Catch as many early-boot crashes as we can by starting tombstoned
immediately after /data is mounted.
Bug: http://b/139864948
Test: adb shell su 0 dmesg | grep "starting service"
Change-Id: I7f8821102191a445e87020f3efa59a2e0620d9db
We are currently creating test partitions via ImageManager, which
unfortunately leads to a lot of hacks in my local tree to test
first-stage init. ImageManager devices look nothing like real partitions
in super, and it's a lot of work to massage it to act the same.
Instead, let's create an actual super partition. We can do this via
ImageManager, and give it its very own partition table. To make this
work, we need a special IPartitionOpener that will redirect requests for
"super" to our temporary device. A new .cpp file has been added to house
this. A few other tidbits have moved there as well.
This makes setup code a bit more complicated, but now our tests will
behave much closer to a real device.
Bug: 139204329
Test: libsnapshot_test gtest
Change-Id: I2f40109cd3c1d6343d6f52979789e25f9afd533a
Before ImageManager was introduced, gsid avoided using PartitionOpener
when writing to external media. PartitionOpener couldn't interact with
non-boot devices, because it prepends /dev/block/by-name. We hacked
around this in both gsid and in first-stage init, which manually detects
the problem and prepends /dev/block instead.
After the ImageManager refactoring, sdcard support broke in gsid,
because it started relying on PartitionOpener. Let's fix this by allowing
/dev/block for mmcblk* names in PartitionOpener.
Bug: 139204329
Test: fiemap_image_test gtest
Change-Id: Ic1cbdbe0a18fc09522ee38cc62b35fd8193ce250
It's been a long standing issue that init cannot respond to property
set messages when it is running a builtin command. This is
particularly problematic when the commands involve IPC to vold or
other daemons, as it prevents them from being able to set properties.
This change has init run property service in a thread, which
eliminates the above issue.
This change may also serve as a starting block to running property
service in an entirely different process to better isolate init from
handling property requests.
Test: CF boots, walleye boots, properties are set appropriately
Change-Id: Id9534a5916abb2f7d2a49cda54e33c1b69c50c2f
A lib with 'nopreload' option in public.libraries.txt is not preloaded
during zygote. This is useful for seldom used public libraries; they
don't contribute to the zygote startup time and only affect the apps
that they are used.
Bug: 132911956
Test: libnativeloader_test
Change-Id: I6f97c90e6721aec7f2f96c8fc7b963b34f8edd3e
Ensure that off64_t is defined when building adb for mac
Fixes: 139646013
Test: make # On mac
Change-Id: I32b68738d71c8a63a35c22434092933dcabac1ce
(cherry picked from commit 2167b9082d)
host_init_verifier statically links libprocessgroup, which has shared
dependency on libjsoncpp. Right now it also has shared dependency on
libjsoncpp, resulting in ODR violation:
ERROR: AddressSanitizer: odr-violation (0x7fce895f3220):
[1] size=8 'Json::kNullRef' external/jsoncpp/src/lib_json/json_value.cpp:37:22
[2] size=8 'Json::kNullRef' external/jsoncpp/src/lib_json/json_value.cpp:37:22
These globals were registered at these points:
[1]:
#0 0x56276926ef0d (/usr/local/google/buildbot/src/android/master/out/host/linux-x86/bin/host_init_verifier+0xd1f0d)
#1 0x7fce894ac1d9 (/lib64/ld-linux-x86-64.so.2+0x101d9)
[2]:
#0 0x56276926ef0d (/usr/local/google/buildbot/src/android/master/out/host/linux-x86/bin/host_init_verifier+0xd1f0d)
#1 0x7fce894ac1d9 (/lib64/ld-linux-x86-64.so.2+0x101d9)
Move host_init_verifier to libjsoncpp_headers to fix the ODR violation.
Bug: 139546461
Bug: 131328001
Test: Run ASAN host_init_verifier
Change-Id: I54a51138d61fff37f092ffaa0758817600a0af9d
The test suite is still quite buggy if interrupted. This fixes a number
of issues (such as bad ordering of setup calls), and refactors things to
add more ASSERTs.
Bug: 139204329
Test: libsnapshot_test gtest
Change-Id: I224608715c29f343b34512a9ac1143f0dde932e9
Convert the string field to an enum. We still write a string back to the
state file.
Bug: N/A
Test: libsnapshot_test gtest
Change-Id: I7cc1cb597dacd7d6faaaba05fb01c0a86bd54c8f
IPartitionOpener is useful for writing gtests, however, it can't easily
be used with CreateLogicalPartition. fs_mgr assumes the super partition
name will map to /dev/block/by-name/super whereas in tests we want to
redirect it to a different block device.
This CL makes two changes. First, it adds a new method to IPartitionOpener
to return a "device string" for a device name. The string must either be
an absolute path (for example /dev/block/by-name/super) or a major:minor
sequence, since device-mapper will accept either.
