During device probing GParted always explicitly calls ped_device_sync()
to flush the caches for coherency between the whole disk device and the
partition devices [1]. As the GParted_Core::flush_device() comment
explains, since v3.1.61-gfb99ba5 [2], libparted flushes the devices
every time a device is opened [3], so also explicitly doing this in
GParted is unnecessary. Therefore stop explicitly flushing the devices
in GParted, except when using libparted 3.1 and older which doesn't do
it itself.
The ped_device_open() and ped_device_close() needed for the
ped_device_sync() is also a trigger of device changes and udev rule
execution performing LVM Volume Group activation, because libparted
opens the device read-write [4]. This is another reason to remove it
when possible. However even when eliminated it does not solve the issue
of LVM VG activation because other triggers remain. Do want this change
first though so that the sequence of libparted calls isn't changed
immediately after documenting them and fixing the issue and so that
there is no doubt that this change doesn't fix the issue.
Removing this extra device flush saves a little bit of time, depending
on the speed of the drive and number of partitions. Sample savings on
my desktop:
Drive type and partitions Saving (seconds)
----------------------------- ----------------
fast SSD with 3 partitions 0.05
slow SSD with 12 partitions 0.27
HDD with 1 partition 0.05
VHDD in VM with 1 partition 0.14
VHDD in VM with 10 partitions 0.58
Also the settle_device() call in flush_device() needs to be kept to wait
for device changes and udev rule execution whether it is GParted
explicitly flushing the device or just libparted automatically doing it
for cache coherency in ped_device_get(). This is because:
1. Libparted <= 3.2 opens the whole disk device read-write for
ped_device_get() [5]. (This was changed with parted
v3.2.26-g44d5ae0 [6] to open the whole disk device read-only).
2. Libparted up to and including the latest 3.6 release still opens all
partition devices read-write for ped_device_get() [7].
3. A whole disk device FAT32 file system looks enough like a partition
table that both the Linux kernel and libparted think it is
partitioned:
# mkfs.fat -F32 /dev/sdb
mkfs.fat 4.2 (2021-01-31)
# grep sdb /proc/partitions
8 16 8388608 sdb
8 17 8388607 sdb1
# parted /dev/sdb print
Model: ATA VBOX HARDDISK (scsi)
Disk /dev/sdb: 8590MB
Sector size (logical/physical): 512B/512B
Partition Table: loop
Disk Flags:
Number Start End Size File system Flags
1 0s 16777215s 16777216s fat32
So the ped_device_get() call on a whole disk device FAT32 file system
still triggers device change and udev rule execution which needs to be
waited for, as this is exactly the case fixed previously by commit:
1382e0b828
Wait for udev change on /dev/DISK when querying whole device FS (!46)
[1] 3bea067596
Flush devices when scanning to prevent reading stale signatures (#723842)
[2] Revert "linux-commit: do not unnecessarily open partition device nodes"
http://git.savannah.gnu.org/cgit/parted.git/commit/?id=fb99ba5ebd0dc34204fc9f1014131d5d494805bc
[3] parted libparted/arch/linux.c:_device_open()
https://git.savannah.gnu.org/cgit/parted.git/tree/libparted/arch/linux.c?h=v3.6#n1752
1709 static int
1710 linux_open (PedDevice* dev)
1711 {
1712 return _device_open (dev, RW_MODE);
1713 }
1714
1715 static int
1716 _device_open (PedDevice* dev, int flags)
...
1752 _flush_cache (dev);
[4] parted libparted/device.c:ped_device_open() v3.6
https://git.savannah.gnu.org/cgit/parted.git/tree/libparted/device.c?h=v3.6#n226
parted libparted/arch/linux.c v3.6
https://git.savannah.gnu.org/cgit/parted.git/tree/libparted/arch/linux.c?h=v3.6
ped_device_open(...)
ped_architecture->dev_ops->open(...) = linux_open(...)
_device_open(..., RW_MODE)
open(..., O_RDWR)
[5] parted libparted/device.c:ped_device_get() v3.2
https://git.savannah.gnu.org/cgit/parted.git/tree/libparted/device.c?h=v3.2#n149
parted libparted/arch/linux.c v3.2
https://git.savannah.gnu.org/cgit/parted.git/tree/libparted/arch/linux.c?h=v3.2
ped_device_get(...)
ped_architecture->dev_ops->_new(...) = linux_new()
init_ide(...)
init_scsi(...)
init_generic(...)
ped_device_open(...)
ped_architecture->dev_ops->open(...) = linux_open(...)
open(..., O_RDWR)
[6] libparted: Use read only when probing devices on linux (#1245144)
http://git.savannah.gnu.org/cgit/parted.git/commit/?id=44d5ae0115c4ecfe3158748309e9912c5aede92d
[7] parted libparted/arch/linux.v v3.6
http://git.savannah.gnu.org/cgit/parted.git/tree/libparted/arch/linux.c?h=v3.6#n1660
1660 static void
1661 _flush_cache (PedDevice* dev)
...
