This adds an API to fs_mgr for reading dm-linear tables out of device trees
and issuing device-mapper ioctls. The device tree code will be
implemented separately. The dm-linear structures in fs_mgr are organized
assuming we may want to pull them from sources other than DT (for
example, text files, binary blobs, or something hardcoded for testing).
File systems which are mounted from these logical partitions have
specific fstab requirements. The block device must be a partition name,
and if Verified Boot is used, that name must match the vbmeta partition name.
Second, the entry must have the "logical" fs_mgr flag. Example fstab
entry:
vendor /vendor ext4 ro wait,logical
Example fstab entry in device tree:
vendor {
compatible = "android,fstab";
dev = "vendor";
type = "ext4";
mnt_flags = "ro";
fs_mgr_flags = "wait,slotselect,avb,logical";
};
Bug: 78914864
Test: N/A
Change-Id: I4d8878ea8858f26310119616cadc3ee0dd08566c
Remove "verity" from device-mapper helper functions so it's clear they can be
re-used for non-verity targets.
Bug: 78914864
Test: AVB device still boots
Change-Id: Id8474b2c6e23e828eff563263ebb409031cde17e
DM_READONLY_FLAG should only be used when loading tables, and
DM_STATUS_TABLE_FLAG should only be used when querying a table's status. This
patch cleans up how we set flags to reflect when the kernel actually cares
about them.
Bug: 78914864
Test: AVB device still boots
Change-Id: I809d8c2ef2105849ebdd095bbe7f08f15ae63465
The total sectors that we want to format is used in different meanings from
various users. This notifies its size based on 4096 bytes explicitly.
Bug: 76407663
Change-Id: I4e00f2e2289c1381f0238d2a4acb606a0ab551a9
Merged-In: I4e00f2e2289c1381f0238d2a4acb606a0ab551a9
Reported-by: katao@xiaomi.com
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>
Test: Modified Pixel to support and use Speck encryption, booted.
Bug: 77227248
Change-Id: I6ec272db5e4f16e9dd4f759203c14ad2a8772cd1
Merged-In: I6ec272db5e4f16e9dd4f759203c14ad2a8772cd1
With treble, the vendor kernel and ramdisk are supposed to be used.
But vendors always find new and interesting ways to break things:
$ cat /proc/device-tree/firmware/android/fstab/system/dev; echo
/by-name/system_a
$
Change-Id: I826a96ebd625b52563d3cc970bf99d9b26dc093e
This patch adds sysfs_path=%s in fstab to get the sysfs entry path.
One example is that vold is able to conduct UFS-specific operations given the
below fstab entry.
/dev/block/bootdevice/by-name/userdata /data f2fs noatime,nosuid,nodev,discard
latemount,wait,check,fileencryption=ice,quota,formattable,sysfs_path=/sys/devices/platform/soc/1d84000.ufshc
Bug: 63264275
Bug: 68721792
Change-Id: I5097873893948f12296ac5eaa0c2a59a6e6dd9e1
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>
Don't use the FDE flow to support metadata encryption; just use the
vold service which directly mounts the volume.
Bug: 63927601
Test: Boot Taimen to SUW with and without metadata encryption.
Change-Id: Idf9c27a69872cd7a9e2fb76df09a91d8e5ef4896
If we're setting up the number of reserved blocks, we also want to
set our new AID_DISK_RESERVED as the GID that's allowed to use those
blocks.
Test: builds, boots
Bug: 62024591
Change-Id: Iaabfa7d63ad9ff0b9732e2b9996937607d622fe2
Some devices, such as android on chromebooks, don't need an fstab.
Test: Ensure no error messages are seen from fs_mgr when fstab is
missing.
Change-Id: Ifadb2193743a61d03f1becefd6bc81a61eb45081
The fstab struct wasn't properly being freed.
Test: Ensure a user of fs_mgr (vold) runs without errors.
