Renamed GRUB_ENABLE_LINUX_PARTUUID to GRUB_DISABLE_LINUX_PARTUUID
>     - Updated the 10_linux logic so GRUB_ENABLE_LINUX_PARTUUID and
>         GRUB_ENABLE_LINUX_UUID would behave more independently.
>     - Documented interactions of GRUB_DISABLE_LINUX_UUID,
>         GRUB_DISABLE_LINUX_PARTUUID, and initramfs detection in commit
>         message.
>     - Added optional patch that sets GRUB_DISABLE_LINUX_PARTUUID to true
>         by default.
>     - Fixed merge conflicts between this patchset and upstream's master
>         branch.

Could you provide summary for current version not only for previous ones?

Anyway, I think that most of the patches are now in good shape.
I am looking forward for next version.

Thank you for doing the work.

Daniel

Re: [GRUB PARTUUID PATCH V9 4/5] Update grub script template files

[Nick Vinson]

On 04/10/2018 01:52 PM, Daniel Kiper wrote:
> On Sat, Apr 07, 2018 at 04:28:13PM -0700, Nicholas Vinson wrote:
>> Update grub-mkconfig.in and 10_linux.in to support grub-probe's new
>> partuuid target.  Update grub.texi documentation.  The following table
>> shows how GRUB_DISABLE_LINUX_UUID, GRUB_DISABLE_LINUX_PARTUUID, and
>> initramfs detection interact:
>>
>> Initramfs  GRUB_DISABLE_LINUX_PARTUUID  GRUB_DISABLE_LINUX_UUID  Linux Root
>> detected   Set                          Set                      ID Method
>>
>> False      False                        False                    part UUID
>> False      False                        True                     part UUID
>> False      True                         False                    dev name
>> False      True                         True                     dev name
>> True       False                        False                    fs UUID
>> True       False                        True                     part UUID
>> True       True                         False                    fs UUID
>> True       True                         True                     dev name
> 
> What will happen if GRUB_DISABLE_LINUX_PARTUUID and/or GRUB_DISABLE_LINUX_UUID
> are not set? I think that you can avoid that by setting defaults. You do that
> for GRUB_DISABLE_LINUX_PARTUUID in next patch but GRUB_DISABLE_LINUX_UUID
> does not have any default.
> 

If they're not set, then that's the same as them being set to 'False'.
I should have worded my table above a bit differently and used Yes/No
instead of True/False as that is really what it is trying to convey.
I.E. If there is no initramfs detected and GRUB_DISABLE_LINUX_PARTUUID
is not set and GRUB_DISABLE_LINUX_UUID is not set, set the kernel root
variable to "root=PARTUUID=...".

In the scripts, the only value explicitly checked for is 'true'.
Anything else (including unset) is considered false.

>> Signed-off-by: Nicholas Vinson <nvinson234@gmail.com>
>> ---
>>  docs/grub.texi          | 10 ++++++++++
>>  util/grub-mkconfig.in   |  3 +++
>>  util/grub.d/10_linux.in | 18 +++++++++++++++---
> 
> Could you update util/grub.d/20_linux_xen.in too?

Let me see what I can do.  I have never worked with xen before, so I do
not know much about it.

> 
>>  3 files changed, 28 insertions(+), 3 deletions(-)
>>
>> diff --git a/docs/grub.texi b/docs/grub.texi
>> index 65b4bbeda..06f0afe45 100644
>> --- a/docs/grub.texi
>> +++ b/docs/grub.texi
>> @@ -1424,6 +1424,16 @@ the Linux kernel, using a @samp{root=UUID=...} kernel parameter.  This is
>>  usually more reliable, but in some cases it may not be appropriate.  To
>>  disable the use of UUIDs, set this option to @samp{true}.
>>
>> +@item GRUB_DISABLE_LINUX_PARTUUID
>> +If @command{grub-mkconfig} cannot identify the root filesystem via its
>> +universally-unique indentifier (UUID), @command{grub-mkconfig} will use the UUID
>> +of the partition containing the filesystem to identify the root filesystem to
>> +the Linux kernel via a @samp{root=PARTUUID=...} kernel parameter.  This is not
>> +as reliable as using the filesystem UUID, but is more reliable than using the
>> +Linux device names.  When enabled, this option requires the Linux kernel version
> 
> s/When enabled, /When enabled, GRUB_DISABLE_LINUX_PARTUUID set to false, /
> or something like that. Otherwise it is unclear.

