Since evicting inodes triggers writes to the backing file, which uses
the mi_owner field from the mount_info struct, make sure inodes are
evicted before we free the mount_info data
Test: incfs_test
Bug: 270117845
Change-Id: I673b2e0e04b5adc3998caf6f22443598a30338af
Signed-off-by: Paul Lawrence <paullawrence@google.com>
(cherry picked from commit 7899985277527b29c47929a6d6a89c5c89b406ad)
(cherry picked from commit faf3626b8e34df3dfff3a99e6582a9abd24410ce)
Signed-off-by: Lee Jones <joneslee@google.com>
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:17 +07:00
Syzbot recently found a number of issues related to incremental-fs
(see bug numbers below). All have to do with the fact that incr-fs
allows mounts of the same source and target multiple times.
This is a design decision and the user space component "Data Loader"
expects this to work for app re-install use case.
The mounting depth needs to be controlled, however, and only allowed
to be two levels deep. In case of more than two mount attempts the
driver needs to return an error.
In case of the issues listed below the common pattern is that the
reproducer calls:
mount("./file0", "./file0", "incremental-fs", 0, NULL)
many times and then invokes a file operation like chmod, setxattr,
or open on the ./file0. This causes a recursive call for all the
mounted instances, which eventually causes a stack overflow and
a kernel crash:
BUG: stack guard page was hit at ffffc90000c0fff8
kernel stack overflow (double-fault): 0000 [#1] PREEMPT SMP KASAN
This change also cleans up the mount error path to properly clean
allocated resources and call deactivate_locked_super(), which
causes the incfs_kill_sb() to be called, where the sb is freed.
Bug: 211066171
Bug: 213140206
Bug: 213215835
Bug: 211914587
Bug: 211213635
Bug: 213137376
Bug: 211161296
Signed-off-by: Tadeusz Struk <tadeusz.struk@linaro.org>
Change-Id: I08d9b545a2715423296bf4beb67bdbbed78d1be1
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Tadeusz Strukauthored and Sorayukiicommited 2026-05-15 18:07:17 +07:00
This reverts commit 4ab5bac1598e3ed91a6267f6cada336467312112 and
2da7ed1f781983a2bc818f1d6f7cc9f2fa3ff1d2.
This is to fix the incrementalinstall test.
Can now install the same apk twice, and repeated installs are stable.
Bug: 217661925
Bug: 219731048
Signed-off-by: Tadeusz Struk <tadeusz.struk@linaro.org>
Change-Id: I77f038b296041b329060ff9350e4c9711f5ed91d
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Tadeusz Strukauthored and Sorayukiicommited 2026-05-15 18:07:17 +07:00
Cleanup incremental-fs left overs on umount, otherwise incr-fs will
complain as below:
BUG: Dentry {i=47a,n=.incomplete} still in use [unmount of incremental-fs]
This requires vfs_rmdir() of the special index dir.
Since set_anon_super() was used in incfs_mount_fs() the incfs_kill_sb()
should use kill_anon_super() instead of generic_shutdown_super()
otherwise it will leak the pseudo dev_t that set_anon_super() allocates.
Bug: 211066171
Signed-off-by: Tadeusz Struk <tadeusz.struk@linaro.org>
Change-Id: I7ea54db63513fc130e1997cbf79121015ee12405
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Tadeusz Strukauthored and Sorayukiicommited 2026-05-15 18:07:16 +07:00
Syzbot recently found a number of issues related to incremental-fs
(see bug numbers below). All have to do with the fact that incr-fs
allows mounts of the same source and target multiple times.
The correct behavior for a file system is to allow only one such
mount, and then every subsequent attempt should fail with a -EBUSY
error code. In case of the issues listed below the common pattern
is that the reproducer calls:
mount("./file0", "./file0", "incremental-fs", 0, NULL)
many times and then invokes a file operation like chmod, setxattr,
or open on the ./file0. This causes a recursive call for all the
mounted instances, which eventually causes a stack overflow and
a kernel crash:
BUG: stack guard page was hit at ffffc90000c0fff8
kernel stack overflow (double-fault): 0000 [#1] PREEMPT SMP KASAN
The reason why many mounts with the same source and target are
possible is because the incfs_mount_fs() as it is allocates a new
super_block for every call, regardless of whether a given mount already
exists or not. This happens every time the sget() function is called
with a test param equal to NULL.