Second, CreateLogicalPartition now accepts an optional IPartitionOpener.
When converting block devices to paths, it uses the opener instead of
automatically prepending /dev/block/by-name.
Bug: 139204329
Test: liblp_test gtest
libsnapshot_test gtest
Change-Id: Id6b3120cc2ef5c0dd941b29ff96215ad3c8ec848
* changes:
liblp: Don't halve super partition for virtual a/b devices
liblp: MetadataBuilder::GetBlockDevicePartitionName
liblp_test_static: GetMockedInstance->GetMockedPropertyFetcher
libsnapshot: UnmapSnapshot: don't delete device that doesn't exist.
liblp: change some functions to use string_view
Beside using the 1% threshold for determining that a filesystem
should be an overlayfs candidate, do so also if the free space
is below 8MB. This will cover nearly empty filesystems that can
not be right-sized below the smallest ext4 filesystem size.
Adjust the test to cover the new cases.
Should test all system partitions, including new /system_ext and
any others that get added in the future.
Fix instance of dir that were supposed to be die. Using || die
does not work as expected within a function, so use if ! <cond>.
Test: adb-remount-test.sh
Bug: 139471424
Change-Id: If3e4cac5dce7bee998727052130eaf362e9e2fb5
this makes libsparse compile with MacOS clang++
Test: "libsparse" compiles and works fine under both Linux and MacOS
Change-Id: Ifcf018b856d2fe5f1dac7eeddd72de2810be66c5
This change fixes a bug that neuralnetworks namespace is linked to the
default namespace in [vendor] section. Default namespace is for
/vendor/lib in [vendor] section. System libraries are available in the
'system' namespace.
Bug: 139113342
Test: N/A
Change-Id: I1b6751e5e33522f9829f3d9ad7371efdce01b0f0
Explicitly initialize the pointer member `app_main_namespace_` of the
class LibraryNamespaces.
Test: atest libnativeloader_tests
Change-Id: I3932e7fb3c4074a127fcdbdedbd9c88652a7e36c
For Virtual A/B devices, rename partitions / groups ending in
source slot to target slot, so that the metadata can later
be written to target metadata slot.
Test: liblp_test_static
Bug: 135752105
Change-Id: I6a7b25e8da2808b1831f317760f4345df6b22689
For Virtual A/B devices, we snapshot the partitions. Hence
there is no need to prioritize using the second half of super
partition for _b partitions.
Test: liblp_test_static
Bug: 135752105
Change-Id: I8bd669e6ffa0a4d114a1b386133179bd2819099b
Allow to retrieve name of a block device partition at
a certain index. If the index is out of range, return
empty string.
This is needed for an hypothetical edge case when an
intersection of two LinearExtent needs to be computed,
and the two metadata have a different list of block
device names. In reality, Virtual A/B devices are always
launched with dynamic partitions, so it will always return
"super" for index 0 and "" otherwise.
Test: liblp_test_static
Bug: 135752105
Change-Id: I9ea59edefdc41d0e69e3644aa2452676372938b4
Implement an epoll-backed fdevent_context, that reduces overhead when
we're polling many file descriptors. FdeventTest.smoke goes from ~5.2s
to ~3.3s when run on the host (after this patch's modification to change
it from chaining 10 file descriptors together to 512).
Test: adb_test on host
Test: adbd_test on blueline
Test: test_adb.py
Test: test_device.py
Change-Id: Iacf0093aa7bebea31e447c2cb012af72d8c3297e
As a side effect, delete FDE_ACTIVE, which was always set on every
fdevent, and FDE_PENDING, which was an internal implementation detail.
This patch removes spin detection, which will be reimplemented
separately later.
Test: adb_test on host
Test: adbd_test on blueline
Change-Id: I40be3504ce03c4fae5e071fa018542a051b7511d
Otherwise, DeleteDevice will just return false and it will
be stuck here.
Test: apply OTA after first boot, update_engine can unmap system_b
Bug: 135752105
Change-Id: If2b00336f4b948ba5f57a5e129fc070a739848fa
Converting all of them to string_view requires some
non-trivial amount of refactoring (because there is not
an implicit conversion from string_view to string), and
hence may be done later if necessary.
Test: builds
Bug: 135752105
Change-Id: I936ae7117500fadc7b34a5b91b279bc863d9919d
adb uses its own implementation of Windows HANDLE->int mapping,
and it doesn't play well with _get_osfhandle() function used in
MappedFile::FromFd().
This CL adds another function that accepts raw handle, so adb
can pass it directly
+ make constant functions 'const'
+ make the MappedFile movable, as nothing prevents it from being
one
Test: libbase_test
Change-Id: Ifde4a4094b910e9c7b431126ecf3fef5aa3bb4a6