1673 for (i = 1; i < lpn; i++) {
...
1680 if (!_partition_is_mounted_by_path(name)) {
1681 fd = open (name, WR_MODE, 0);
Closes#259 - Trying to deactivate LVM PV fails
[Only the options being used by GParted are quoted here from the help
output. More options are available.]
$ bcachefs fsck --help
bcachefs fsck - filesystem check and repair
Usage: bcachefs fsck [OPTION]... <device>
Options:
-y Assume "yes" to all questions
-f Force checking even if filesystem is marked clean
-v Be verbose
Closes!123 - Add support for bcachefs, single device file systems only
Add reading of the bcachefs file system label, not the per device label,
and the external UUID. These match what blkid reports.
Example without a label:
# bcachefs format /dev/sdb1
# bcachefs show-super /dev/sdb1 | egrep -i 'Label:|UUID:|Device:'
External UUID: 3316bc9a-d129-42b6-a80e-9649874bca73
Internal UUID: 656eebe5-10a9-4f12-94c8-aab2fdc54732
Label:
Device: 0
Label: (none)
UUID: cd436a8d-82eb-4993-a317-b39ea0d6bd2e
# blkid /dev/sdb1
/dev/sdb1: UUID="3316bc9a-d129-42b6-a80e-9649874bca73" BLOCK_SIZE="512" UUID_SUB="cd436a8d-82eb-4993-a317-b39ea0d6bd2e" TYPE="bcachefs" PARTUUID="7962e584-34c9-4088-8a00-a651af517089"
Example with a label:
# bcachefs format --force -L 'test label' /dev/sdb1
# bcachefs show-super /dev/sdb1 | egrep -i 'Label:|UUID:|Device:'
External UUID: 3d7bdabe-2616-4545-affc-1aba0f8fb4a7
Internal UUID: 9cc95d3e-7991-4f78-9dd0-850cb9749e34
Label: test label
Device: 0
Label: (none)
UUID: 784d1bd0-5769-4fbb-ad32-07894d381bba
# blkid /dev/sdb1
/dev/sdb1: UUID="3d7bdabe-2616-4545-affc-1aba0f8fb4a7" LABEL="test label" BLOCK_SIZE="512" UUID_SUB="784d1bd0-5769-4fbb-ad32-07894d381bba" TYPE="bcachefs" PARTUUID="7962e584-34c9-4088-8a00-a651af517089"
Closes!123 - Add support for bcachefs, single device file systems only
Currently bcachefs-tools only provides a method to report the file
system usage while it is mounted. We won't make GParted mount a
bcachefs to read it's usage as we want to keep GParted's scanning as a
read-only activity. Therefore GParted can't report the usage of an
unmounted bcachefs.
# bcachefs format /dev/sdb1
# bcachefs fs usage /dev/sdb1
error opening /dev/sdb1: not a bcachefs filesystem
# echo $?
1
# bcachefs fs usage --help
bcachefs fs usage - display detailed filesystem usage
Usage: bcachefs fs usage [OPTION]... <mountpoint>
...
# mount /dev/sdb1 /mnt/1
# bcachefs fs usage /mnt/1
Filesystem: a61a8302-9a79-4c24-a9e6-486e7fcc78f5
Size: 987842560
Used: 12713984
Online reserved: 0
Data type Required/total Durability Devices
btree: 1/1 1 [sdb1] 1048576
(no label) (device 0): sdb1 rw
data buckets fragmented
free: 1061027840 8095
sb: 3149824 25 126976
journal: 8388608 64
btree: 1048576 8
user: 0 0
cached: 0 0
parity: 0 0
stripe: 0 0
need_gc_gens: 0 0
need_discard: 0 0
capacity: 1073741824 8192
# echo $?
0
Closes!123 - Add support for bcachefs, single device file systems only
Set the minimum file system size to 16 MiB as creating a bcachefs that
size succeeds:
$ truncate -s $((16*1024*1024)) /tmp/disk.img
$ bcachefs format /tmp/disk.img
...
initializing new filesystem
going read-write
initializing freespace
$ echo $?
0
Where as creating a smaller file system fails for most sizes below that:
$ rm /tmp/disk.img
$ truncate -s $((15*1024*1024)) /tmp/disk.img
$ bcachefs format /tmp/disk.img
...
mounting version 1.6: btree_subvolume_children
initializing new filesystem
going read-write
bch2_trans_mark_dev_sb(): error ENOSPC_disk_reservation
bch2_fs_initialize(): error marking superblocks ENOSPC_disk_reservation
bch2_fs_initialize(): error ENOSPC_disk_reservation
bch2_fs_start(): error starting filesystem ENOSPC_disk_reservation
error opening /tmp/disk.img: ENOSPC_disk_reservation
$ echo $?