The correct behavior for an FS mount implementation is to call
appropriate mount vfs call for it's type, i.e. mount_bdev() for
a block device backed FS, mount_single() for a pseudo file system,
like sysfs that is mounted in a single, well know location, or
mount_nodev() for other special purpose FS like overlayfs.
In case of incremental-fs the open coded mount logic doesn't check
for abusive mount attempts such as overlays.
To fix this issue the logic needs to be changed to pass a proper
test function to sget() call, which then checks if a super_block
for a mount instance has already been allocated and also allows
the VFS to properly verify invalid mount attempts.
Bug: 211066171
Bug: 213140206
Bug: 213215835
Bug: 211914587
Bug: 211213635
Bug: 213137376
Bug: 211161296
Signed-off-by: Tadeusz Struk <tadeusz.struk@linaro.org>
Change-Id: I66cfc3f1b5aaffb32b0845b2dad3ff26fe952e27
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Tadeusz Strukauthored and Sorayukiicommited 2026-05-15 18:07:16 +07:00
Use same selinux scheme as incfs v2
Fix memory leak
Bug: 174692664
Test: incfs_test passes
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I6058ddad9d43ba01b2eabd7d3c576f2cc9b42292
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:16 +07:00
READ/WRITE_ONCE are for atomic data types, not for structures. Fix this
up by doing a memcpy to make it explicit just how messy this copy is...
This fixes a build error on 5.8-rc1, as things are more strict, odds are
it's also wrong in other kernel versions as well...
Cc: Daniel Mentz <danielmentz@google.com>
Cc: Paul Lawrence <paullawrence@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: I7ecd3d05bd94c936dd5e69c63028458786f37a78
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Greg Kroah-Hartmanauthored and Sorayukiicommited 2026-05-15 18:07:16 +07:00
It's still magic number issue which cannot be compatible with
arm-32 platform, although we try to fix it in Iae4f3877444
("ANDROID: Incremental fs: magic number compatible 32-bit"),
there is still incompatible scenario, such as: get_incfs_node(),
it will return NULL then kernel exception will be trigger because
of NULL pointer access. (inode_set() -> get_incfs_node(), then used
node->xxx directly)
We change magic number directly, otherwise, we must fix above issues
one by one.
Bug: 159772865
Fixes: Iae4f3877444("ANDROID: Incremental fs: magic number compatible 32-bit")
Signed-off-by: Peng Zhou <Peng.Zhou@mediatek.com>
Signed-off-by: mtk81325 <peng.zhou@mediatek.com>
Change-Id: I71f279c1bb55ea296ab33a47644f30df4a9f60a6
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
mtk81325authored and Sorayukiicommited 2026-05-15 18:07:16 +07:00
Incfs's matic is bigger than 32-bit, but super block structure's
s_magic is unsigned long which is 32-bit in ARM 32-bit platform.
Do the cast for magic.
Bug: 159772865
Signed-off-by: Peng Zhou <Peng.Zhou@mediatek.com>
Signed-off-by: mtk81325 <peng.zhou@mediatek.com>
Change-Id: Iae4f38774440c7d6ae44529d4f0f8ebb2ec5dacc
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
mtk81325authored and Sorayukiicommited 2026-05-15 18:07:16 +07:00
Incremental fs appears to not depend on pkcs7 anymore.
Bug: 151584760
Signed-off-by: Daniel Mentz <danielmentz@google.com>
Change-Id: I809b4b5651d84ca70fd8bf837765e33df8547418
Signed-off-by: Paul Lawrence <paullawrence@google.com>
(cherry picked from commit 83c1d9116ec0d695b02f36cda51f305257718e09)
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Daniel Mentzauthored and Sorayukiicommited 2026-05-15 18:07:15 +07:00
This reverts commit ab185e45f637ba0b0239268f1130890c8837981d.
This change used the PageChecked flag to mark the Merkle tree as
checked. However, f2fs uses this internally. This caused file system
hangs on devices after installs.
Test: incfs_test passes, installs no longer hang
Bug: 157589629
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I980a700d65eb4f4a77434715d61dda4b8e80658c
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:15 +07:00
Waking up the waiters accounts for 80+% of the total logging
time, and about 40% of overall read_single_page() with no
signature verification. By throttling it to once every 16ms
we get back all read performance, reduce the waiter's CPU
usage and still leave it enough time to pull the logs out.