Change-Id: I4dcb8ae2ab3e831fbdb13372eb31a67a5d9fb735
Current syntax of the fstab in device tree (fstab_dt) assumes the
node name is the mount point, which doesn't allow subdir:
vendor { <== using "vendor/abc" leads to syntax error in device tree
compatible = "android,vendor";
dev = "/dev/block/platform/soc.0/f9824900.sdhci/by-name/vendor";
type = "ext4";
mnt_flags = "ro,barrier=1,inode_readahead_blks=8";
fsmgr_flags = "wait";
};
This CL adds a new field "mnt_point" in the fstab_dt configuration:
vendor_abc {
compatible = "android,vendor_abc";
dev = "/dev/block/platform/soc.0/f9824900.sdhci/by-name/abc";
type = "ext4";
mnt_flags = "ro,barrier=1,inode_readahead_blks=8";
mnt_point = "/vendor/abc"; <== new field
fsmgr_flags = "wait";
};
The new field is optional and will fallback to using node name as mount
point if it is absent.
Note that this CL also sorts fstab_dt by mount point, because
/vendor needs to be mounted first, which contains a directory
/vendor/abc, before /vendor/abc can be mounted.
Bug: 69955336
Test: first stage mount /vendor/abc on a device
Change-Id: Ie2e519f9801f211a7a221622f32c82bedd00353f
The client of FirstStageMount class should check the existence of fstab
in device-tree prior to using it. So raising a FATAL error inside
FirstStageMount when failed to parsing the fstab, in order to expose
more accurate error messages.
Also fixing a comment in fs_mgr, where it might happen in either
non-A/B or A/B.
Bug: 69102431
Test: boot sailfish
Change-Id: Ifb525beaa3423b118644778bfe0f84bff9465303
This change adds fstab.${ro.boot.hardware.platform} file into list of
searched fstab filenames.
The ${ro.boot.hardware.platform} helps consolidate init sequence based
on platform. Potentially it allows same build serve different devices.
Bug: 68663168
Change-Id: I822d5bc432e823218ebc640093cfbe5e9271d3b1
Previously there is no vboot 1.0 metadata for ENG builds. It relies on
is_device_secure() to query "ro.secure" and skip setting up dm-verity
if the value is 0 (meaning ENG build).
This change will be submitted together with other changes to add vboot
1.0 metadata for ENG builds with a "disable magic". The resulting
metadata will be the same as triggering an "adb disable-verity" on an
USERDEBUG image.
Bug: 63056044
Test: boot sailfish eng/userdebug builds
Change-Id: I35eef771e1b30bfc6d01b8ed76b40c942fe7b783
This is needed if they will ever handle ro. properties that have
values longer than 92 characters.
Bug: 23102347
Bug: 34954705
Test: read and write properties with value length > 92 characters
Change-Id: I44aa135c97ec010f12162c30f743387810ae2c5d
fs_mgr isn't usable by C now as we've migrated all fs_mgr clients in AOSP
to CPP. Removing the support for C clients.
Bug: 67024518
Bug: 67444051
Test: boot sailfish
Change-Id: Iac57f51eb6a524c6dcdc625c83af75602b4f9b3b
No fs_mgr client is using fs_mgr_get_entry_for_mount_point_after(), which
supports matching a mount point iteratively by specifying a starting entry
in the fstab. So removes it.
Also switch the public API from C/C++ to C++-only:
fs_mgr_get_entry_for_mount_point(struct fstab* fstab, const char* path);
==> fs_mgr_get_entry_for_mount_point(struct fstab* fstab, const std::string& path);
Bug: 67024518
Test: boot sailfish, manually test fs_mgr_get_entry_for_mount_point().
Change-Id: I38e18fa6bd9485b6cd31b55474406a511df3f506
Currently when searching a mount point from a fstab, it checks whether the
mount point of an fstab entry is the prefix of a given path, and the
next char after the match is either '\0' or '/'.
However, it will always return the fstab entry with mount point
"/vendor" when searching path is "/vendor", "/vendor/abc" or "/vendor/cde"
in the following fstab. Because "/vendor" is the prefix of "/vendor/abc"
and "/vendor/cde", and the next char after the match is '/'.
/dev/block/platform/.../by-name/vendor /vendor ext4 ro wait
/dev/block/platform/.../by-name/abc /vendor/abc ext4 ro wait
/dev/block/platform/.../by-name/cde /vendor/cde ext4 ro wait
Fix this by performing an exact match when searching the mount point.
Bug: 63912287
Test: boot sailfish
Change-Id: I504655f5c71790c5d528085de416ce3c30d21fea
David Anderson