1
Closes!123 - Add support for bcachefs, single device file systems only
To make it clear they are all member variables. Add missing
m_fs_block_size POD (Plain Old Data) member to the constructor
initialisation list so that it's never uninitialised. And stop
unnecessarily assigning constant false to m_force_auto_64bit in
get_filesystem_support() because it is already initialised to false by
the constructor.
To make it clear it is a member variable and continue the theme that
variables of type FSType have fstype in their name [1][2].
[1] 58fb230fb0
Also rename FS.filesystem member to fstype (!52)
[2] b0f92be638
Rename Partition.filesystem member to fstype (!52)
To make it clear m_old_mkudffs is a member variable.
To make it clear they are xfs member variables. Also add a default
constructor with an initialisation list so POD (Plain Old Data) member
variable m_src_used is never left uninitialised, as it was previously
only assigned in xfs::copy(). m_dest_mount_point is a Glib::ustring
object which is default constructed to a zero length string so doesn't
need adding to the xfs default constructor initialisation list.
* C++ FAQ / Should my constructors use "initialization lists" or
"assignment"?
https://isocpp.org/wiki/faq/ctors#init-lists
"Initialization lists. In fact, constructors should initialize as
a rule all member objects in the initialization list."
To make it clear m_old_mkudffs is a member variable. Also stop
unnecessarily assigning constant false to m_old_mkudffs in
get_filesystem_support() because it is already initialised to false by
the default constructor.
gpart scans a drive trying to guess the location of partitions when an
MBR partition table is lost [1]. However the tool is unmaintained,
takes hours or days of 100% CPU time to scan a drive and provides no
progress indication [2][3][4]. We keep recommending killing the gpart
process and using TestDisk [5] instead.
Therefore remove Device > Attempt Data Rescue and the use of gpart from
GParted.
[1] Gpart
https://github.com/baruch/gpart
[2] Have you had a good or bad experience with Dev->Attempt Data Rescue?
http://gparted-forum.surf4.info/viewtopic.php?id=17992
No good, only bad experiences using gpart were reported.
[3] Gparted does not say anything
http://gparted-forum.surf4.info/viewtopic.php?id=17749
Forum user reported waiting 48 hours with no progress indication.
We recommended using TestDisk.
[4] How cancel Data Rescue process?
http://gparted-forum.surf4.info/viewtopic.php?id=18143
Forum user reported it will take 3 days to scan their external 480GB
drive. We recommended using TestDisk instead.
[5] TestDisk, Data Recovery
https://www.cgsecurity.org/wiki/TestDiskCloses!118 - Remove Attempt Data Rescue and use of gpart
In C++11, nullptr [1] is the strongly typed value to use instead of the
macro NULL [2]. Use everywhere [3][4].
[1] nullptr, the pointer literal (since C++11)
https://en.cppreference.com/w/cpp/language/nullptr
[2] NULL
https://en.cppreference.com/w/cpp/types/NULL
[3] Bjarne Stroustrup's C++ Style and Technique FAQ, Should I use NULL
or 0?
https://www.stroustrup.com/bs_faq2.html#null
"In C++, the definition of NULL is 0, so there is only an
aesthetic difference. I prefer to avoid macros, so I use 0.
Another problem with NULL is that people sometimes mistakenly
believe that it is different from 0 and/or not an integer. In
pre-standard code, NULL was/is sometimes defined to something
unsuitable and therefore had/has to be avoided. That's less
common these days.
If you have to name the null pointer, call it nullptr; that's
what it's called in C++11. Then, "nullptr" will be a keyword.
"
[4] What is nullptr in C++? Advantages, Use Cases & Examples
https://favtutor.com/blogs/nullptr-cpp
"Advantages of nullptr
...
Compatible: Null pointers are compatible with null pointer
constants in the C style (such as NULL and 0). This implies
that old C code that uses these constants and null pointers can
communicate with each other in C++.
"
Closes!117 - Require C++11 compilation
get_text() only performs const access on the ProgressBar object so
return the member string by constant reference.
Previously done for other string returning getters, even though the
value is assigned to a variable and doesn't save anything:
1f6e81295b
Return constant reference from OperationDetail::get_description() (!94)
The only use of the reference returned from
OperationDetail::get_progressbar() is to call const methods
ProgressBar::running(), ::get_fraction() and ::get_text(). Therefore
make OperationDetail::get_progressbar() return a const reference.