Bug: 155996534
Test: adb install megacity.apk & dd from the installed apk
Signed-off-by: Yurii Zubrytskyi <zyy@google.com>
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I4a118dc226d7ca318cf099ba3e239f0120bb23c2
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Yurii Zubrytskyiauthored and Sorayukiicommited 2026-05-15 18:07:15 +07:00
If an incfs file is created, then the file system is sync'd,on opening
the incfs file inode_set reads the size from the backing file from
within iget5_locked, causing this error.
Test: incfs_test passes, this no longer occurs
Bug: 156413528
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I8939c4afa514d39d251c044d7680cfc69272669e
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:15 +07:00
With a verified file (use incfs_perf to create a verified file), throughput
measured using dd after dropping caches increases from 200M/s to 290M/s
Test: incfs_test passes
Bug: 155996534
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I7abb5ad92e4167f82f3452acc9db322fec8307dd
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:15 +07:00
As was, chmod would change the cached inode's mode, which would
persist until the inode was uncached.
Fix to change mode of backing file, but make sure mount files
are read only, backing files are always writeable.
Test: App no longer fails with incfs errors
Bug: 154972299
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I40517331f24329484387c6b880f1517f887b29f6
(cherry picked from commit fe4fae35fe307a15cacc5e6693a98bf5140e643b)
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:14 +07:00
incfs only syncs at createfile time. This was making createfile take
a very long time. It also appears to offer little actual value -
whether we flush or no, if the device crashes the header will be
partial.
Bug: 15356649
Test: incfs_test passes, createfile takes less than half the time
Change-Id: I8f1fa138226868ebfb4a6a41254444af453070c8
Signed-off-by: Paul Lawrence <paullawrence@google.com>
(cherry picked from commit 4cc78c93ada6d0d8744c5b1ae12fddb50ea6a620)
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:14 +07:00
Bug: 153560805
Test: incfs_test passes on qemu and Pixel 4
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I1b55341e4e4247a74f3f539b9d190fef0ca409b8
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:14 +07:00
Read log buffer can have multiple threads doing any of these
operations simultaneously:
- Polling for changes
- Reading log records
- Adding new log records
- Updating log buffer size, or enabling/disabling it completely
As we don't control the userspace, and it turns out that they
all currently originate from different processes, code needs to
be safe against parallel access to a read buffer and a request
for reallocating it.
This CL add an r/w spinlock to protect the buffer and its size.
Each remount takes the write lock, while everything else takes
a read lock. Remount makes sure it doesn't take too long by
preallocating and precalculating all updates, while other
operations don't care much about their critical section size -
they all can still run together.
Bug: 152633648
Test: manual remount + reading
Signed-off-by: Yurii Zubrytskyi <zyy@google.com>
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I7271b4cb89f1ae2cbee6e5b073758f344c4ba66a
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Yurii Zubrytskyiauthored and Sorayukiicommited 2026-05-15 18:07:14 +07:00
This led to a 20x speed improvement on QEMU. 512 is somewhat
arbitrary - most of the gains are already there reading 64 records
at a time, but since the record size is 10 bytes, 512 is just over
a page and seems a good choice.
Bug: 153170997
Test: incfs_test passes. Adding logging to incfs_get_filled_blocks
to measure performance shows a 20x improvement
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: Ifb2da77cfd8c9d653c7047ba1eb7f39d795fa1c2
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:14 +07:00
Found by sparse
Bug: 153174547
Test: make C=2 fs/incfs/incrementalfs.ko no errors, incfs_test pass
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I9ff4f4f35975fe09936724488b96cd8bdeeb719e
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:13 +07:00
This change was never cherry-picked to other branches,
is just syntactic sugar, and clashes with another change. Revert,
and then apply cleanly over all branches as needed.
This reverts commit 2664a43d88d91eb09f212cb26e111656a4acee49.
Change-Id: I51dca9cb046cc15302c22e8f73e58dcf4849f84d
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:13 +07:00
Since INCFS_IOC_GET_FILLED_BLOCKS potentially leaks information about usage
patterns, and is only useful to someone filling the file, best protect it in
the same way as INCFS_IOC_FILL_BLOCKS.