I'll clean up the wording in the next release.

Thanks,
Nicholas Vinson

> 
> Daniel
> 

Re: [GRUB PARTUUID PATCH V9 0/5] Add PARTUUID detection support

[Nick Vinson]

On 04/10/2018 01:56 PM, Daniel Kiper wrote:
> On Sat, Apr 07, 2018 at 04:28:09PM -0700, Nicholas Vinson wrote:
>> Changes from Patch v8:
>>     - Renamed GRUB_ENABLE_LINUX_PARTUUID to GRUB_DISABLE_LINUX_PARTUUID
>>     - Updated the 10_linux logic so GRUB_ENABLE_LINUX_PARTUUID and
>>         GRUB_ENABLE_LINUX_UUID would behave more independently.
>>     - Documented interactions of GRUB_DISABLE_LINUX_UUID,
>>         GRUB_DISABLE_LINUX_PARTUUID, and initramfs detection in commit
>>         message.
>>     - Added optional patch that sets GRUB_DISABLE_LINUX_PARTUUID to true
>>         by default.
>>     - Fixed merge conflicts between this patchset and upstream's master
>>         branch.
> 
> Could you provide summary for current version not only for previous ones?

I can add a summary section before the change log.  I'll make sure to do
so with the next patchset.

Thanks,
Nicholas Vinson

> 
> Anyway, I think that most of the patches are now in good shape.
> I am looking forward for next version.
> 
> Thank you for doing the work.
> 
> Daniel
> 

Re: [GRUB PARTUUID PATCH V9 4/5] Update grub script template files

[Daniel Kiper]

On Tue, Apr 10, 2018 at 08:00:04PM -0700, Nick Vinson wrote:
> On 04/10/2018 01:52 PM, Daniel Kiper wrote:
> > On Sat, Apr 07, 2018 at 04:28:13PM -0700, Nicholas Vinson wrote:
> >> Update grub-mkconfig.in and 10_linux.in to support grub-probe's new
> >> partuuid target.  Update grub.texi documentation.  The following table
> >> shows how GRUB_DISABLE_LINUX_UUID, GRUB_DISABLE_LINUX_PARTUUID, and
> >> initramfs detection interact:
> >>
> >> Initramfs  GRUB_DISABLE_LINUX_PARTUUID  GRUB_DISABLE_LINUX_UUID  Linux Root
> >> detected   Set                          Set                      ID Method
> >>
> >> False      False                        False                    part UUID
> >> False      False                        True                     part UUID
> >> False      True                         False                    dev name
> >> False      True                         True                     dev name
> >> True       False                        False                    fs UUID
> >> True       False                        True                     part UUID
> >> True       True                         False                    fs UUID
> >> True       True                         True                     dev name
> >
> > What will happen if GRUB_DISABLE_LINUX_PARTUUID and/or GRUB_DISABLE_LINUX_UUID
> > are not set? I think that you can avoid that by setting defaults. You do that
> > for GRUB_DISABLE_LINUX_PARTUUID in next patch but GRUB_DISABLE_LINUX_UUID
> > does not have any default.
> >
>
> If they're not set, then that's the same as them being set to 'False'.
> I should have worded my table above a bit differently and used Yes/No
> instead of True/False as that is really what it is trying to convey.

IMO it will be more confusing. I think that I would use lowercase
false/true as it is used in the script and below the table I would
add a note that <VARIABLE_UNSET> == false or something like that.