As described in the previous commit "Clear progress bar text when
starting the bar (#230)" progress bar data is either reporting bytes
copied or fraction complete. The bytes copied case gets in progress
text like this:
544.00 MiB of 1.00 GiB copied (00:00:11 remaining)
But the fraction complete gets no text.
Now also generate time remaining text for progress bars only reporting
fraction complete. As with the bytes copied text only add the time
remaining estimate after 5 seconds have passed. Looks like:
(00:01:59 remaining)
This is most useful for NTFS partition copy and resize operations which
can take a while depending on the amount of data involved.
STAT_FORMATTED is only used inside snap_to_mebibyte() to suppress
enforcement that partition boundaries must not overlay the MBR or EBRs
when merely formatting existing partitions. However since commit [1],
snap_to_mebibyte() is only called inside the dialogs composing Create
New, Copy / Paste into New and Resize / Move operations and never when
composing a Format operation or any other operation which doesn't change
partition boundaries. Therefore remove STAT_FORMATTED.
[1] 7c94b7d920
Snap partition boundaries before dialogs update FS usage (#48)
Now Win_GParted::m_display_device.partitions is an identical copy of
Win_GParted::display_partitions with the same lifetime. That's wasteful
and pointless. Therefore remove the later and use the former in it's
place.
Closes#227 - Unable to allocate 1 MiB between partitions when moving to
the right
The Create New and Paste dialogs also create partitions and have to
honour currently composed partitions while doing so. Therefore they
must have a Device object containing the currently composed partition
layout for passing into snap_to_alignment() and below. So copy the
current Device object when refreshing the visual at the same time
visual_partitions is generated and use in all 3 dialogs which compose
new partitions.
Note that Create New and Paste aren't subject to the same bug as Resize/
Move was because the code in snap_to_mebibyte() [1] checked the
partition object being composed has status STAT_REAL. This is true for
partition objects created by the Resize/Move dialog, but not true for
the Create New and Paste dialogs which set status to STAT_NEW and
STAT_COPY respectively instead.
[1] Dialog_Base_Partition::snap_to_mebibyte() lines 418 to 438
https://gitlab.gnome.org/GNOME/gparted/-/blob/GPARTED_1_5_0/src/Dialog_Base_Partition.cc#L418Closes#227 - Unable to allocate 1 MiB between partitions when moving to
the right
Using Automake variable EXTRA_DIST [1] to list the GParted header files
seems overly general. Instead use noinst_HEADERS [2] as it better
describes GParted header files. Header files which need to be
distributed in the archive, but not part of an installed library so not
to be installed below /usr/include.
[1] GNU Automake manual, 14.1 Basics of Distribution
https://www.gnu.org/software/automake/manual/html_node/Basics-of-Distribution.html
"..., it is still common to have files to be distributed which
are not found by the automatic rules. You should listed these
files in the EXTRA_DIST variable. You can mention files in
subdirectories in EXTRA_DIST.
"
[2] GNU Automake manual, 9.2 Header files
https://www.gnu.org/software/automake/manual/html_node/Headers.html
"Usually, only header files that accompany installed libraries
need to be installed. Headers used by programs or convenience
libraries are not installed. The noinst_HEADERS variable can be
used for such headers. However, when the header belongs to a
single convenience library or program, we recommend listing it
in the program's or library's _SOURCES variable (see Defining
program sources) instead of in noinst_HEADERS. This is clearer
for the Makefile.am reader. noinst_HEADERS would be the right
variable to use in a directory containing only headers and no
associated library or program.
All header files must be listed somewhere; in a _SOURCES
variable or in a _HEADERS variable. Missing ones will not
appear in the distribution.
"
Initially just testing erasing of Intel Software RAID signatures.
Chosen because it was expected to work, but turned out not to be true in
all cases.
The code needs to initialise GParted_Core::mainthread, construct
Gtk::Main() and execute xvfb-run because of this call chain:
GParted_Core::erase_filesystem_signatures()
GParted_Core::settle_device()
Utils::execute_command ("udevadm settle ...")
status.foreground = (Glib::Thread::self() == GParted_Core::mainthread)
Gtk::Main::run()
This was also needed when testing file system interface classes as
discussed in commits [1][2].
The test fails like this:
$ ./test_EraseFileSystemSignatures
...