Add useful field data_block_out as well
Test: incfs_test passes
Bug: 152983639
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I126a8cf711e56592479093e9aadbfd0e7f700752
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:13 +07:00
When read log is 0 sized, we still need to init the wait queue to avoid
kernel panics if someone does decide to poll on the read log.
Test: Added test for this condition, incfs_test crashes
With fix, incfs_test doesn't crash
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Bug: 152909243
Change-Id: Ic3250523bb7ddb1839f8e95852c17103e5ffb782
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:13 +07:00
When returning incomplete results index_out has to be usable to
call the function again and resume from the same location. This
means that if the output buffer was too small the function needs
to check for that when encountering the _beginning_ of a next
output range, not the end of it. Otherwise resuming from the
end of the range that didn't fit into the buffer would cause
the call to never return that range
+ Make the backing file header flags update thread safe
Bug: 152691988
Test: libincfs-test, incfs_test passes
Signed-off-by: Yurii Zubrytskyi <zyy@google.com>
Change-Id: I351156beba0b74e1942a39117279d3fcdb5e0c78
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Yurii Zubrytskyiauthored and Sorayukiicommited 2026-05-15 18:07:13 +07:00
Without these, you can't unmount a volume on which incfs was
mounted and the tests run.
Also incfs_tests would fail sporadically without the fix to
test_inode
Test: Run incfs_test and unmount underlying volume 1000 times
Bug: 152636070
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: I88f11f5d4269c22d9073e5eb671d0c7cc4629f6c
(cherry picked from commit c062bc8e769f0f6e47cc41fb9b7ab30e3e7f2689)
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:13 +07:00
Provide a securable way to open a file for filling
Test: incfs_test passes
Bug: 138149732
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: Ib4b6fd839ad30ce08e31121d19e2c0d7066d302f
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:12 +07:00
Filling blocks is not equivalent to writing a file, since they are
constrained by the root hash. selinux policy may wish to treat them
differently, for instance.
Test: incfs_test passes
Bug: 138149732
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: Ic369b84b92547b1cfefe422bd881c4e466090aed
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:12 +07:00
They provide no value and simply duplicate a check in copy_from/to_user
Test: incfs_test passes
Bug: 138149732
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Change-Id: Icc6054a2d6a495c9a03cd1507dda1ab8ca0b0dc4
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:12 +07:00
To make selinux work, add xattr support. This is a bit clunky -
it seems like it would be better for the log and pending read
functionality to be ioctls rather than this mixture of real
and virtual files.
Bug: 133435829
Change-Id: I56579fabe2ae7efb88f0344553948dc9573299aa
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:11 +07:00
- added chmod() to +0222 to make all backing files and dirs
writable. vold/system_server have a umask that clears those
flags, making incfs unusable
Signed-off-by: Yurii Zubrytskyi <zyy@google.com>
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Bug: 133435829
Change-Id: Id9258401570cc2cc7cd5735aace89d379a9b043d
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Yurii Zubrytskyiauthored and Sorayukiicommited 2026-05-15 18:07:11 +07:00
Fix all sparse errors in fs/incfs except
fs/incfs/integrity.c:192:9: warning: Variable length array is used
Test: incfs_test passes
Bug: 133435829
Change-Id: I9c2e26e4e1a06a894977f11a3c8559b968dd115e
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:11 +07:00
Fixed incfs_test build errors
Fixed Kconfig errors
Readded .gitignore
Test: With just enabling CONFIG_INCREMENTAL_FS, both defconfig and
cuttlefish_defconfig build and incfs_test runs and passes
Bug: 133435829
Change-Id: Id3247ffcc63a095f66dcedf554017a06c5a9ce4a
Signed-off-by: Paul Lawrence <paullawrence@google.com>
Signed-off-by: Sorayukii <sorayukii69@gmail.com>
Paul Lawrenceauthored and Sorayukiicommited 2026-05-15 18:07:10 +07:00
backslashxx
authored and
Sorayukii
commited 
Paul Lawrence
authored and
Tadeusz Struk
authored and
Lee Jones
authored and
Greg Kroah-Hartman
authored and
mtk81325
authored and
Daniel Mentz
authored and
Yurii Zubrytskyi
authored and
Eugene Zemtsov
authored and