> I.E. If there is no initramfs detected and GRUB_DISABLE_LINUX_PARTUUID
> is not set and GRUB_DISABLE_LINUX_UUID is not set, set the kernel root
> variable to "root=PARTUUID=...".
>
> In the scripts, the only value explicitly checked for is 'true'.
> Anything else (including unset) is considered false.

OK, perfect!

> >> Signed-off-by: Nicholas Vinson <nvinson234@gmail.com>
> >> ---
> >>  docs/grub.texi          | 10 ++++++++++
> >>  util/grub-mkconfig.in   |  3 +++
> >>  util/grub.d/10_linux.in | 18 +++++++++++++++---
> >
> > Could you update util/grub.d/20_linux_xen.in too?
>
> Let me see what I can do.  I have never worked with xen before, so I do
> not know much about it.

This should be easy. However, if you are not sure please drop me a line.

> >>  3 files changed, 28 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/docs/grub.texi b/docs/grub.texi
> >> index 65b4bbeda..06f0afe45 100644
> >> --- a/docs/grub.texi
> >> +++ b/docs/grub.texi
> >> @@ -1424,6 +1424,16 @@ the Linux kernel, using a @samp{root=UUID=...} kernel parameter.  This is
> >>  usually more reliable, but in some cases it may not be appropriate.  To
> >>  disable the use of UUIDs, set this option to @samp{true}.
> >>
> >> +@item GRUB_DISABLE_LINUX_PARTUUID
> >> +If @command{grub-mkconfig} cannot identify the root filesystem via its
> >> +universally-unique indentifier (UUID), @command{grub-mkconfig} will use the UUID
> >> +of the partition containing the filesystem to identify the root filesystem to
> >> +the Linux kernel via a @samp{root=PARTUUID=...} kernel parameter.  This is not
> >> +as reliable as using the filesystem UUID, but is more reliable than using the
> >> +Linux device names.  When enabled, this option requires the Linux kernel version
> >
> > s/When enabled, /When enabled, GRUB_DISABLE_LINUX_PARTUUID set to false, /
> > or something like that. Otherwise it is unclear.
>
> I'll clean up the wording in the next release.

Thanks a lot!

By the way, should not we merge patches 4 and 5 into one patch?

Daniel

Re: [GRUB PARTUUID PATCH V9 4/5] Update grub script template files

[Nick Vinson]

On 04/11/2018 01:31 AM, Daniel Kiper wrote:
> On Tue, Apr 10, 2018 at 08:00:04PM -0700, Nick Vinson wrote:
>> On 04/10/2018 01:52 PM, Daniel Kiper wrote:
>>> On Sat, Apr 07, 2018 at 04:28:13PM -0700, Nicholas Vinson wrote:
>>>> Update grub-mkconfig.in and 10_linux.in to support grub-probe's new
>>>> partuuid target.  Update grub.texi documentation.  The following table
>>>> shows how GRUB_DISABLE_LINUX_UUID, GRUB_DISABLE_LINUX_PARTUUID, and
>>>> initramfs detection interact:
>>>>
>>>> Initramfs  GRUB_DISABLE_LINUX_PARTUUID  GRUB_DISABLE_LINUX_UUID  Linux Root
>>>> detected   Set                          Set                      ID Method
>>>>
>>>> False      False                        False                    part UUID
>>>> False      False                        True                     part UUID
>>>> False      True                         False                    dev name
>>>> False      True                         True                     dev name
>>>> True       False                        False                    fs UUID
>>>> True       False                        True                     part UUID
>>>> True       True                         False                    fs UUID
>>>> True       True                         True                     dev name
>>>
>>> What will happen if GRUB_DISABLE_LINUX_PARTUUID and/or GRUB_DISABLE_LINUX_UUID
>>> are not set? I think that you can avoid that by setting defaults. You do that
>>> for GRUB_DISABLE_LINUX_PARTUUID in next patch but GRUB_DISABLE_LINUX_UUID
>>> does not have any default.
>>>
>>
>> If they're not set, then that's the same as them being set to 'False'.
>> I should have worded my table above a bit differently and used Yes/No
>> instead of True/False as that is really what it is trying to convey.
> 
> IMO it will be more confusing. I think that I would use lowercase
> false/true as it is used in the script and below the table I would
> add a note that <VARIABLE_UNSET> == false or something like that.

