When allocating a partition with a size that is unaligned (to the
optimal alignment), the remaining sectors are wasted since they are
never reallocated. This is because the free list is guaranteed to only
contain optimally-aligned regions. Unfortunately this means when a
partition is resized, we are wasting a small amount of space each time.
On a non-A/B device, this could wind up being significant.
For example, with an alignment of 512KiB, a 4KiB partition at offset 0
will waste 508KiB of space. The next extent to be allocated by any
partition will start at the next 512KiB.
To address this, we check if the last extent for a partition can be
extended to cover the difference between its last sector and the next
optimally aligned sector. We also verify that this region was not
allocated to any other partition, and does not appear in the free list,
to make sure we're not stealing space that will be used somewhere else.
Bug: 120434950
Test: liblp_test gtest
Change-Id: I88689889d44a4d2c51e659241918aaf2c064e049
FIBMAP blocks are returned in FIGETBSZ units, which means the number of
blocks also needs to be determined by FIGETBSZ. Using the stat blocksize
is incorrect. On VFAT, FIGETBSZ returns 512 whereas st_blksize is 32768.
Bug: 126230649
Test: fiemap_writer_test gtest
Change-Id: Id0a667936ff9c0b60b1e8f72920cf62ceece1657
The path no longer exists, and thus the path doesn't need to be
unmounted upon remounting.
Bug: 125549215
Test: system/core/fs_mgr/tests/adb-remount-test.sh
Change-Id: I7b263c755ad7eeaa63a00ad9795a134707698625
We allocate VFAT file space by seeking and writing a single byte. It
turns out this is not fast enough to avoid invoking the progress
callback, so instead, we should seek in increments.
Bug: 126230649
Test: fiemap_writer_test gtest
Change-Id: I2b363af4ebde873411e8a3acd22ca68d175afe3f
When this patch landed, it was thought that over-allocating partitions
had no undesired side effects. However there are cases where it is now
causing problems. When flashing a GSI for VTS, the vbmeta is appended to
the end of the partition. If libfs_avb cannot find this vbmeta by
looking at the end of the partition, verification will fail. Similar use
cases with verifying other dynamic partitions are expected to be
problematic, so for now we revert to block-size alignment.
This reverts commit dccfdca1e1.
Bug: 125113249
Test: liblp_test gtest
Change-Id: Ibfd66734b63265e9fbb2ccf6b670c168c36190bc
FIBMAP returns block numbers, whereas fiemap_writer was treating them as
byte offsets.
Bug: 126230649
Test: gsi_tool install to vfat works
Change-Id: Ice80128213425aa9601e381d0fbe53f9581bf596
This adds few new methods to introduce write support to SplitFiemap:
(1) Write(), which has an internal cursor to stream data into the split
files. As the end of one file is reached, the next is opened.
(2) Flush(), which calls fsync() on each internal FiemapWriter.
(3) HasPinnedExtents(), which calls the same on each internal
FiemapWriter.
Included are some tests for edge cases in Write().
Bug: 126230649
Test: fiemap_writer_test gtest
Change-Id: I9fd509215975dbbb20a44b020315d3c1b287d1a0
For devices that do not support overlayfs that are system as root,
need to pull /dev/root device from / mount point in fstab.
Test: adb-remount-test.sh
Bug: 126256072
Change-Id: I198c0e8c60770508737318afde95db2966294ff5
Report any discrepancy in the active slot.
Fix a problem with problematic error propagation for adb_cat()
Test: adb-remount-test.sh
Bug: 126256072
Change-Id: I8a5d4e364945c5e60d252333886987b8dca0cfb3
Can supply a specific partition to remount. Partitions can be
specified by name or mount point. Some extra work to differentiate
an unknown partition, invalid partition, or one that is covered by
overlayfs.
Test: adb-remount-test.sh
Bug: 122602260
Change-Id: Iab6f51c2b5ebe01f1cea3fb235445d5e2f495365
This introduces a new SplitFiemap class that will divide an allocation
request across multiple FiemapWriters. This is primarily useful on
filesystems that have onerous restrictions on maximum file sizes. Vfat,
for example, supports a maximum of 4GiB, which is too small to satisfy
larger userdata size requests.
Bug: 126230649
Test: fiemap_writer_test gtest
Change-Id: I3c95d341e4e94e0c44bbf0e8553c34ccfdcd155b
When flashing in fastbootd, we create a new MetadataBuilder using the
given super_empty.img and attempt to import the existing partition
table. This will fail if there is some incompatibility in the partition
layout or partition quotas.
This import code was accidentally double-accounting partitions when
determining if they could fit within the group quota, preventing
"fastboot flashall" once partitions reached a certain size.
Bug: 126930319
Test: liblp_test gtest
Change-Id: I89a69cba110b62719197c9a4885cfc5bcf8f009f
Rather than require block-size alignment, instead bump the requested
file size to the necessary alignment. This ensures that the final block
is usable without placing onerous restrictions on the caller to figure
out the file system's block size.
This will require callers (namely, gsid) to track the actual desired
image size separately from the flie size.