[ RUN ] EraseFileSystemSignaturesTest.IntelSoftwareRAIDAligned
[ OK ] EraseFileSystemSignaturesTest.IntelSoftwareRAIDAligned (155 ms)
[ RUN ] EraseFileSystemSignaturesTest.IntelSoftwareRAIDUnaligned
test_EraseFileSystemSignatures.cc:286: Failure
Failed
image_contains_all_zeros(): First non-zero bytes:
0x00001A00 "Intel Raid ISM C" 49 6E 74 65 6C 20 52 61 69 64 20 49 53 4D 20 43
test_EraseFileSystemSignatures.cc:320: Failure
Value of: image_contains_all_zeros()
Actual: false
Expected: true
[ FAILED ] EraseFileSystemSignaturesTest.IntelSoftwareRAIDUnaligned (92 ms)
Manually write the same test image:
$ python << 'EOF'
signature = b'Intel Raid ISM Cfg Sig. '
import os
fd = os.open('/tmp/test.img', os.O_CREAT|os.O_WRONLY)
os.ftruncate(fd, 16*1024*1024 - 512)
os.lseek(fd, -(2*512), os.SEEK_END)
os.write(fd, signature)
os.close(fd)
EOF
Run gpartedbin /tmp/test.img and Format to > Cleared. GParted continues
to display the the image file as containing an ataraid signature.
$ blkid /tmp/test.img
/tmp/test.img: TYPE="isw_raid_member"
$ hexdump -C /tmp/test.img
00000000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00fffa00 49 6e 74 65 6c 20 52 61 69 64 20 49 53 4d 20 43 |Intel Raid ISM C|
00fffa10 66 67 20 53 69 67 2e 20 00 00 00 00 00 00 00 00 |fg Sig. ........|
00fffa20 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|
*
00fffe00
This signature is not being cleared when the device/partition/image size
is 512 bytes smaller than a whole MiB because the last 3.5 KiB is left
unwritten. This is because the last block of zeros written is 8 KiB
aligned to 4 KiB at the end of the device.
[1] a97c23c57c
Add initial create ext2 only FileSystem interface class test (!49)
[2] 8db9a83b39
Run test program under xvfb-run to satisfy need for an X11 display (!49)
Closes#220 - Format to Cleared not clearing "pdc" ataraid signature
'btrfs filesystem show' only used to report rounded human readable size
figures. Therefore the actual figure could have been anywhere within
the rounding limit. GParted also applied a heuristic to snap the file
system size figure to the partition size if the partition size was
within the rounding limit of the reported file system size [1].
btrfs-progs v4.1 added the --raw option to print the figures in bytes
[2][3][4].
# btrfs filesystem show --raw /dev/sdb1
Label: none uuid: 003a619e-856f-4b9c-bd29-4d0ae0296d66
Total devices 2 FS bytes used 178765824
devid 1 size 2147483648 used 239861760 path /dev/sdb1
devid 2 size 2147483648 used 436207616 path /dev/sdc1
Since the oldest supported distributions now use btrfs-progs v4.5.3 and
later (see the distribution End-of-Life table in the previous commit
message), unconditionally use this to get accurate figures.
[1] 7fc16a1b69
Handle btrfs tools rounding of figures (#499202)
[2] btrfs-progs: Allow "filesystem show" command to handle different units
https://git.kernel.org/pub/scm/linux/kernel/git/kdave/btrfs-progs.git/commit/?id=15379fa2257bf937cf7830c0b1b79f2daf5df72c
[3] btrfs-progs: docs: new size options for fi show
https://git.kernel.org/pub/scm/linux/kernel/git/kdave/btrfs-progs.git/commit/?id=81225f11d9ea58590476612e69211113ddb9b943
[4] Btrfs progs release 4.1
https://lore.kernel.org/linux-btrfs/20150622150023.GX6761@twin.jikos.cz/Closes!105 - Update used btrfs file system commands, new minimum is
btrfs-progs 4.5
Create function to replace repeated code which optionally removes
trailing new line character from a string.
Closes!105 - Update used btrfs file system commands, new minimum is
btrfs-progs 4.5
add_mountpoint_entry() doesn't modify the passed strings so use
pass-by-constant-reference. This avoids pass-by-value and having to
construct copies of the strings just to pass them to this method.
A user received the following error when attempting to resize a mounted
btrfs file system on their NixOS distribution:
Shrink /dev/nvme0n1p3 from 933.38 GiB to 894.32 GiB (ERROR)
+ calibrate /dev/nvme0n1p3 00:00:00 (SUCCESS)
+ btrfs filesystem resize 1:937759744K '/etc/machine-id' (ERROR)
ERROR: not a directory: /etc/machine-id
ERROR: resize works on mounted filesystems and accepts only
directories as argument. Passing file containing a btrfs image
would resize the underlying filesystem instead of the image.
In the partition table section of the gparted_details /dev/nvme0n1p3 was
reported with these mount points:
/etc/machine-id, /etc/NetworkManager/system-connections,
/etc/ssh/ssh_host_ed25519_key, /etc/ssh/ssh_host_ed25519_key.pub,
/etc/ssh/ssh_host_rsa_key, /etc/ssh/ssh_host_rsa_key.pub, /home,
/nix, /nix/store, /state, /var
The user had a common configuration of NixOS which boots with an empty
tmpfs as root with a few bind mounted files and directories to provide
the needed persistent data [1][2].