Ack.  I will update the commit comment.

> 
>> I.E. If there is no initramfs detected and GRUB_DISABLE_LINUX_PARTUUID
>> is not set and GRUB_DISABLE_LINUX_UUID is not set, set the kernel root
>> variable to "root=PARTUUID=...".
>>
>> In the scripts, the only value explicitly checked for is 'true'.
>> Anything else (including unset) is considered false.
> 
> OK, perfect!
> 
>>>> Signed-off-by: Nicholas Vinson <nvinson234@gmail.com>
>>>> ---
>>>>  docs/grub.texi          | 10 ++++++++++
>>>>  util/grub-mkconfig.in   |  3 +++
>>>>  util/grub.d/10_linux.in | 18 +++++++++++++++---
>>>
>>> Could you update util/grub.d/20_linux_xen.in too?
>>
>> Let me see what I can do.  I have never worked with xen before, so I do
>> not know much about it.
> 
> This should be easy. However, if you are not sure please drop me a line.
> 
>>>>  3 files changed, 28 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/docs/grub.texi b/docs/grub.texi
>>>> index 65b4bbeda..06f0afe45 100644
>>>> --- a/docs/grub.texi
>>>> +++ b/docs/grub.texi
>>>> @@ -1424,6 +1424,16 @@ the Linux kernel, using a @samp{root=UUID=...} kernel parameter.  This is
>>>>  usually more reliable, but in some cases it may not be appropriate.  To
>>>>  disable the use of UUIDs, set this option to @samp{true}.
>>>>
>>>> +@item GRUB_DISABLE_LINUX_PARTUUID
>>>> +If @command{grub-mkconfig} cannot identify the root filesystem via its
>>>> +universally-unique indentifier (UUID), @command{grub-mkconfig} will use the UUID
>>>> +of the partition containing the filesystem to identify the root filesystem to
>>>> +the Linux kernel via a @samp{root=PARTUUID=...} kernel parameter.  This is not
>>>> +as reliable as using the filesystem UUID, but is more reliable than using the
>>>> +Linux device names.  When enabled, this option requires the Linux kernel version
>>>
>>> s/When enabled, /When enabled, GRUB_DISABLE_LINUX_PARTUUID set to false, /
>>> or something like that. Otherwise it is unclear.
>>
>> I'll clean up the wording in the next release.
> 
> Thanks a lot!
> 
> By the way, should not we merge patches 4 and 5 into one patch?

We should.

I split it so that if #5 proved controversial for some reason and
received a lot of resistance, it could be dropped and the remainder of
the patchset could be merged.

There does not seem to be any issue with setting
GRUB_DISABLE_LINUX_PARTUUID to a default value, so there is no reason to
keep it separate.

Thanks,
Nicholas Vinson

> 
> Daniel
> 

Re: [GRUB PARTUUID PATCH V9 4/5] Update grub script template files

[Daniel Kiper]