This patch also updates tests to use the actual filesize of the
filesystem, rather than hardcoded 4096.
Bug: 126230649
Test: fiemap_writer_test gtest
Change-Id: I000cca274718c3ceac526d7c3392fe3a23bb42bc
We should check if the fs_mgr option starts with "avb_keys" before
"avb". Otherwise, it will treat "avb_keys" as "avb" fs_mgr option.
Bug: 112103720
Test: atest fs_mgr_unit_test
Change-Id: I88446222fa88e8ecfcd6f96d30ad4336ebe146a8
Deal with first version of verity for legacy products.
Test: system/core/fs_mgr/tests/adb-remount-test.sh
Bug: 120448575
Bug: 123079041
Change-Id: I7a2dd8309cbb19751fdbb05d4efc30c486615e04
The -R flag tells remount it can reboot to disable verity or to
run fsck on an ext4 deduped filesystem, or both.
Testing may include a manual component because adb-remount-test.sh
needs to run from a device in an enable-verity state to test this.
Only recognizes chained avb.
Test: adb-remount-test.sh
Bug: 122602260
Change-Id: I6ce4372532d9b933dcca9e2bec544d525b76c4d9
Allow selection of an fstab file. This is to mirror the built-in
functionality associated with init mount_all <fstab> command.
Test: adb-remount-test.sh
Bug: 122602260
Change-Id: I5fc2f3707c1dafd687c826eaccbaab03a408035b
Logging normally goes to stderr, also send output redirected to the
Android Logger. Required if command is exec'd, useful if commanded
from shell.
Test: manual confirmation of both outputs.
Bug: 122602260
Change-Id: Ibc2e14bd4fad561514c0c33741da8ca6f00af3f3
There is currently no good option for callers to setup overlayfs
on-device, it is automated as part of the adb services. Add a
remount command that does what is needed that simulates the salient
behaviors of the adb remount command.
Clean up some noise restoring device to original state when done.
Test: adb-remount-test.sh
Bug: 122602260
Change-Id: Idf213800a8182cb1c51600c8f574df8a8cd68d4a
Handle a device in recovery mode gracefully. Handle a device
that fails to boot into a commanded state more gracefully.
Deal with regression where die() calls restore(), and we can not
have this under the conditions where we are ignoring the error
in a subshell.
Test: adb-remount-test.sh
Bug: 118225373
Bug: 123079041
Change-Id: Ie37beb245d0ec55eb00757cdb93da34ff9c42827
The FstabEntry.avb_key is renamed to FstabEntry.avb_keys, to
allow specifying multiple avb keys, separated by ':'
(because ',' is already used by fstab parsing).
Bug: 124013032
Test: boot live GSI with multiple allowed AVB keys
Change-Id: Iacd3472a1d5a659dfecf09ea6074d622658f4d0b
The original adb-remount-test.sh when certifying kernels allowed a
pass on 4.4 kernels because it added new content, and missed a test
for overriding existing content. When the test was added to confirm
APEX control of libc.so, it serendipitously added a check for
overriding existing content, which the 4.4 kernel did not allow to
pass. Update the tests and documentation to reflect this new state
of affairs.
Summary: 4.4 kernel overlayfs driver worked partially without the
patch for override_creds.
Test: adb-remount-test.sh
Bug: 126256072
Change-Id: I979ea59a12bc0b9926826b9b09a7893ab3b9ee7f
This allows fiemap_writer_test to be run on any read-write partition or
block device.
Bug: 126230649
Test: fiemap_writer_test
Change-Id: Ie002cb77c84c683e5f242695882ba78d510eca36
Make it easier to collect test execution time.
Clean up some noise restoring device to original state.
Test: adb-remount-test.sh --print-time
Bug: 123079041
Change-Id: I56f12698ff25362dcefcf8a6ddd8f96a23b37f34
init first stage manufactures a / mount pointing to /dev/root and
we can not use that for checking the filesystem. Result is /system
overlay is mounted in second stage.
This adjustment to special case /dev/root is safe for overlayfs
handling because we will also gate on content in
/mnt/scratch/overlay/system, which will not be set during runtime
setup because check will have the proper block device at that time.
Test: adb-remount-test.sh
Bug: 123079041
Change-Id: I9c6674163955155f26ad2b0c9b78e9ff6507a514
The copy of /data/* doesn't work now, so put the fstab content
into the unittest source instead.
Also replacing ReadFstabFromFd with ReadFstabFromFile, to prevent
race.
Finally, two test cases are temporarily disabled due to read
fstab file error (root cause is still unknown).
Bug: 124837435
Test: atest 124837435
Change-Id: Ib6a3d931a48bffd8be23bda23fa4492babafd166
This commit extracts the security patch level (SPL), e.g.,
com.android.build.system.security_patch = 2019-04-05 from AVB property
descriptors when attempting to mount a standalone image (e.g., live
GSI). Then compares the SPL between the old system.img and the new live
system.img for rollback protection.
Bug: 122705329
Test: boot an old Live GSI, checks rollback is detected
Change-Id: I7aae58c0b2062a3ff57ed932ad58e7b604453fed
David Anderson