Re-create an equivalent situation:
1. Create a btrfs file system and mount it:
# mkfs.btrfs /dev/sdb1
# mkdir /mnt/store
# mount /dev/sdb1 /mnt/store
2. Bind mount a file from this file system else where in the hierarchy.
The only criteria is that this mount point sorts before /mnt/store.
# echo 'Test contents' > /mnt/store/test
# touch /boot/test
# mount --bind /mnt/store/test /boot/test
The kernel reports these mount mounts:
# grep sdb1 /proc/mounts
/dev/sdb1 /mnt/store btrfs rw,seclabel,relatime,space_cache=v2,subvolid=5,subvol=/ 0 0
/dev/sdb1 /boot/test btrfs rw,seclabel,relatime,space_cache=v2,subvolid=5,subvol=/ 0 0
3. Use GParted to resize this mounted btrfs file system. It fails with
the above error.
GParted read the mount points from /proc/mounts and sorted them. (See
the end of Mount_Info::load_cache() for the sorting). When resizing the
btrfs file system GParted just used the first sorted mount point. This
was the file /etc/machine-id for the user and file /boot/test in the
re-creation, hence the error.
Fix by selecting the first directory mount point to pass to the btrfs
resize command.
[1] NixOS tmpfs as root
https://elis.nu/blog/2020/05/nixos-tmpfs-as-root/
[2] Erase your darlings
https://grahamc.com/blog/erase-your-darlingsCloses#193 - path used to resize btrfs needs to be a directory
Continuing from the state in the previous commit, create an ext4 file
system using the previously created external journal and mount it.
# mke2fs -t ext4 -J device=/dev/sdb1 -L test-ext4 /dev/sdb2
# mount /dev/sdb2 /mnt/2
Did some experimenting with trying to create a second file system using
the same external journal which is already in use.
# mke2fs -t ext4 -J device=/dev/sdb1 -L 2nd-test-ext4 /dev/sdb3
...
/dev/sdb1 is apparently in use by the system; will not make a journal here!
# exit $?
1
Examined the source code of mke2fs and found that it performs an
exclusive read-only open of the named journal block device to check if
it is in use by the system or not [1]. Use the same method in GParted.
Not used alternative method would be to mark the jbd active when the
ext3/4 file system using it is active, but that requires working out the
linkage between them. That can be done using either blkid or dumpe2fs
output but that involves parsing more fields and caching more data so is
much more code than just testing the block device busy status using the
same method which mke2fs uses.
Matching UUIDs via blkid output.
# blkid /dev/sdb1 /dev/sdb2
/dev/sdb1: LABEL="test-jbd" UUID="6e52858e-0479-432f-80a1-de42f9a4093e" TYPE="jbd"
/dev/sdb2: LABEL="test-ext4" UUID="cea5c2cd-b21c-4abf-a497-8c073bb12300" EXT_JOURNAL="6e52858e-0479-432f-80a1-de42f9a4093e" TYPE="ext4"
Matching UUIDs via dumpe2fs output.
# dumpe2fs -h /dev/sdb1 | egrep 'Filesystem UUID|Journal users'
dumpe2fs 1.46.3 (27-Jul-2021)
Filesystem UUID: 6e52858e-0479-432f-80a1-de42f9a4093e
Journal users: cea5c2cd-b21c-4abf-a497-8c073bb12300
# dumpe2fs -h /dev/sdb2 | egrep 'Filesystem UUID|Journal UUID'
dumpe2fs 1.46.3 (27-Jul-2021)
Filesystem UUID: cea5c2cd-b21c-4abf-a497-8c073bb12300
Journal UUID: 6e52858e-0479-432f-80a1-de42f9a4093e
If GParted was going to show the journal to file system linkage in the
UI then doing this would be needed. However so far there has only been
a single reported case of a GParted user using an external journal,
therefore adding the code complexity for this feature is not currently
justified. The simple busy detection method used by mke2fs is all that
is needed.
[1] mke2fs source code
https://git.kernel.org/pub/scm/fs/ext2/e2fsprogs.git/
misc/mke2fs.c:main()
check_mount(journal_device, force, _("journal"));
misc/util.c:check_mount()
ext2fs_check_if_mounted(device, &mount_flags);
lib/ext2fs/ismounted.c:ext2fs_check_if_mounted()
ext2fs_check_mount_point(file, mount_flags, NULL, 0);
lib/ext2fs/ismounted.c:ext2fs_check_if_mounted()
if (stat(device, &st_buf) == 0 &&
ext2fsP_is_disk_device(st_buf.st_mode)) {
int fd = open(device, O_RDONLY | O_EXCL);
if (fd >= 0) {
/*
* The device is not busy so it's
* definitelly not mounted. No need to
* to perform any more checks.