On Wed, Apr 11, 2018 at 07:45:01AM -0700, Nick Vinson wrote:
> On 04/11/2018 01:31 AM, Daniel Kiper wrote:
> > On Tue, Apr 10, 2018 at 08:00:04PM -0700, Nick Vinson wrote:
> >> On 04/10/2018 01:52 PM, Daniel Kiper wrote:
> >>> On Sat, Apr 07, 2018 at 04:28:13PM -0700, Nicholas Vinson wrote:
> >>>> Update grub-mkconfig.in and 10_linux.in to support grub-probe's new
> >>>> partuuid target.  Update grub.texi documentation.  The following table
> >>>> shows how GRUB_DISABLE_LINUX_UUID, GRUB_DISABLE_LINUX_PARTUUID, and
> >>>> initramfs detection interact:
> >>>>
> >>>> Initramfs  GRUB_DISABLE_LINUX_PARTUUID  GRUB_DISABLE_LINUX_UUID  Linux Root
> >>>> detected   Set                          Set                      ID Method
> >>>>
> >>>> False      False                        False                    part UUID
> >>>> False      False                        True                     part UUID
> >>>> False      True                         False                    dev name
> >>>> False      True                         True                     dev name
> >>>> True       False                        False                    fs UUID
> >>>> True       False                        True                     part UUID
> >>>> True       True                         False                    fs UUID
> >>>> True       True                         True                     dev name
> >>>
> >>> What will happen if GRUB_DISABLE_LINUX_PARTUUID and/or GRUB_DISABLE_LINUX_UUID
> >>> are not set? I think that you can avoid that by setting defaults. You do that
> >>> for GRUB_DISABLE_LINUX_PARTUUID in next patch but GRUB_DISABLE_LINUX_UUID
> >>> does not have any default.
> >>>
> >>
> >> If they're not set, then that's the same as them being set to 'False'.
> >> I should have worded my table above a bit differently and used Yes/No
> >> instead of True/False as that is really what it is trying to convey.
> >
> > IMO it will be more confusing. I think that I would use lowercase
> > false/true as it is used in the script and below the table I would
> > add a note that <VARIABLE_UNSET> == false or something like that.
>
> Ack.  I will update the commit comment.

Thanks. May I ask you to put similar table into docs/grub.texi?

Daniel

Re: [GRUB PARTUUID PATCH V9 4/5] Update grub script template files

[Nick Vinson]

On 04/16/2018 04:47 AM, Daniel Kiper wrote:
> On Wed, Apr 11, 2018 at 07:45:01AM -0700, Nick Vinson wrote:
>> On 04/11/2018 01:31 AM, Daniel Kiper wrote:
>>> On Tue, Apr 10, 2018 at 08:00:04PM -0700, Nick Vinson wrote:
>>>> On 04/10/2018 01:52 PM, Daniel Kiper wrote:
>>>>> On Sat, Apr 07, 2018 at 04:28:13PM -0700, Nicholas Vinson wrote:
>>>>>> Update grub-mkconfig.in and 10_linux.in to support grub-probe's new
>>>>>> partuuid target.  Update grub.texi documentation.  The following table
>>>>>> shows how GRUB_DISABLE_LINUX_UUID, GRUB_DISABLE_LINUX_PARTUUID, and
>>>>>> initramfs detection interact:
>>>>>>
>>>>>> Initramfs  GRUB_DISABLE_LINUX_PARTUUID  GRUB_DISABLE_LINUX_UUID  Linux Root
>>>>>> detected   Set                          Set                      ID Method
>>>>>>
>>>>>> False      False                        False                    part UUID
>>>>>> False      False                        True                     part UUID
>>>>>> False      True                         False                    dev name
>>>>>> False      True                         True                     dev name
>>>>>> True       False                        False                    fs UUID
>>>>>> True       False                        True                     part UUID
>>>>>> True       True                         False                    fs UUID
>>>>>> True       True                         True                     dev name
>>>>>
>>>>> What will happen if GRUB_DISABLE_LINUX_PARTUUID and/or GRUB_DISABLE_LINUX_UUID
>>>>> are not set? I think that you can avoid that by setting defaults. You do that
>>>>> for GRUB_DISABLE_LINUX_PARTUUID in next patch but GRUB_DISABLE_LINUX_UUID
>>>>> does not have any default.
>>>>>
>>>>
>>>> If they're not set, then that's the same as them being set to 'False'.
>>>> I should have worded my table above a bit differently and used Yes/No
>>>> instead of True/False as that is really what it is trying to convey.
>>>
>>> IMO it will be more confusing. I think that I would use lowercase
>>> false/true as it is used in the script and below the table I would
>>> add a note that <VARIABLE_UNSET> == false or something like that.
>>
>> Ack.  I will update the commit comment.
> 
> Thanks. May I ask you to put similar table into docs/grub.texi?

Done.

Thanks,
Nicholas Vinson
> 
> Daniel
>