*/
close(fd);
*mount_flags = 0;
return 0;
} else if (errno == EBUSY) {
busy = 1;
}
}
Closes#89 - GParted doesn't recognise EXT4 fs journal partition
A user reported that they were using an external journal with an ext4
file system, but that GParted didn't recognise it. (They had the jbd
on an Intel Optane drive and the ext4 file system on an SSD).
Create a jbd like this:
# mke2fs -O journal_dev -L test-jbd /dev/sdb1
# blkid /dev/sdb1
/dev/sdb1: LABEL="test-jbd" UUID="6e52858e-0479-432f-80a1-de42f9a4093e" TYPE="jbd"
Add recognition of jbd. Use Blue Shadow colour, the same as ext4,
because jbd is primarily used by ext3/4 [1][2]. jbd is also used by
ocfs2 [1][3] and lustre [4][5] clustered file systems, but they are very
unlikely to encountered by GParted users. Also xfs [6] and jfs [7] can
have external journals so if recognition of them is ever added they will
get the same colour as their respective file systems too.
[1] Journaling block device
https://en.wikipedia.org/wiki/Journaling_block_device
"JBD is filesystem-independent. ext3, ext4 and OCFS2 are known to
use JBD"
[2] https://ext4.wiki.kernel.org/index.php/Frequently_Asked_Questions#What_are_the_key_differences_between_jbd_and_jbd2.3F
[3] OCFS2: The Oracle Clustered File System, Version 2
https://www.kernel.org/doc/ols/2006/ols2006v1-pages-289-302.pdf
"Metadata journaling is done on a per node basis with JBD"
[4] Efficient Object Storage Journaling in a Distributed Parallel File
System
https://www.usenix.org/legacy/event/fast10/tech/full_papers/oral.pdf
[5] Lustre Software Release 2.x Operations Manual
https://doc.lustre.org/lustre_manual.pdf
6.4.2. Choosing Parameters for an External Journal
[6] mkfs.xfs(8) - construct an XFS filesystem
https://man7.org/linux/man-pages/man8/mkfs.xfs.8.html
"OPTIONS
...
logdev=device
This is used to specify that the log section should reside on
the device separate from the data section. The internal=1 and
logdev options are mutually exclusive.
"
[7] jfs_mkfs(8) - create a JFS formatted partition
https://manpages.debian.org/testing/jfsutils/jfs_mkfs.8.en.html
"OPTIONS
...
-j journal_device
Create the external JFS journal on journal_device, ...
"
Closes#89 - GParted doesn't recognise EXT4 fs journal partition
A bcache device provides accelerated access to a backing device in a one
to one relationship. Multiple bcache backing devices can be attached to
and accelerated by the same cache device. Extending the setup from the
previous commit, create an additional backing device and attach it to
the same cache.
# bcache make -B /dev/sdb2
# bcache attach /dev/sdc1 /dev/sdb2
# bcache show
Name Type State Bname AttachToDev
/dev/sdb2 1 (data) clean(running) bcache1 /dev/sdc1
/dev/sdb1 1 (data) clean(running) bcache0 /dev/sdc1
/dev/sdc1 3 (cache) active N/A N/A
List a couple of bcache specific sysfs files which identify registered
(active) bcache devices (components).
# ls -l /sys/block/sd?/sd??/bcache/{dev,set}
lrwxrwxrwx. 1 root root 0 Jan 7 10:08 /sys/block/sdb/sdb1/bcache/dev -> ../../../../../../../../../../virtual/block/bcache0
lrwxrwxrwx. 1 root root 0 Jan 7 11:53 /sys/block/sdb/sdb2/bcache/dev -> ../../../../../../../../../../virtual/block/bcache1
lrwxrwxrwx. 1 root root 0 Jan 7 11:53 /sys/block/sdc/sdc1/bcache/set -> ../../../../../../../../../../../fs/bcache/9945e165-0604-4f29-94bd-b155d01080ad
As was done with previous software block devices [1][2][3][4] show the
bcache (access) device as the mount point of a backing device
(component). Use the /sys/block/DEV[/PTN]/bcache/dev sysfs symlinks to
provide the bcache device names. Bcache cache devices (components)
don't get mount points because they aren't accessible.
[1] commit 8083f11d84
Display LVM2 VGNAME as the PV's mount point (#160787)
[2] commit f6c2f00df7
Populate member mount point with SWRaid array device (#756829)
[3] commit 538c866d09
Display array device as mount point of mdadm started ATARAID members
(#75)
[4] commit 538c866d09
Display array device as mount point of mdadm started ATARAID members
(#75)
Closes#183 - Basic support for bcache
Make (format as) bcache backing device (-B) and cache device (-C) and
implicitly attach the backing device to the cache to enable caching, all
in one.
# bcache make -B /dev/sdb1 -C /dev/sdc1
# bcache show
Name Type State Bname AttachToDev
/dev/sdb1 1 (data) clean(running) bcache0 /dev/sdc1
/dev/sdc1 3 (cache) active N/A N/A
After experimenting with 'bcache unregister', 'bcache register' and
stracing 'bcache show' the bcache kernel module creates the sysfs
directory /sys/block/DEV[/PTN]/bcache and it's contents only when the
bcache device is registered with the kernel (bcache component is
active). Use this to identify whether any bcache device (component)
should be displayed as active or not in GParted.
# ls -ld /sys/block/sd?/sd?1/bcache
drwxr-xr-x. 6 root root 0 Jan 7 10:08 /sys/block/sdb/sdb1/bcache
drwxr-xr-x. 2 root root 0 Jan 7 10:08 /sys/block/sdc/sdc1/bcache
Closes#183 - Basic support for bcache
Use blkid to detect bcache formatted devices. Requires blkid from
util-linux >= 2.24 for detection of bcache devices [1].
Use util-linux's FS images when testing GParted detection.
# wget http://git.kernel.org/cgit/utils/util-linux/util-linux.git/plain/tests/ts/blkid/images-fs/bcache-B.img.xz
# xzcat bcache-B.img.xz > /dev/sdb1
# wget http://git.kernel.org/cgit/utils/util-linux/util-linux.git/plain/tests/ts/blkid/images-fs/bcache-C.img.xz
# xzcat bcache-C.img.xz > /dev/sdc1
# blkid /dev/sdb1 /dev/sdc1
/dev/sdb1: UUID="8fb7f716-4c19-4517-bfbb-6f4a2becad60" TYPE="bcache" PARTUUID="f8f1485e-01"
/dev/sdc1: UUID="7a343627-ac87-4bf0-b76f-46067cbc9b8c" TYPE="bcache" PARTUUID="f46e8c86-01"
To tidy-up after testing GParted detection, stop the bcache device in
case it was automatically started and wipe the signatures. This is to
prevent udev rules from automatically starting the bcache device on
every subsequent reboot.
# echo 1 > /sys/block/sdb/sdb1/bcache/stop
# wipefs -a /dev/sdb1 /dev/sdc1
Closes#183 - Basic support for bcache
Since the only use of SWRaid_Info::get_uuid() assign the returned value
this doesn't actually save any copy construction. Do it for consistency
with the other get_*() methods in SWRaid_Info.
Closes!94 - Make more getter methods use return-by-constant-reference
Have to use a second constant reference variable array_path_2 in
GParted_Core::set_mountpoints() because by design C++ does not implement
rebinding of references [1].
[1] why doesn't C++ allow rebinding a reference?
https://stackoverflow.com/questions/27037744/why-doesnt-c-allow-rebinding-a-referenceCloses!94 - Make more getter methods use return-by-constant-reference
All uses of get_description() copy construct to a local variable, not
assign to a reference, so this doesn't save anything. It is just being
done to be consistent with making other getters return a constant
reference.
Closes!94 - Make more getter methods use return-by-constant-reference
A number of GParted methods named get_*() are returning properties and
are return-by-value. For objects this implies the returned value is
copy constructed and later destroyed when it goes out of scope. Change
them to use return-by-constant-reference to avoid unnecessary object
duplication. Just have to make sure the reference goes out of scope
before the referenced object is destroyed to avoid having a reference
(pointer) to freed memory. Assigning to a local variable instead of
of a local reference still duplicates the object so can be used when the
lifetime of the gotten object needs to be longer that the referenced
object.
Previously done for other getters here:
d948cbcb91
Make get_custom_text() and get_generic_text() return by reference
This change just makes Device::get_path() return a constant reference
and updates uses to match to avoid copy constructing the returned value.
Closes!94 - Make more getter methods use return-by-constant-reference
Creating a grep process to check if a particular mount is still mounted
is an unnecessary overhead. All that is needed is for the Mount_Info
module to refresh it's copy of /proc/mounts and query that.
To keep the code as simple as possible just reload the whole of the
Mount_Info module and query the mount cache to determine if the
particular block device is still mounted at this particular mount point.
This therefore re-reads /proc/mounts (necessary) and /proc/swaps and
/etc/fstab (unnecessary). This is still much less overhead than
creating a separate grep process.
Closes!89 - Fix unmount error when unmounting below a bind mount point
Mike Fleetwood